Covering GiST indexes

1.
/messages/by-id/1516210494.1798.16.camel@nlpgo.com

a.korotkov@postgrespro.ru

almost 8 years ago

In reply to: Andrey Borodin (#1)

Re: Covering GiST indexes

Hi, Andrey!

On Thu, Apr 12, 2018 at 2:00 PM, Andrey Borodin <x4mmm@yandex-team.ru>
wrote:

Looks like we finally have covering indexes! And that's cool!

Thank you!

So I decided to create a thread to discuss covering GiST indexes.

Here's a prototype patch implementing this functionality.
It is quite small (+80 -30) and lacks tests and docs. But it creates a
context.

I have two concerns.
First one is about INDEX_AM_RESERVED_BIT.
B-tree uses it as a base for prefix truncation (I'm not quite sure why it
is usually called suffix truncation, but this is a matter for other thread).
GiST , probably, will not use [pre\su]fix truncation. But I'd like to use
that 13th bit to implement intra-page indexing - a way to improve search
within gist page. See [0,1]

I also think that GiST isn't going to do suffix truncation of keys, because
GiST
is composing keys for every attribute and trying to use them in queries if
even GiST didn't use those attributes in any split. Depending on data
distribution,
key representation, query and so on that keys may appear useful or not.
And GiST doesn't have any legal interface to determine that in advance.

I think that GiST needs another feature: not suffix truncation, but
different
attribute representation in leaf pages and internal pages. For example,
now GiST on points stores boxes in leafs. That's a great waste of space.
So, we might just have different tuple descriptors for internal and leaf
pages of GiST, which could have both different attribute types and
different counts of attributes. Thankfully GiST pages don't have high keys
and we don't have to mix tuples of different types on the same page
(unlike B-tree).

Second, currently including indexes do not allow same attributes in both

keys and include parts.
# create index on x using gist(c) include (c);
ERROR: included columns must not intersect with key columns

But it makes sense for example for geometries like PostGIS. Index keys are
truncated to small MBRs while having whole complex geometry right in an
index could be handy.

The issue discovered in [1] seems to be related. Thus, after fix provided
there when
column is indexed without index-only scan support, then it can't be used in
index-only
scan anyway (if even it's indexed another time with index-only scan
support).
So, I think we need a better fix for [1] first.

Another thing that could be done for PostGIS geometries is just another
opclass which
stores geometries "as is" in leafs. As I know, geometries contain MBRs
inside their
own, so there is no need to store extra MBR. I think the reason why PostGIS
doesn't have such opclass yet is that geometries are frequently large and
easily
can exceed maximum size of index tuple. The same problem applies to
coverting
indexes too. Possible solution here might be to support external toasted
attributes
in covering indexes. But I think that still requires quite amount of work.

------
Alexander Korotkov
Postgres Professional: http://www.postgrespro.com
The Russian Postgres Company

Peter Geoghegan

pg@bowt.ie

almost 8 years ago

In reply to: Andrey Borodin (#1)

Re: Covering GiST indexes

On Thu, Apr 12, 2018 at 4:00 AM, Andrey Borodin <x4mmm@yandex-team.ru> wrote:

I have two concerns.
First one is about INDEX_AM_RESERVED_BIT.
B-tree uses it as a base for prefix truncation (I'm not quite sure why it is usually called suffix truncation, but this is a matter for other thread).

Since you brought it up, and since I pushed that particular
terminology, I should acknowledge that the original 1977 Bayer paper
on suffix truncation calls a B-Tree with suffix truncation a prefix
B-Tree. However, more recent work seems to consistently refer to the
technique as suffix truncation, while also referring to more advanced
techniques for compressing (not truncating) leaf tuples as prefix
compression.

I suggested suffix truncation because it seemed to be the dominant way
of referring to the technique. And, because it seemed more logical:
the suffix is what gets truncated away.

--
Peter Geoghegan

Darafei "Komяpa" Praliaskouski

me@komzpa.net

almost 8 years ago

In reply to: Alexander Korotkov (#4)

Re: Covering GiST indexes

Another thing that could be done for PostGIS geometries is just another
opclass which
stores geometries "as is" in leafs. As I know, geometries contain MBRs
inside their
own, so there is no need to store extra MBR. I think the reason why
PostGIS
doesn't have such opclass yet is that geometries are frequently large and
easily
can exceed maximum size of index tuple.

Geometry datatype layout was designed with TOASTing in mind: most of data
is stored in the header, including type, SRID, box and some other flags, so
getting just several first bytes tells you a lot.

PostGIS datasets are often of a mixed binary length: in buildings, for
example, it is quite common to have a lot of four corner houses, and just
one mapped as a circle, that digitizing software decided to make via
720-point polygon. Since reading it from TOAST for an index would require a
seek of some kind, it may be as efficient to just look it up in the table.

This way a lossy decompress function can help with index only scans: up to
some binary length, try to store the original geometry in the index. After
that, store a shape that has less points in it but covers slightly larger
area, plus a flag that it's not precise. There are a lot of ways to
generate a covering shape with less points: obvious is a box, less obvious
is non axis aligned box, a collection of boxes for a multipart shape, an
outer ring for an area with lots of holes, a convex hull or a smallest
enclosing k-gon.

In GIS there is a problem of border of Russia: the country overlaps over
180 meridian and has a complex border shape. if you take a box of it, it
spans from -180 to 180. If you query any spot in US or in Europe, you'll
have it intersecting with your area, require a full recheck, complete
detoast and decompression, and then "no, it's not a thing we need, skip".
Allowing anything better than a box would help. If we're allowing a complex
shape - we've got the container for it, geometry.

If we're storing geometry in index and original's small, why not allow
complete Index Only Scan on it, and let it skip recheck? :)

Darafei Praliaskouski,
GIS Engineer / Juno Minsk

a.korotkov@postgrespro.ru

almost 8 years ago

In reply to: Darafei "Komяpa" Praliaskouski (#6)

Re: Covering GiST indexes

On Thu, Apr 12, 2018 at 11:20 PM, Darafei "Komяpa" Praliaskouski <
me@komzpa.net> wrote:

Another thing that could be done for PostGIS geometries is just another

opclass which
stores geometries "as is" in leafs. As I know, geometries contain MBRs
inside their
own, so there is no need to store extra MBR. I think the reason why
PostGIS
doesn't have such opclass yet is that geometries are frequently large and
easily
can exceed maximum size of index tuple.

Geometry datatype layout was designed with TOASTing in mind: most of data
is stored in the header, including type, SRID, box and some other flags, so
getting just several first bytes tells you a lot.

PostGIS datasets are often of a mixed binary length: in buildings, for
example, it is quite common to have a lot of four corner houses, and just
one mapped as a circle, that digitizing software decided to make via
720-point polygon. Since reading it from TOAST for an index would require a
seek of some kind, it may be as efficient to just look it up in the table.

This way a lossy decompress function can help with index only scans: up to
some binary length, try to store the original geometry in the index. After
that, store a shape that has less points in it but covers slightly larger
area, plus a flag that it's not precise. There are a lot of ways to
generate a covering shape with less points: obvious is a box, less obvious
is non axis aligned box, a collection of boxes for a multipart shape, an
outer ring for an area with lots of holes, a convex hull or a smallest
enclosing k-gon.

In GIS there is a problem of border of Russia: the country overlaps over
180 meridian and has a complex border shape. if you take a box of it, it
spans from -180 to 180. If you query any spot in US or in Europe, you'll
have it intersecting with your area, require a full recheck, complete
detoast and decompression, and then "no, it's not a thing we need, skip".
Allowing anything better than a box would help. If we're allowing a complex
shape - we've got the container for it, geometry.

If we're storing geometry in index and original's small, why not allow
complete Index Only Scan on it, and let it skip recheck? :)

So, as I get the idea is that geometries has very different sizes
from very small to very large. And it would be nice to store small
geometries in the index "as is". And then for small geomentries
we can do index only scan and recheck while for large geometries
we have to visit heap for fetching geometry.

That can be done in custom opclass without work in PostgreSQL
core except index only scan. Right now optimizer expects index
to be always capable to return original datum for some column,
or to be never capable to do this. It doesn't allow this decision to
be done in runtime. Allowing this would require patch to PostgreSQL
core. This patch shouldn't be hard at the executor size. But
optimizer part of this patch seems hard, because I don't know
how to estimate fraction of index keys, which can be used to
reconstruct original datums. Probably that would require
GiST compress method to return some statistics...

------
Alexander Korotkov
Postgres Professional: http://www.postgrespro.com
The Russian Postgres Company

x4mmm@yandex-team.ru

almost 8 years ago

In reply to: Teodor Sigaev (#2)

1 attachment(s)

Re: Covering GiST indexes

12 апр. 2018 г., в 17:03, Teodor Sigaev <teodor@sigaev.ru> написал(а):

Interesting work. I don't have a time now to learn deep your patch, so, add it to next commitfest, pls.

Thanks! Sure, I'll add it.

First of all I'd like to see more tests in patch, not only CREATE INDEX.

Here's V2, with basic set of tests.
Currently, I'm investigating what to document and more places to tests.

Also, I do not use generic function index_truncate_tuple(), because it deforms and then forms tuple, but GiST usually have already deformed tuple.

Best regards, Andrey Borodin.

Attachments:

0001-Covering-GiST-v2.patchapplication/octet-stream; name=0001-Covering-GiST-v2.patch; x-unix-mode=0644Download

From df93b2de26fb59b4bcc7ac5899ad7b5ce06140bd Mon Sep 17 00:00:00 2001
From: Andrey Borodin <amborodin@acm.org>
Date: Wed, 11 Apr 2018 17:43:15 +0500
Subject: [PATCH] Covering GiST v2

---
 src/backend/access/gist/gist.c                     |  34 ++++-
 src/backend/access/gist/gistbuild.c                |   2 +-
 src/backend/access/gist/gistget.c                  |   4 +-
 src/backend/access/gist/gistscan.c                 |  12 +-
 src/backend/access/gist/gistsplit.c                |  12 +-
 src/backend/access/gist/gistutil.c                 |  48 +++++--
 src/include/access/gist_private.h                  |   5 +-
 src/test/regress/expected/amutils.out              |   4 +-
 src/test/regress/expected/index_including.out      |   8 +-
 src/test/regress/expected/index_including_gist.out | 145 +++++++++++++++++++++
 src/test/regress/parallel_schedule                 |   2 +-
 src/test/regress/serial_schedule                   |   1 +
 src/test/regress/sql/index_including.sql           |   6 +-
 src/test/regress/sql/index_including_gist.sql      |  85 ++++++++++++
 14 files changed, 328 insertions(+), 40 deletions(-)
 create mode 100644 src/test/regress/expected/index_including_gist.out
 create mode 100644 src/test/regress/sql/index_including_gist.sql

diff --git a/src/backend/access/gist/gist.c b/src/backend/access/gist/gist.c
index 9007d65ad2..0303d022bf 100644
--- a/src/backend/access/gist/gist.c
+++ b/src/backend/access/gist/gist.c
@@ -74,7 +74,7 @@ gisthandler(PG_FUNCTION_ARGS)
 	amroutine->amclusterable = true;
 	amroutine->ampredlocks = true;
 	amroutine->amcanparallel = false;
-	amroutine->amcaninclude = false;
+	amroutine->amcaninclude = true;
 	amroutine->amkeytype = InvalidOid;
 
 	amroutine->ambuild = gistbuild;
@@ -169,7 +169,7 @@ gistinsert(Relation r, Datum *values, bool *isnull,
 	oldCxt = MemoryContextSwitchTo(giststate->tempCxt);
 
 	itup = gistFormTuple(giststate, r,
-						 values, isnull, true /* size is currently bogus */ );
+						 values, isnull, true /* size is currently bogus */, false);
 	itup->t_tid = *ht_ctid;
 
 	gistdoinsert(r, itup, 0, giststate);
@@ -1377,8 +1377,8 @@ gistSplit(Relation r,
 						IndexTupleSize(itup[0]), GiSTPageSize,
 						RelationGetRelationName(r))));
 
-	memset(v.spl_lisnull, true, sizeof(bool) * giststate->tupdesc->natts);
-	memset(v.spl_risnull, true, sizeof(bool) * giststate->tupdesc->natts);
+	memset(v.spl_lisnull, true, sizeof(bool) * giststate->truncTupdesc->natts);
+	memset(v.spl_risnull, true, sizeof(bool) * giststate->truncTupdesc->natts);
 	gistSplitByKey(r, page, itup, len, giststate, &v, 0);
 
 	/* form left and right vector */
@@ -1401,7 +1401,7 @@ gistSplit(Relation r,
 		ROTATEDIST(res);
 		res->block.num = v.splitVector.spl_nright;
 		res->list = gistfillitupvec(rvectup, v.splitVector.spl_nright, &(res->lenlist));
-		res->itup = gistFormTuple(giststate, r, v.spl_rattr, v.spl_risnull, false);
+		res->itup = gistFormTuple(giststate, r, v.spl_rattr, v.spl_risnull, false, true);
 	}
 
 	if (!gistfitpage(lvectup, v.splitVector.spl_nleft))
@@ -1423,7 +1423,7 @@ gistSplit(Relation r,
 		ROTATEDIST(res);
 		res->block.num = v.splitVector.spl_nleft;
 		res->list = gistfillitupvec(lvectup, v.splitVector.spl_nleft, &(res->lenlist));
-		res->itup = gistFormTuple(giststate, r, v.spl_lattr, v.spl_lisnull, false);
+		res->itup = gistFormTuple(giststate, r, v.spl_lattr, v.spl_lisnull, false, true);
 	}
 
 	return res;
@@ -1457,8 +1457,10 @@ initGISTstate(Relation index)
 	giststate->scanCxt = scanCxt;
 	giststate->tempCxt = scanCxt;	/* caller must change this if needed */
 	giststate->tupdesc = index->rd_att;
+	giststate->truncTupdesc = CreateTupleDescCopyConstr(index->rd_att);
+	giststate->truncTupdesc->natts = IndexRelationGetNumberOfKeyAttributes(index);
 
-	for (i = 0; i < index->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(index); i++)
 	{
 		fmgr_info_copy(&(giststate->consistentFn[i]),
 					   index_getprocinfo(index, i + 1, GIST_CONSISTENT_PROC),
@@ -1526,6 +1528,24 @@ initGISTstate(Relation index)
 			giststate->supportCollation[i] = DEFAULT_COLLATION_OID;
 	}
 
+	for (; i < index->rd_att->natts; i++)
+	{
+		giststate->consistentFn[i].fn_oid = InvalidOid;
+		giststate->unionFn[i].fn_oid = InvalidOid;
+		giststate->compressFn[i].fn_oid = InvalidOid;
+		giststate->decompressFn[i].fn_oid = InvalidOid;
+		giststate->penaltyFn[i].fn_oid = InvalidOid;
+		giststate->picksplitFn[i].fn_oid = InvalidOid;
+		giststate->equalFn[i].fn_oid = InvalidOid;
+		giststate->distanceFn[i].fn_oid = InvalidOid;
+		giststate->fetchFn[i].fn_oid = InvalidOid;
+
+		if (OidIsValid(index->rd_indcollation[i]))
+			giststate->supportCollation[i] = index->rd_indcollation[i];
+		else
+			giststate->supportCollation[i] = DEFAULT_COLLATION_OID;
+	}
+
 	MemoryContextSwitchTo(oldCxt);
 
 	return giststate;
diff --git a/src/backend/access/gist/gistbuild.c b/src/backend/access/gist/gistbuild.c
index 434f15f014..f4545f8069 100644
--- a/src/backend/access/gist/gistbuild.c
+++ b/src/backend/access/gist/gistbuild.c
@@ -469,7 +469,7 @@ gistBuildCallback(Relation index,
 	oldCtx = MemoryContextSwitchTo(buildstate->giststate->tempCxt);
 
 	/* form an index tuple and point it at the heap tuple */
-	itup = gistFormTuple(buildstate->giststate, index, values, isnull, true);
+	itup = gistFormTuple(buildstate->giststate, index, values, isnull, true, false);
 	itup->t_tid = htup->t_self;
 
 	if (buildstate->bufferingMode == GIST_BUFFERING_ACTIVE)
diff --git a/src/backend/access/gist/gistget.c b/src/backend/access/gist/gistget.c
index c4e8a3b913..afb8d31c37 100644
--- a/src/backend/access/gist/gistget.c
+++ b/src/backend/access/gist/gistget.c
@@ -805,12 +805,14 @@ gistgetbitmap(IndexScanDesc scan, TIDBitmap *tbm)
  *
  * Opclasses that implement a fetch function support index-only scans.
  * Opclasses without compression functions also support index-only scans.
+ * Included attributes can be returned.
  */
 bool
 gistcanreturn(Relation index, int attno)
 {
 	if (OidIsValid(index_getprocid(index, attno, GIST_FETCH_PROC)) ||
-		!OidIsValid(index_getprocid(index, attno, GIST_COMPRESS_PROC)))
+		!OidIsValid(index_getprocid(index, attno, GIST_COMPRESS_PROC))||
+		attno > IndexRelationGetNumberOfKeyAttributes(index))
 		return true;
 	else
 		return false;
diff --git a/src/backend/access/gist/gistscan.c b/src/backend/access/gist/gistscan.c
index 4d97ff1d5d..fb93e8d6a6 100644
--- a/src/backend/access/gist/gistscan.c
+++ b/src/backend/access/gist/gistscan.c
@@ -158,6 +158,7 @@ gistrescan(IndexScanDesc scan, ScanKey key, int nkeys,
 	if (scan->xs_want_itup && !scan->xs_hitupdesc)
 	{
 		int			natts;
+		int			nkeyatts;
 		int			attno;
 
 		/*
@@ -167,13 +168,22 @@ gistrescan(IndexScanDesc scan, ScanKey key, int nkeys,
 		 * types.
 		 */
 		natts = RelationGetNumberOfAttributes(scan->indexRelation);
+		nkeyatts = IndexRelationGetNumberOfKeyAttributes(scan->indexRelation);
 		so->giststate->fetchTupdesc = CreateTemplateTupleDesc(natts, false);
-		for (attno = 1; attno <= natts; attno++)
+		for (attno = 1; attno <= nkeyatts; attno++)
 		{
 			TupleDescInitEntry(so->giststate->fetchTupdesc, attno, NULL,
 							   scan->indexRelation->rd_opcintype[attno - 1],
 							   -1, 0);
 		}
+
+		for (; attno <= natts; attno++)
+		{
+			/* taking opcintype from giststate->tupdesc */
+			TupleDescInitEntry(so->giststate->fetchTupdesc, attno, NULL,
+					TupleDescAttr(so->giststate->tupdesc, attno - 1)->atttypid,
+							   -1, 0);
+		}
 		scan->xs_hitupdesc = so->giststate->fetchTupdesc;
 
 		/* Also create a memory context that will hold the returned tuples */
diff --git a/src/backend/access/gist/gistsplit.c b/src/backend/access/gist/gistsplit.c
index a7038cca67..02918536bf 100644
--- a/src/backend/access/gist/gistsplit.c
+++ b/src/backend/access/gist/gistsplit.c
@@ -207,7 +207,7 @@ placeOne(Relation r, GISTSTATE *giststate, GistSplitVector *v,
 	gistDeCompressAtt(giststate, r, itup, NULL, (OffsetNumber) 0,
 					  identry, isnull);
 
-	for (; attno < giststate->tupdesc->natts; attno++)
+	for (; attno < giststate->truncTupdesc->natts; attno++)
 	{
 		float		lpenalty,
 					rpenalty;
@@ -485,7 +485,7 @@ gistUserPicksplit(Relation r, GistEntryVector *entryvec, int attno, GistSplitVec
 	 */
 	v->spl_dontcare = NULL;
 
-	if (attno + 1 < giststate->tupdesc->natts)
+	if (attno + 1 < giststate->truncTupdesc->natts)
 	{
 		int			NumDontCare;
 
@@ -657,7 +657,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 		 */
 		v->spl_risnull[attno] = v->spl_lisnull[attno] = true;
 
-		if (attno + 1 < giststate->tupdesc->natts)
+		if (attno + 1 < giststate->truncTupdesc->natts)
 			gistSplitByKey(r, page, itup, len, giststate, v, attno + 1);
 		else
 			gistSplitHalf(&v->splitVector, len);
@@ -683,7 +683,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 				v->splitVector.spl_left[v->splitVector.spl_nleft++] = i;
 
 		/* Compute union keys, unless outer recursion level will handle it */
-		if (attno == 0 && giststate->tupdesc->natts == 1)
+		if (attno == 0 && giststate->truncTupdesc->natts == 1)
 		{
 			v->spl_dontcare = NULL;
 			gistunionsubkey(giststate, itup, v);
@@ -700,7 +700,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 			 * Splitting on attno column is not optimal, so consider
 			 * redistributing don't-care tuples according to the next column
 			 */
-			Assert(attno + 1 < giststate->tupdesc->natts);
+			Assert(attno + 1 < giststate->truncTupdesc->natts);
 
 			if (v->spl_dontcare == NULL)
 			{
@@ -771,7 +771,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 	 * that PickSplit (or the special cases above) produced correct union
 	 * datums.
 	 */
-	if (attno == 0 && giststate->tupdesc->natts > 1)
+	if (attno == 0 && giststate->truncTupdesc->natts > 1)
 	{
 		v->spl_dontcare = NULL;
 		gistunionsubkey(giststate, itup, v);
diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c
index 55cccd247a..9180062e9a 100644
--- a/src/backend/access/gist/gistutil.c
+++ b/src/backend/access/gist/gistutil.c
@@ -160,7 +160,7 @@ gistMakeUnionItVec(GISTSTATE *giststate, IndexTuple *itvec, int len,
 
 	evec = (GistEntryVector *) palloc((len + 2) * sizeof(GISTENTRY) + GEVHDRSZ);
 
-	for (i = 0; i < giststate->tupdesc->natts; i++)
+	for (i = 0; i < giststate->truncTupdesc->natts; i++)
 	{
 		int			j;
 
@@ -221,7 +221,7 @@ gistunion(Relation r, IndexTuple *itvec, int len, GISTSTATE *giststate)
 
 	gistMakeUnionItVec(giststate, itvec, len, attr, isnull);
 
-	return gistFormTuple(giststate, r, attr, isnull, false);
+	return gistFormTuple(giststate, r, attr, isnull, false, true);
 }
 
 /*
@@ -296,7 +296,7 @@ gistDeCompressAtt(GISTSTATE *giststate, Relation r, IndexTuple tuple, Page p,
 {
 	int			i;
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		Datum		datum;
 
@@ -329,7 +329,7 @@ gistgetadjusted(Relation r, IndexTuple oldtup, IndexTuple addtup, GISTSTATE *gis
 	gistDeCompressAtt(giststate, r, addtup, NULL,
 					  (OffsetNumber) 0, addentries, addisnull);
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		gistMakeUnionKey(giststate, i,
 						 oldentries + i, oldisnull[i],
@@ -355,7 +355,7 @@ gistgetadjusted(Relation r, IndexTuple oldtup, IndexTuple addtup, GISTSTATE *gis
 	if (neednew)
 	{
 		/* need to update key */
-		newtup = gistFormTuple(giststate, r, attr, isnull, false);
+		newtup = gistFormTuple(giststate, r, attr, isnull, false, true);
 		newtup->t_tid = oldtup->t_tid;
 	}
 
@@ -442,7 +442,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 		zero_penalty = true;
 
 		/* Loop over index attributes. */
-		for (j = 0; j < r->rd_att->natts; j++)
+		for (j = 0; j < IndexRelationGetNumberOfKeyAttributes(r); j++)
 		{
 			Datum		datum;
 			float		usize;
@@ -470,7 +470,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 				result = i;
 				best_penalty[j] = usize;
 
-				if (j < r->rd_att->natts - 1)
+				if (j < IndexRelationGetNumberOfKeyAttributes(r) - 1)
 					best_penalty[j + 1] = -1;
 
 				/* we have new best, so reset keep-it decision */
@@ -500,7 +500,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 		 * If we looped past the last column, and did not update "result",
 		 * then this tuple is exactly as good as the prior best tuple.
 		 */
-		if (j == r->rd_att->natts && result != i)
+		if (j == IndexRelationGetNumberOfKeyAttributes(r) && result != i)
 		{
 			if (keep_current_best == -1)
 			{
@@ -570,7 +570,7 @@ gistdentryinit(GISTSTATE *giststate, int nkey, GISTENTRY *e,
 
 IndexTuple
 gistFormTuple(GISTSTATE *giststate, Relation r,
-			  Datum attdata[], bool isnull[], bool isleaf)
+			  Datum attdata[], bool isnull[], bool isleaf, bool isTruncated)
 {
 	Datum		compatt[INDEX_MAX_KEYS];
 	int			i;
@@ -579,7 +579,7 @@ gistFormTuple(GISTSTATE *giststate, Relation r,
 	/*
 	 * Call the compress method on each attribute.
 	 */
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		if (isnull[i])
 			compatt[i] = (Datum) 0;
@@ -602,7 +602,23 @@ gistFormTuple(GISTSTATE *giststate, Relation r,
 		}
 	}
 
-	res = index_form_tuple(giststate->tupdesc, compatt, isnull);
+	if (!isTruncated)
+	{
+		/*
+		* Allocate each included attribute.
+		*/	
+		for (; i < r->rd_att->natts; i++)
+		{
+			if (isnull[i])
+				compatt[i] = (Datum) 0;
+			else
+			{
+				compatt[i] = attdata[i];
+			}
+		}
+	}
+
+	res = index_form_tuple(isTruncated?giststate->truncTupdesc:giststate->tupdesc, compatt, isnull);
 
 	/*
 	 * The offset number on tuples on internal pages is unused. For historical
@@ -644,7 +660,7 @@ gistFetchTuple(GISTSTATE *giststate, Relation r, IndexTuple tuple)
 	bool		isnull[INDEX_MAX_KEYS];
 	int			i;
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		Datum		datum;
 
@@ -679,6 +695,14 @@ gistFetchTuple(GISTSTATE *giststate, Relation r, IndexTuple tuple)
 			fetchatt[i] = (Datum) 0;
 		}
 	}
+
+	/*
+	 * Get each included attribute.
+	 */
+	for (; i < r->rd_att->natts; i++)
+	{
+		fetchatt[i] = index_getattr(tuple, i + 1, giststate->tupdesc, &isnull[i]);
+	}
 	MemoryContextSwitchTo(oldcxt);
 
 	return heap_form_tuple(giststate->fetchTupdesc, fetchatt, isnull);
diff --git a/src/include/access/gist_private.h b/src/include/access/gist_private.h
index 36ed7244ba..c1714a6ccc 100644
--- a/src/include/access/gist_private.h
+++ b/src/include/access/gist_private.h
@@ -79,6 +79,8 @@ typedef struct GISTSTATE
 	MemoryContext tempCxt;		/* short-term context for calling functions */
 
 	TupleDesc	tupdesc;		/* index's tuple descriptor */
+	TupleDesc	truncTupdesc;	/* truncated tuple descriptor
+								 * for internal pages */
 	TupleDesc	fetchTupdesc;	/* tuple descriptor for tuples returned in an
 								 * index-only scan */
 
@@ -458,7 +460,8 @@ extern IndexTuple gistgetadjusted(Relation r,
 				IndexTuple addtup,
 				GISTSTATE *giststate);
 extern IndexTuple gistFormTuple(GISTSTATE *giststate,
-			  Relation r, Datum *attdata, bool *isnull, bool isleaf);
+			  Relation r, Datum *attdata, bool *isnull, bool isleaf,
+			  bool isTruncated);
 
 extern OffsetNumber gistchoose(Relation r, Page p,
 		   IndexTuple it,
diff --git a/src/test/regress/expected/amutils.out b/src/test/regress/expected/amutils.out
index 24cd3c5e2e..0c481f94ab 100644
--- a/src/test/regress/expected/amutils.out
+++ b/src/test/regress/expected/amutils.out
@@ -75,7 +75,7 @@ select prop,
  can_unique         | f  |       | 
  can_multi_col      | t  |       | 
  can_exclude        | t  |       | 
- can_include        | f  |       | 
+ can_include        | t  |       | 
  bogus              |    |       | 
 (19 rows)
 
@@ -158,7 +158,7 @@ select amname, prop, pg_indexam_has_property(a.oid, prop) as p
  gist   | can_unique    | f
  gist   | can_multi_col | t
  gist   | can_exclude   | t
- gist   | can_include   | f
+ gist   | can_include   | t
  gist   | bogus         | 
  hash   | can_order     | f
  hash   | can_unique    | f
diff --git a/src/test/regress/expected/index_including.out b/src/test/regress/expected/index_including.out
index 1d253ee77d..e4b42ae200 100644
--- a/src/test/regress/expected/index_including.out
+++ b/src/test/regress/expected/index_including.out
@@ -292,22 +292,20 @@ SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl' ORDER BY indexname;
 
 DROP TABLE tbl;
 /*
- * 7. Check various AMs. All but btree must fail.
+ * 7. Check various AMs. All but btree and gist must fail.
  */
 CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
 CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "brin" does not support included columns
-CREATE INDEX on tbl USING gist(c3) INCLUDE (c4);
-ERROR:  access method "gist" does not support included columns
+CREATE INDEX on tbl USING gist(c4) INCLUDE (c3);
 CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
 ERROR:  access method "spgist" does not support included columns
 CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "gin" does not support included columns
 CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "hash" does not support included columns
-CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
+CREATE INDEX on tbl USING rtree(c4) INCLUDE (c3, c1);
 NOTICE:  substituting access method "gist" for obsolete method "rtree"
-ERROR:  access method "gist" does not support included columns
 CREATE INDEX on tbl USING btree(c1, c2) INCLUDE (c3, c4);
 DROP TABLE tbl;
 /*
diff --git a/src/test/regress/expected/index_including_gist.out b/src/test/regress/expected/index_including_gist.out
new file mode 100644
index 0000000000..5b87da9bb0
--- /dev/null
+++ b/src/test/regress/expected/index_including_gist.out
@@ -0,0 +1,145 @@
+/*
+ * 1.1. test CREATE INDEX with buffered build
+ */
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+-- must fail because of intersection of key and included columns
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c4);
+ERROR:  included columns must not intersect with key columns
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+                                  pg_get_indexdef                                  
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+ c1 | c2 | c3 |     c4      
+----+----+----+-------------
+  1 |  2 |  3 | (2,3),(1,2)
+  2 |  4 |  6 | (4,5),(2,3)
+  3 |  6 |  9 | (6,7),(3,4)
+  4 |  8 | 12 | (8,9),(4,5)
+(4 rows)
+
+SET enable_bitmapscan TO off;
+EXPLAIN SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                                     QUERY PLAN                                     
+------------------------------------------------------------------------------------
+ Index Only Scan using tbl_gist_idx on tbl_gist  (cost=0.27..36.41 rows=8 width=44)
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+/*
+ * 1.2. test CREATE INDEX with inserts
+ */
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+-- must fail because of intersection of key and included columns
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c4);
+ERROR:  included columns must not intersect with key columns
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+                                  pg_get_indexdef                                  
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+ c1 | c2 | c3 |     c4      
+----+----+----+-------------
+  1 |  2 |  3 | (2,3),(1,2)
+  2 |  4 |  6 | (4,5),(2,3)
+  3 |  6 |  9 | (6,7),(3,4)
+  4 |  8 | 12 | (8,9),(4,5)
+(4 rows)
+
+SET enable_bitmapscan TO off;
+EXPLAIN SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                                     QUERY PLAN                                     
+------------------------------------------------------------------------------------
+ Index Only Scan using tbl_gist_idx on tbl_gist  (cost=0.28..36.41 rows=8 width=44)
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+/*
+ * 2. CREATE INDEX CONCURRENTLY
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX CONCURRENTLY tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                     indexdef                                      
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+DROP TABLE tbl_gist;
+/*
+ * 3. REINDEX
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                   indexdef                                    
+-------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c3)
+(1 row)
+
+REINDEX INDEX tbl_gist_idx;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                   indexdef                                    
+-------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c3)
+(1 row)
+
+ALTER TABLE tbl_gist DROP COLUMN c1;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+ indexdef 
+----------
+(0 rows)
+
+DROP TABLE tbl_gist;
+/*
+ * 4. Update, delete values in indexed table.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+UPDATE tbl_gist SET c1 = 100 WHERE c1 = 2;
+UPDATE tbl_gist SET c1 = 1 WHERE c1 = 3;
+DELETE FROM tbl_gist WHERE c1 = 5 OR c3 = 12;
+DROP TABLE tbl_gist;
+/*
+ * 5. Alter column type.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+ALTER TABLE tbl_gist ALTER c1 TYPE bigint;
+ALTER TABLE tbl_gist ALTER c3 TYPE bigint;
+\d tbl_gist
+              Table "public.tbl_gist"
+ Column |  Type   | Collation | Nullable | Default 
+--------+---------+-----------+----------+---------
+ c1     | bigint  |           |          | 
+ c2     | integer |           |          | 
+ c3     | bigint  |           |          | 
+ c4     | box     |           |          | 
+Indexes:
+    "tbl_gist_idx" gist (c4) INCLUDE (c1, c3)
+
+DROP TABLE tbl_gist;
diff --git a/src/test/regress/parallel_schedule b/src/test/regress/parallel_schedule
index 0d3a27ed41..9bf4292f18 100644
--- a/src/test/regress/parallel_schedule
+++ b/src/test/regress/parallel_schedule
@@ -55,7 +55,7 @@ test: copy copyselect copydml
 # ----------
 test: create_misc create_operator create_procedure
 # These depend on the above two
-test: create_index create_view index_including
+test: create_index create_view index_including index_including_gist
 
 # ----------
 # Another group of parallel tests
diff --git a/src/test/regress/serial_schedule b/src/test/regress/serial_schedule
index 20027c131c..ce85776a5c 100644
--- a/src/test/regress/serial_schedule
+++ b/src/test/regress/serial_schedule
@@ -66,6 +66,7 @@ test: create_operator
 test: create_procedure
 test: create_index
 test: index_including
+test: index_including_gist
 test: create_view
 test: create_aggregate
 test: create_function_3
diff --git a/src/test/regress/sql/index_including.sql b/src/test/regress/sql/index_including.sql
index caedc9866d..fdda3cdd79 100644
--- a/src/test/regress/sql/index_including.sql
+++ b/src/test/regress/sql/index_including.sql
@@ -165,15 +165,15 @@ SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl' ORDER BY indexname;
 DROP TABLE tbl;
 
 /*
- * 7. Check various AMs. All but btree must fail.
+ * 7. Check various AMs. All but btree and gist must fail.
  */
 CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
 CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
-CREATE INDEX on tbl USING gist(c3) INCLUDE (c4);
+CREATE INDEX on tbl USING gist(c4) INCLUDE (c3);
 CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
 CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
 CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
-CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
+CREATE INDEX on tbl USING rtree(c4) INCLUDE (c3, c1);
 CREATE INDEX on tbl USING btree(c1, c2) INCLUDE (c3, c4);
 DROP TABLE tbl;
 
diff --git a/src/test/regress/sql/index_including_gist.sql b/src/test/regress/sql/index_including_gist.sql
new file mode 100644
index 0000000000..33e3924eaa
--- /dev/null
+++ b/src/test/regress/sql/index_including_gist.sql
@@ -0,0 +1,85 @@
+/*
+ * 1.1. test CREATE INDEX with buffered build
+ */
+
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+-- must fail because of intersection of key and included columns
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c4);
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO off;
+EXPLAIN SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+
+/*
+ * 1.2. test CREATE INDEX with inserts
+ */
+
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+-- must fail because of intersection of key and included columns
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c4);
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO off;
+EXPLAIN SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+
+/*
+ * 2. CREATE INDEX CONCURRENTLY
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX CONCURRENTLY tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+DROP TABLE tbl_gist;
+
+
+/*
+ * 3. REINDEX
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+REINDEX INDEX tbl_gist_idx;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+ALTER TABLE tbl_gist DROP COLUMN c1;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+DROP TABLE tbl_gist;
+
+/*
+ * 4. Update, delete values in indexed table.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+UPDATE tbl_gist SET c1 = 100 WHERE c1 = 2;
+UPDATE tbl_gist SET c1 = 1 WHERE c1 = 3;
+DELETE FROM tbl_gist WHERE c1 = 5 OR c3 = 12;
+DROP TABLE tbl_gist;
+
+/*
+ * 5. Alter column type.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+ALTER TABLE tbl_gist ALTER c1 TYPE bigint;
+ALTER TABLE tbl_gist ALTER c3 TYPE bigint;
+\d tbl_gist
+DROP TABLE tbl_gist;
+
-- 
2.15.1 (Apple Git-101)

x4mmm@yandex-team.ru

over 7 years ago

In reply to: Andrey Borodin (#8)

1 attachment(s)

Re: Covering GiST indexes

Hi!

13 апр. 2018 г., в 17:36, Andrey Borodin <x4mmm@yandex-team.ru> написал(а):

Here's V2, with basic set of tests.
Currently, I'm investigating what to document and more places to tests.

Here is v3 version of the patch. I've fixed some comments and added some words to docs.

Best regards, Andrey Borodin.

Attachments:

0001-Covering-GiST-v3.patchapplication/octet-stream; name=0001-Covering-GiST-v3.patch; x-unix-mode=0644Download

From 79a3159436c1323ec68d154e6169aba5174aa636 Mon Sep 17 00:00:00 2001
From: Andrey Borodin <amborodin@acm.org>
Date: Thu, 5 Jul 2018 20:35:43 +0400
Subject: [PATCH] Covering GiST v3

---
 doc/src/sgml/ref/create_index.sgml                 |   2 +-
 doc/src/sgml/textsearch.sgml                       |   6 +
 src/backend/access/gist/gist.c                     |  34 ++++-
 src/backend/access/gist/gistbuild.c                |   2 +-
 src/backend/access/gist/gistget.c                  |   4 +-
 src/backend/access/gist/gistscan.c                 |  12 +-
 src/backend/access/gist/gistsplit.c                |  12 +-
 src/backend/access/gist/gistutil.c                 |  48 +++++--
 src/include/access/gist_private.h                  |   5 +-
 src/test/regress/expected/amutils.out              |   4 +-
 src/test/regress/expected/index_including.out      |   8 +-
 src/test/regress/expected/index_including_gist.out | 145 +++++++++++++++++++++
 src/test/regress/parallel_schedule                 |   2 +-
 src/test/regress/serial_schedule                   |   1 +
 src/test/regress/sql/index_including.sql           |   6 +-
 src/test/regress/sql/index_including_gist.sql      |  84 ++++++++++++
 16 files changed, 334 insertions(+), 41 deletions(-)
 create mode 100644 src/test/regress/expected/index_including_gist.out
 create mode 100644 src/test/regress/sql/index_including_gist.sql

diff --git a/doc/src/sgml/ref/create_index.sgml b/doc/src/sgml/ref/create_index.sgml
index 91692325a5..cb25613967 100644
--- a/doc/src/sgml/ref/create_index.sgml
+++ b/doc/src/sgml/ref/create_index.sgml
@@ -178,7 +178,7 @@ CREATE [ UNIQUE ] INDEX [ CONCURRENTLY ] [ [ IF NOT EXISTS ] <replaceable class=
        </para>
 
        <para>
-        Currently, only the B-tree index access method supports this feature.
+        Currently, B-tree and GiST index access methods supports this feature.
         In B-tree indexes, the values of columns listed in the
         <literal>INCLUDE</literal> clause are included in leaf tuples which
         are linked to the heap tuples, but are not included into pivot tuples
diff --git a/doc/src/sgml/textsearch.sgml b/doc/src/sgml/textsearch.sgml
index 8075ea94e7..c73603e846 100644
--- a/doc/src/sgml/textsearch.sgml
+++ b/doc/src/sgml/textsearch.sgml
@@ -3674,6 +3674,12 @@ SELECT plainto_tsquery('supernovae stars');
    retrieved to see if the match is correct.
   </para>
 
+  <para>
+   A GiST index can be covering, i.e. use <literal>INCLUDE</literal> clause.
+   Included columns can have data types without GiST operator class. Included
+   attributes will be stored uncompressed.
+  </para>
+
   <para>
    Lossiness causes performance degradation due to unnecessary fetches of table
    records that turn out to be false matches.  Since random access to table
diff --git a/src/backend/access/gist/gist.c b/src/backend/access/gist/gist.c
index 8a42effdf7..bbaac5de8c 100644
--- a/src/backend/access/gist/gist.c
+++ b/src/backend/access/gist/gist.c
@@ -74,7 +74,7 @@ gisthandler(PG_FUNCTION_ARGS)
 	amroutine->amclusterable = true;
 	amroutine->ampredlocks = true;
 	amroutine->amcanparallel = false;
-	amroutine->amcaninclude = false;
+	amroutine->amcaninclude = true;
 	amroutine->amkeytype = InvalidOid;
 
 	amroutine->ambuild = gistbuild;
@@ -169,7 +169,7 @@ gistinsert(Relation r, Datum *values, bool *isnull,
 	oldCxt = MemoryContextSwitchTo(giststate->tempCxt);
 
 	itup = gistFormTuple(giststate, r,
-						 values, isnull, true /* size is currently bogus */ );
+						 values, isnull, true /* size is currently bogus */, false);
 	itup->t_tid = *ht_ctid;
 
 	gistdoinsert(r, itup, 0, giststate);
@@ -1377,8 +1377,8 @@ gistSplit(Relation r,
 						IndexTupleSize(itup[0]), GiSTPageSize,
 						RelationGetRelationName(r))));
 
-	memset(v.spl_lisnull, true, sizeof(bool) * giststate->tupdesc->natts);
-	memset(v.spl_risnull, true, sizeof(bool) * giststate->tupdesc->natts);
+	memset(v.spl_lisnull, true, sizeof(bool) * giststate->truncTupdesc->natts);
+	memset(v.spl_risnull, true, sizeof(bool) * giststate->truncTupdesc->natts);
 	gistSplitByKey(r, page, itup, len, giststate, &v, 0);
 
 	/* form left and right vector */
@@ -1401,7 +1401,7 @@ gistSplit(Relation r,
 		ROTATEDIST(res);
 		res->block.num = v.splitVector.spl_nright;
 		res->list = gistfillitupvec(rvectup, v.splitVector.spl_nright, &(res->lenlist));
-		res->itup = gistFormTuple(giststate, r, v.spl_rattr, v.spl_risnull, false);
+		res->itup = gistFormTuple(giststate, r, v.spl_rattr, v.spl_risnull, false, true);
 	}
 
 	if (!gistfitpage(lvectup, v.splitVector.spl_nleft))
@@ -1423,7 +1423,7 @@ gistSplit(Relation r,
 		ROTATEDIST(res);
 		res->block.num = v.splitVector.spl_nleft;
 		res->list = gistfillitupvec(lvectup, v.splitVector.spl_nleft, &(res->lenlist));
-		res->itup = gistFormTuple(giststate, r, v.spl_lattr, v.spl_lisnull, false);
+		res->itup = gistFormTuple(giststate, r, v.spl_lattr, v.spl_lisnull, false, true);
 	}
 
 	return res;
@@ -1457,8 +1457,10 @@ initGISTstate(Relation index)
 	giststate->scanCxt = scanCxt;
 	giststate->tempCxt = scanCxt;	/* caller must change this if needed */
 	giststate->tupdesc = index->rd_att;
+	giststate->truncTupdesc = CreateTupleDescCopyConstr(index->rd_att);
+	giststate->truncTupdesc->natts = IndexRelationGetNumberOfKeyAttributes(index);
 
-	for (i = 0; i < index->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(index); i++)
 	{
 		fmgr_info_copy(&(giststate->consistentFn[i]),
 					   index_getprocinfo(index, i + 1, GIST_CONSISTENT_PROC),
@@ -1526,6 +1528,24 @@ initGISTstate(Relation index)
 			giststate->supportCollation[i] = DEFAULT_COLLATION_OID;
 	}
 
+	for (; i < index->rd_att->natts; i++)
+	{
+		giststate->consistentFn[i].fn_oid = InvalidOid;
+		giststate->unionFn[i].fn_oid = InvalidOid;
+		giststate->compressFn[i].fn_oid = InvalidOid;
+		giststate->decompressFn[i].fn_oid = InvalidOid;
+		giststate->penaltyFn[i].fn_oid = InvalidOid;
+		giststate->picksplitFn[i].fn_oid = InvalidOid;
+		giststate->equalFn[i].fn_oid = InvalidOid;
+		giststate->distanceFn[i].fn_oid = InvalidOid;
+		giststate->fetchFn[i].fn_oid = InvalidOid;
+
+		if (OidIsValid(index->rd_indcollation[i]))
+			giststate->supportCollation[i] = index->rd_indcollation[i];
+		else
+			giststate->supportCollation[i] = DEFAULT_COLLATION_OID;
+	}
+
 	MemoryContextSwitchTo(oldCxt);
 
 	return giststate;
diff --git a/src/backend/access/gist/gistbuild.c b/src/backend/access/gist/gistbuild.c
index 434f15f014..f4545f8069 100644
--- a/src/backend/access/gist/gistbuild.c
+++ b/src/backend/access/gist/gistbuild.c
@@ -469,7 +469,7 @@ gistBuildCallback(Relation index,
 	oldCtx = MemoryContextSwitchTo(buildstate->giststate->tempCxt);
 
 	/* form an index tuple and point it at the heap tuple */
-	itup = gistFormTuple(buildstate->giststate, index, values, isnull, true);
+	itup = gistFormTuple(buildstate->giststate, index, values, isnull, true, false);
 	itup->t_tid = htup->t_self;
 
 	if (buildstate->bufferingMode == GIST_BUFFERING_ACTIVE)
diff --git a/src/backend/access/gist/gistget.c b/src/backend/access/gist/gistget.c
index c4e8a3b913..afb8d31c37 100644
--- a/src/backend/access/gist/gistget.c
+++ b/src/backend/access/gist/gistget.c
@@ -805,12 +805,14 @@ gistgetbitmap(IndexScanDesc scan, TIDBitmap *tbm)
  *
  * Opclasses that implement a fetch function support index-only scans.
  * Opclasses without compression functions also support index-only scans.
+ * Included attributes can be returned.
  */
 bool
 gistcanreturn(Relation index, int attno)
 {
 	if (OidIsValid(index_getprocid(index, attno, GIST_FETCH_PROC)) ||
-		!OidIsValid(index_getprocid(index, attno, GIST_COMPRESS_PROC)))
+		!OidIsValid(index_getprocid(index, attno, GIST_COMPRESS_PROC))||
+		attno > IndexRelationGetNumberOfKeyAttributes(index))
 		return true;
 	else
 		return false;
diff --git a/src/backend/access/gist/gistscan.c b/src/backend/access/gist/gistscan.c
index 4d97ff1d5d..fb93e8d6a6 100644
--- a/src/backend/access/gist/gistscan.c
+++ b/src/backend/access/gist/gistscan.c
@@ -158,6 +158,7 @@ gistrescan(IndexScanDesc scan, ScanKey key, int nkeys,
 	if (scan->xs_want_itup && !scan->xs_hitupdesc)
 	{
 		int			natts;
+		int			nkeyatts;
 		int			attno;
 
 		/*
@@ -167,13 +168,22 @@ gistrescan(IndexScanDesc scan, ScanKey key, int nkeys,
 		 * types.
 		 */
 		natts = RelationGetNumberOfAttributes(scan->indexRelation);
+		nkeyatts = IndexRelationGetNumberOfKeyAttributes(scan->indexRelation);
 		so->giststate->fetchTupdesc = CreateTemplateTupleDesc(natts, false);
-		for (attno = 1; attno <= natts; attno++)
+		for (attno = 1; attno <= nkeyatts; attno++)
 		{
 			TupleDescInitEntry(so->giststate->fetchTupdesc, attno, NULL,
 							   scan->indexRelation->rd_opcintype[attno - 1],
 							   -1, 0);
 		}
+
+		for (; attno <= natts; attno++)
+		{
+			/* taking opcintype from giststate->tupdesc */
+			TupleDescInitEntry(so->giststate->fetchTupdesc, attno, NULL,
+					TupleDescAttr(so->giststate->tupdesc, attno - 1)->atttypid,
+							   -1, 0);
+		}
 		scan->xs_hitupdesc = so->giststate->fetchTupdesc;
 
 		/* Also create a memory context that will hold the returned tuples */
diff --git a/src/backend/access/gist/gistsplit.c b/src/backend/access/gist/gistsplit.c
index a7038cca67..02918536bf 100644
--- a/src/backend/access/gist/gistsplit.c
+++ b/src/backend/access/gist/gistsplit.c
@@ -207,7 +207,7 @@ placeOne(Relation r, GISTSTATE *giststate, GistSplitVector *v,
 	gistDeCompressAtt(giststate, r, itup, NULL, (OffsetNumber) 0,
 					  identry, isnull);
 
-	for (; attno < giststate->tupdesc->natts; attno++)
+	for (; attno < giststate->truncTupdesc->natts; attno++)
 	{
 		float		lpenalty,
 					rpenalty;
@@ -485,7 +485,7 @@ gistUserPicksplit(Relation r, GistEntryVector *entryvec, int attno, GistSplitVec
 	 */
 	v->spl_dontcare = NULL;
 
-	if (attno + 1 < giststate->tupdesc->natts)
+	if (attno + 1 < giststate->truncTupdesc->natts)
 	{
 		int			NumDontCare;
 
@@ -657,7 +657,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 		 */
 		v->spl_risnull[attno] = v->spl_lisnull[attno] = true;
 
-		if (attno + 1 < giststate->tupdesc->natts)
+		if (attno + 1 < giststate->truncTupdesc->natts)
 			gistSplitByKey(r, page, itup, len, giststate, v, attno + 1);
 		else
 			gistSplitHalf(&v->splitVector, len);
@@ -683,7 +683,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 				v->splitVector.spl_left[v->splitVector.spl_nleft++] = i;
 
 		/* Compute union keys, unless outer recursion level will handle it */
-		if (attno == 0 && giststate->tupdesc->natts == 1)
+		if (attno == 0 && giststate->truncTupdesc->natts == 1)
 		{
 			v->spl_dontcare = NULL;
 			gistunionsubkey(giststate, itup, v);
@@ -700,7 +700,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 			 * Splitting on attno column is not optimal, so consider
 			 * redistributing don't-care tuples according to the next column
 			 */
-			Assert(attno + 1 < giststate->tupdesc->natts);
+			Assert(attno + 1 < giststate->truncTupdesc->natts);
 
 			if (v->spl_dontcare == NULL)
 			{
@@ -771,7 +771,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 	 * that PickSplit (or the special cases above) produced correct union
 	 * datums.
 	 */
-	if (attno == 0 && giststate->tupdesc->natts > 1)
+	if (attno == 0 && giststate->truncTupdesc->natts > 1)
 	{
 		v->spl_dontcare = NULL;
 		gistunionsubkey(giststate, itup, v);
diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c
index 55cccd247a..54fce0acf2 100644
--- a/src/backend/access/gist/gistutil.c
+++ b/src/backend/access/gist/gistutil.c
@@ -160,7 +160,7 @@ gistMakeUnionItVec(GISTSTATE *giststate, IndexTuple *itvec, int len,
 
 	evec = (GistEntryVector *) palloc((len + 2) * sizeof(GISTENTRY) + GEVHDRSZ);
 
-	for (i = 0; i < giststate->tupdesc->natts; i++)
+	for (i = 0; i < giststate->truncTupdesc->natts; i++)
 	{
 		int			j;
 
@@ -221,7 +221,7 @@ gistunion(Relation r, IndexTuple *itvec, int len, GISTSTATE *giststate)
 
 	gistMakeUnionItVec(giststate, itvec, len, attr, isnull);
 
-	return gistFormTuple(giststate, r, attr, isnull, false);
+	return gistFormTuple(giststate, r, attr, isnull, false, true);
 }
 
 /*
@@ -296,7 +296,7 @@ gistDeCompressAtt(GISTSTATE *giststate, Relation r, IndexTuple tuple, Page p,
 {
 	int			i;
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		Datum		datum;
 
@@ -329,7 +329,7 @@ gistgetadjusted(Relation r, IndexTuple oldtup, IndexTuple addtup, GISTSTATE *gis
 	gistDeCompressAtt(giststate, r, addtup, NULL,
 					  (OffsetNumber) 0, addentries, addisnull);
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		gistMakeUnionKey(giststate, i,
 						 oldentries + i, oldisnull[i],
@@ -355,7 +355,7 @@ gistgetadjusted(Relation r, IndexTuple oldtup, IndexTuple addtup, GISTSTATE *gis
 	if (neednew)
 	{
 		/* need to update key */
-		newtup = gistFormTuple(giststate, r, attr, isnull, false);
+		newtup = gistFormTuple(giststate, r, attr, isnull, false, true);
 		newtup->t_tid = oldtup->t_tid;
 	}
 
@@ -442,7 +442,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 		zero_penalty = true;
 
 		/* Loop over index attributes. */
-		for (j = 0; j < r->rd_att->natts; j++)
+		for (j = 0; j < IndexRelationGetNumberOfKeyAttributes(r); j++)
 		{
 			Datum		datum;
 			float		usize;
@@ -470,7 +470,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 				result = i;
 				best_penalty[j] = usize;
 
-				if (j < r->rd_att->natts - 1)
+				if (j < IndexRelationGetNumberOfKeyAttributes(r) - 1)
 					best_penalty[j + 1] = -1;
 
 				/* we have new best, so reset keep-it decision */
@@ -500,7 +500,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 		 * If we looped past the last column, and did not update "result",
 		 * then this tuple is exactly as good as the prior best tuple.
 		 */
-		if (j == r->rd_att->natts && result != i)
+		if (j == IndexRelationGetNumberOfKeyAttributes(r) && result != i)
 		{
 			if (keep_current_best == -1)
 			{
@@ -570,7 +570,7 @@ gistdentryinit(GISTSTATE *giststate, int nkey, GISTENTRY *e,
 
 IndexTuple
 gistFormTuple(GISTSTATE *giststate, Relation r,
-			  Datum attdata[], bool isnull[], bool isleaf)
+			  Datum attdata[], bool isnull[], bool isleaf, bool isTruncated)
 {
 	Datum		compatt[INDEX_MAX_KEYS];
 	int			i;
@@ -579,7 +579,7 @@ gistFormTuple(GISTSTATE *giststate, Relation r,
 	/*
 	 * Call the compress method on each attribute.
 	 */
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		if (isnull[i])
 			compatt[i] = (Datum) 0;
@@ -602,7 +602,23 @@ gistFormTuple(GISTSTATE *giststate, Relation r,
 		}
 	}
 
-	res = index_form_tuple(giststate->tupdesc, compatt, isnull);
+	if (!isTruncated)
+	{
+		/*
+		 * Allocate each included attribute.
+		 */
+		for (; i < r->rd_att->natts; i++)
+		{
+			if (isnull[i])
+				compatt[i] = (Datum) 0;
+			else
+			{
+				compatt[i] = attdata[i];
+			}
+		}
+	}
+
+	res = index_form_tuple(isTruncated?giststate->truncTupdesc:giststate->tupdesc, compatt, isnull);
 
 	/*
 	 * The offset number on tuples on internal pages is unused. For historical
@@ -644,7 +660,7 @@ gistFetchTuple(GISTSTATE *giststate, Relation r, IndexTuple tuple)
 	bool		isnull[INDEX_MAX_KEYS];
 	int			i;
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		Datum		datum;
 
@@ -679,6 +695,14 @@ gistFetchTuple(GISTSTATE *giststate, Relation r, IndexTuple tuple)
 			fetchatt[i] = (Datum) 0;
 		}
 	}
+
+	/*
+	 * Get each included attribute.
+	 */
+	for (; i < r->rd_att->natts; i++)
+	{
+		fetchatt[i] = index_getattr(tuple, i + 1, giststate->tupdesc, &isnull[i]);
+	}
 	MemoryContextSwitchTo(oldcxt);
 
 	return heap_form_tuple(giststate->fetchTupdesc, fetchatt, isnull);
diff --git a/src/include/access/gist_private.h b/src/include/access/gist_private.h
index 36ed7244ba..c1714a6ccc 100644
--- a/src/include/access/gist_private.h
+++ b/src/include/access/gist_private.h
@@ -79,6 +79,8 @@ typedef struct GISTSTATE
 	MemoryContext tempCxt;		/* short-term context for calling functions */
 
 	TupleDesc	tupdesc;		/* index's tuple descriptor */
+	TupleDesc	truncTupdesc;	/* truncated tuple descriptor
+								 * for internal pages */
 	TupleDesc	fetchTupdesc;	/* tuple descriptor for tuples returned in an
 								 * index-only scan */
 
@@ -458,7 +460,8 @@ extern IndexTuple gistgetadjusted(Relation r,
 				IndexTuple addtup,
 				GISTSTATE *giststate);
 extern IndexTuple gistFormTuple(GISTSTATE *giststate,
-			  Relation r, Datum *attdata, bool *isnull, bool isleaf);
+			  Relation r, Datum *attdata, bool *isnull, bool isleaf,
+			  bool isTruncated);
 
 extern OffsetNumber gistchoose(Relation r, Page p,
 		   IndexTuple it,
diff --git a/src/test/regress/expected/amutils.out b/src/test/regress/expected/amutils.out
index 24cd3c5e2e..0c481f94ab 100644
--- a/src/test/regress/expected/amutils.out
+++ b/src/test/regress/expected/amutils.out
@@ -75,7 +75,7 @@ select prop,
  can_unique         | f  |       | 
  can_multi_col      | t  |       | 
  can_exclude        | t  |       | 
- can_include        | f  |       | 
+ can_include        | t  |       | 
  bogus              |    |       | 
 (19 rows)
 
@@ -158,7 +158,7 @@ select amname, prop, pg_indexam_has_property(a.oid, prop) as p
  gist   | can_unique    | f
  gist   | can_multi_col | t
  gist   | can_exclude   | t
- gist   | can_include   | f
+ gist   | can_include   | t
  gist   | bogus         | 
  hash   | can_order     | f
  hash   | can_unique    | f
diff --git a/src/test/regress/expected/index_including.out b/src/test/regress/expected/index_including.out
index b7d1812ec5..cf8fae8474 100644
--- a/src/test/regress/expected/index_including.out
+++ b/src/test/regress/expected/index_including.out
@@ -289,22 +289,20 @@ SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl' ORDER BY indexname;
 
 DROP TABLE tbl;
 /*
- * 7. Check various AMs. All but btree must fail.
+ * 7. Check various AMs. All but btree and gist must fail.
  */
 CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
 CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "brin" does not support included columns
-CREATE INDEX on tbl USING gist(c3) INCLUDE (c4);
-ERROR:  access method "gist" does not support included columns
+CREATE INDEX on tbl USING gist(c4) INCLUDE (c3);
 CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
 ERROR:  access method "spgist" does not support included columns
 CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "gin" does not support included columns
 CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "hash" does not support included columns
-CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
+CREATE INDEX on tbl USING rtree(c4) INCLUDE (c3, c1);
 NOTICE:  substituting access method "gist" for obsolete method "rtree"
-ERROR:  access method "gist" does not support included columns
 CREATE INDEX on tbl USING btree(c1, c2) INCLUDE (c3, c4);
 DROP TABLE tbl;
 /*
diff --git a/src/test/regress/expected/index_including_gist.out b/src/test/regress/expected/index_including_gist.out
new file mode 100644
index 0000000000..5b87da9bb0
--- /dev/null
+++ b/src/test/regress/expected/index_including_gist.out
@@ -0,0 +1,145 @@
+/*
+ * 1.1. test CREATE INDEX with buffered build
+ */
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+-- must fail because of intersection of key and included columns
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c4);
+ERROR:  included columns must not intersect with key columns
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+                                  pg_get_indexdef                                  
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+ c1 | c2 | c3 |     c4      
+----+----+----+-------------
+  1 |  2 |  3 | (2,3),(1,2)
+  2 |  4 |  6 | (4,5),(2,3)
+  3 |  6 |  9 | (6,7),(3,4)
+  4 |  8 | 12 | (8,9),(4,5)
+(4 rows)
+
+SET enable_bitmapscan TO off;
+EXPLAIN SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                                     QUERY PLAN                                     
+------------------------------------------------------------------------------------
+ Index Only Scan using tbl_gist_idx on tbl_gist  (cost=0.27..36.41 rows=8 width=44)
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+/*
+ * 1.2. test CREATE INDEX with inserts
+ */
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+-- must fail because of intersection of key and included columns
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c4);
+ERROR:  included columns must not intersect with key columns
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+                                  pg_get_indexdef                                  
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+ c1 | c2 | c3 |     c4      
+----+----+----+-------------
+  1 |  2 |  3 | (2,3),(1,2)
+  2 |  4 |  6 | (4,5),(2,3)
+  3 |  6 |  9 | (6,7),(3,4)
+  4 |  8 | 12 | (8,9),(4,5)
+(4 rows)
+
+SET enable_bitmapscan TO off;
+EXPLAIN SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                                     QUERY PLAN                                     
+------------------------------------------------------------------------------------
+ Index Only Scan using tbl_gist_idx on tbl_gist  (cost=0.28..36.41 rows=8 width=44)
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+/*
+ * 2. CREATE INDEX CONCURRENTLY
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX CONCURRENTLY tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                     indexdef                                      
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+DROP TABLE tbl_gist;
+/*
+ * 3. REINDEX
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                   indexdef                                    
+-------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c3)
+(1 row)
+
+REINDEX INDEX tbl_gist_idx;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                   indexdef                                    
+-------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c3)
+(1 row)
+
+ALTER TABLE tbl_gist DROP COLUMN c1;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+ indexdef 
+----------
+(0 rows)
+
+DROP TABLE tbl_gist;
+/*
+ * 4. Update, delete values in indexed table.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+UPDATE tbl_gist SET c1 = 100 WHERE c1 = 2;
+UPDATE tbl_gist SET c1 = 1 WHERE c1 = 3;
+DELETE FROM tbl_gist WHERE c1 = 5 OR c3 = 12;
+DROP TABLE tbl_gist;
+/*
+ * 5. Alter column type.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+ALTER TABLE tbl_gist ALTER c1 TYPE bigint;
+ALTER TABLE tbl_gist ALTER c3 TYPE bigint;
+\d tbl_gist
+              Table "public.tbl_gist"
+ Column |  Type   | Collation | Nullable | Default 
+--------+---------+-----------+----------+---------
+ c1     | bigint  |           |          | 
+ c2     | integer |           |          | 
+ c3     | bigint  |           |          | 
+ c4     | box     |           |          | 
+Indexes:
+    "tbl_gist_idx" gist (c4) INCLUDE (c1, c3)
+
+DROP TABLE tbl_gist;
diff --git a/src/test/regress/parallel_schedule b/src/test/regress/parallel_schedule
index 16f979c8d9..a6db2f3abb 100644
--- a/src/test/regress/parallel_schedule
+++ b/src/test/regress/parallel_schedule
@@ -55,7 +55,7 @@ test: copy copyselect copydml
 # ----------
 test: create_misc create_operator create_procedure
 # These depend on the above two
-test: create_index create_view index_including
+test: create_index create_view index_including index_including_gist
 
 # ----------
 # Another group of parallel tests
diff --git a/src/test/regress/serial_schedule b/src/test/regress/serial_schedule
index 42632be675..71ed5efb19 100644
--- a/src/test/regress/serial_schedule
+++ b/src/test/regress/serial_schedule
@@ -66,6 +66,7 @@ test: create_operator
 test: create_procedure
 test: create_index
 test: index_including
+test: index_including_gist
 test: create_view
 test: create_aggregate
 test: create_function_3
diff --git a/src/test/regress/sql/index_including.sql b/src/test/regress/sql/index_including.sql
index b71bcaf936..10374545ba 100644
--- a/src/test/regress/sql/index_including.sql
+++ b/src/test/regress/sql/index_including.sql
@@ -162,15 +162,15 @@ SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl' ORDER BY indexname;
 DROP TABLE tbl;
 
 /*
- * 7. Check various AMs. All but btree must fail.
+ * 7. Check various AMs. All but btree and gist must fail.
  */
 CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
 CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
-CREATE INDEX on tbl USING gist(c3) INCLUDE (c4);
+CREATE INDEX on tbl USING gist(c4) INCLUDE (c3);
 CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
 CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
 CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
-CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
+CREATE INDEX on tbl USING rtree(c4) INCLUDE (c3, c1);
 CREATE INDEX on tbl USING btree(c1, c2) INCLUDE (c3, c4);
 DROP TABLE tbl;
 
diff --git a/src/test/regress/sql/index_including_gist.sql b/src/test/regress/sql/index_including_gist.sql
new file mode 100644
index 0000000000..cd7bc2c422
--- /dev/null
+++ b/src/test/regress/sql/index_including_gist.sql
@@ -0,0 +1,84 @@
+/*
+ * 1.1. test CREATE INDEX with buffered build
+ */
+
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+-- must fail because of intersection of key and included columns
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c4);
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO off;
+EXPLAIN SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+
+/*
+ * 1.2. test CREATE INDEX with inserts
+ */
+
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+-- must fail because of intersection of key and included columns
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c4);
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO off;
+EXPLAIN SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+
+/*
+ * 2. CREATE INDEX CONCURRENTLY
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX CONCURRENTLY tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+DROP TABLE tbl_gist;
+
+
+/*
+ * 3. REINDEX
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+REINDEX INDEX tbl_gist_idx;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+ALTER TABLE tbl_gist DROP COLUMN c1;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+DROP TABLE tbl_gist;
+
+/*
+ * 4. Update, delete values in indexed table.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+UPDATE tbl_gist SET c1 = 100 WHERE c1 = 2;
+UPDATE tbl_gist SET c1 = 1 WHERE c1 = 3;
+DELETE FROM tbl_gist WHERE c1 = 5 OR c3 = 12;
+DROP TABLE tbl_gist;
+
+/*
+ * 5. Alter column type.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+ALTER TABLE tbl_gist ALTER c1 TYPE bigint;
+ALTER TABLE tbl_gist ALTER c3 TYPE bigint;
+\d tbl_gist
+DROP TABLE tbl_gist;
-- 
2.15.2 (Apple Git-101.1)

#10

Thomas Munro

thomas.munro@enterprisedb.com

over 7 years ago

In reply to: Andrey Borodin (#9)

Re: Covering GiST indexes

On Fri, Jul 6, 2018 at 5:27 AM, Andrey Borodin <x4mmm@yandex-team.ru> wrote:

Here is v3 version of the patch. I've fixed some comments and added some words to docs.

Hi again Andrey,

Cfbot reported the following difference (twice) in the
index_including_gist test:

- ERROR: included columns must not intersect with key columns
+ ERROR: relation "tbl_gist_idx" already exists

Well, both of those errors are valid complaints in this case... maybe
the order of checks changed in master since you wrote the patch?
Probably better to use a different name for the index anyway.

--
Thomas Munro
http://www.enterprisedb.com

#11

[0]: https://github.com/postgres/postgres/commit/701fd0bbc98fe8211d36e96f90753985104cd295 <https://github.com/postgres/postgres/commit/701fd0bbc98fe8211d36e96f90753985104cd295>

x4mmm@yandex-team.ru

over 7 years ago

In reply to: Thomas Munro (#10)

1 attachment(s)

Re: Covering GiST indexes

Hi, Thomas!

29 июля 2018 г., в 14:28, Thomas Munro <thomas.munro@enterprisedb.com> написал(а):

On Fri, Jul 6, 2018 at 5:27 AM, Andrey Borodin <x4mmm@yandex-team.ru> wrote:

Here is v3 version of the patch. I've fixed some comments and added some words to docs.

Hi again Andrey,

Cfbot reported the following difference (twice) in the
index_including_gist test:
- ERROR: included columns must not intersect with key columns
+ ERROR: relation "tbl_gist_idx" already exists
Well, both of those errors are valid complaints in this case... maybe
the order of checks changed in master since you wrote the patch?
Probably better to use a different name for the index anyway.

Thanks! The problem appeared with commit 701fd0b [0]https://github.com/postgres/postgres/commit/701fd0bbc98fe8211d36e96f90753985104cd295 <https://github.com/postgres/postgres/commit/701fd0bbc98fe8211d36e96f90753985104cd295> which dropped validation rules checked in failed test. Here's the patch with fixed tests.

Best regards, Andrey Borodin.

Attachments:

0001-Covering-GiST-v4.patchapplication/octet-stream; name=0001-Covering-GiST-v4.patch; x-unix-mode=0644Download

From ed23a403be4076050a5eb0f9c1c2e34b7f9f36df Mon Sep 17 00:00:00 2001
From: Andrey Borodin <amborodin@acm.org>
Date: Sun, 29 Jul 2018 15:44:12 +0500
Subject: [PATCH] Covering GiST v4 Allow GiST index access method to include
 attributes without opclasses. Included attributes are stored uncompressed and
 cannot be used for index search efficiently, but can participate in Index
 Only Scan.

---
 doc/src/sgml/ref/create_index.sgml                 |   2 +-
 doc/src/sgml/textsearch.sgml                       |   6 +
 src/backend/access/gist/gist.c                     |  34 +++--
 src/backend/access/gist/gistbuild.c                |   2 +-
 src/backend/access/gist/gistget.c                  |   4 +-
 src/backend/access/gist/gistscan.c                 |  12 +-
 src/backend/access/gist/gistsplit.c                |  12 +-
 src/backend/access/gist/gistutil.c                 |  48 +++++--
 src/include/access/gist_private.h                  |   5 +-
 src/test/regress/expected/amutils.out              |   4 +-
 src/test/regress/expected/index_including.out      |   8 +-
 src/test/regress/expected/index_including_gist.out | 139 +++++++++++++++++++++
 src/test/regress/parallel_schedule                 |   2 +-
 src/test/regress/serial_schedule                   |   1 +
 src/test/regress/sql/index_including.sql           |   6 +-
 src/test/regress/sql/index_including_gist.sql      |  80 ++++++++++++
 16 files changed, 324 insertions(+), 41 deletions(-)
 create mode 100644 src/test/regress/expected/index_including_gist.out
 create mode 100644 src/test/regress/sql/index_including_gist.sql

diff --git a/doc/src/sgml/ref/create_index.sgml b/doc/src/sgml/ref/create_index.sgml
index 3c1223b324..7adc8852e8 100644
--- a/doc/src/sgml/ref/create_index.sgml
+++ b/doc/src/sgml/ref/create_index.sgml
@@ -181,7 +181,7 @@ CREATE [ UNIQUE ] INDEX [ CONCURRENTLY ] [ [ IF NOT EXISTS ] <replaceable class=
        </para>
 
        <para>
-        Currently, only the B-tree index access method supports this feature.
+        Currently, B-tree and GiST index access methods supports this feature.
         In B-tree indexes, the values of columns listed in the
         <literal>INCLUDE</literal> clause are included in leaf tuples which
         correspond to heap tuples, but are not included in upper-level
diff --git a/doc/src/sgml/textsearch.sgml b/doc/src/sgml/textsearch.sgml
index 6df424c63e..b3f9c1800b 100644
--- a/doc/src/sgml/textsearch.sgml
+++ b/doc/src/sgml/textsearch.sgml
@@ -3674,6 +3674,12 @@ SELECT plainto_tsquery('supernovae stars');
    retrieved to see if the match is correct.
   </para>
 
+  <para>
+   A GiST index can be covering, i.e. use <literal>INCLUDE</literal> clause.
+   Included columns can have data types without GiST operator class. Included
+   attributes will be stored uncompressed.
+  </para>
+
   <para>
    Lossiness causes performance degradation due to unnecessary fetches of table
    records that turn out to be false matches.  Since random access to table
diff --git a/src/backend/access/gist/gist.c b/src/backend/access/gist/gist.c
index 8a42effdf7..bbaac5de8c 100644
--- a/src/backend/access/gist/gist.c
+++ b/src/backend/access/gist/gist.c
@@ -74,7 +74,7 @@ gisthandler(PG_FUNCTION_ARGS)
 	amroutine->amclusterable = true;
 	amroutine->ampredlocks = true;
 	amroutine->amcanparallel = false;
-	amroutine->amcaninclude = false;
+	amroutine->amcaninclude = true;
 	amroutine->amkeytype = InvalidOid;
 
 	amroutine->ambuild = gistbuild;
@@ -169,7 +169,7 @@ gistinsert(Relation r, Datum *values, bool *isnull,
 	oldCxt = MemoryContextSwitchTo(giststate->tempCxt);
 
 	itup = gistFormTuple(giststate, r,
-						 values, isnull, true /* size is currently bogus */ );
+						 values, isnull, true /* size is currently bogus */, false);
 	itup->t_tid = *ht_ctid;
 
 	gistdoinsert(r, itup, 0, giststate);
@@ -1377,8 +1377,8 @@ gistSplit(Relation r,
 						IndexTupleSize(itup[0]), GiSTPageSize,
 						RelationGetRelationName(r))));
 
-	memset(v.spl_lisnull, true, sizeof(bool) * giststate->tupdesc->natts);
-	memset(v.spl_risnull, true, sizeof(bool) * giststate->tupdesc->natts);
+	memset(v.spl_lisnull, true, sizeof(bool) * giststate->truncTupdesc->natts);
+	memset(v.spl_risnull, true, sizeof(bool) * giststate->truncTupdesc->natts);
 	gistSplitByKey(r, page, itup, len, giststate, &v, 0);
 
 	/* form left and right vector */
@@ -1401,7 +1401,7 @@ gistSplit(Relation r,
 		ROTATEDIST(res);
 		res->block.num = v.splitVector.spl_nright;
 		res->list = gistfillitupvec(rvectup, v.splitVector.spl_nright, &(res->lenlist));
-		res->itup = gistFormTuple(giststate, r, v.spl_rattr, v.spl_risnull, false);
+		res->itup = gistFormTuple(giststate, r, v.spl_rattr, v.spl_risnull, false, true);
 	}
 
 	if (!gistfitpage(lvectup, v.splitVector.spl_nleft))
@@ -1423,7 +1423,7 @@ gistSplit(Relation r,
 		ROTATEDIST(res);
 		res->block.num = v.splitVector.spl_nleft;
 		res->list = gistfillitupvec(lvectup, v.splitVector.spl_nleft, &(res->lenlist));
-		res->itup = gistFormTuple(giststate, r, v.spl_lattr, v.spl_lisnull, false);
+		res->itup = gistFormTuple(giststate, r, v.spl_lattr, v.spl_lisnull, false, true);
 	}
 
 	return res;
@@ -1457,8 +1457,10 @@ initGISTstate(Relation index)
 	giststate->scanCxt = scanCxt;
 	giststate->tempCxt = scanCxt;	/* caller must change this if needed */
 	giststate->tupdesc = index->rd_att;
+	giststate->truncTupdesc = CreateTupleDescCopyConstr(index->rd_att);
+	giststate->truncTupdesc->natts = IndexRelationGetNumberOfKeyAttributes(index);
 
-	for (i = 0; i < index->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(index); i++)
 	{
 		fmgr_info_copy(&(giststate->consistentFn[i]),
 					   index_getprocinfo(index, i + 1, GIST_CONSISTENT_PROC),
@@ -1526,6 +1528,24 @@ initGISTstate(Relation index)
 			giststate->supportCollation[i] = DEFAULT_COLLATION_OID;
 	}
 
+	for (; i < index->rd_att->natts; i++)
+	{
+		giststate->consistentFn[i].fn_oid = InvalidOid;
+		giststate->unionFn[i].fn_oid = InvalidOid;
+		giststate->compressFn[i].fn_oid = InvalidOid;
+		giststate->decompressFn[i].fn_oid = InvalidOid;
+		giststate->penaltyFn[i].fn_oid = InvalidOid;
+		giststate->picksplitFn[i].fn_oid = InvalidOid;
+		giststate->equalFn[i].fn_oid = InvalidOid;
+		giststate->distanceFn[i].fn_oid = InvalidOid;
+		giststate->fetchFn[i].fn_oid = InvalidOid;
+
+		if (OidIsValid(index->rd_indcollation[i]))
+			giststate->supportCollation[i] = index->rd_indcollation[i];
+		else
+			giststate->supportCollation[i] = DEFAULT_COLLATION_OID;
+	}
+
 	MemoryContextSwitchTo(oldCxt);
 
 	return giststate;
diff --git a/src/backend/access/gist/gistbuild.c b/src/backend/access/gist/gistbuild.c
index 434f15f014..f4545f8069 100644
--- a/src/backend/access/gist/gistbuild.c
+++ b/src/backend/access/gist/gistbuild.c
@@ -469,7 +469,7 @@ gistBuildCallback(Relation index,
 	oldCtx = MemoryContextSwitchTo(buildstate->giststate->tempCxt);
 
 	/* form an index tuple and point it at the heap tuple */
-	itup = gistFormTuple(buildstate->giststate, index, values, isnull, true);
+	itup = gistFormTuple(buildstate->giststate, index, values, isnull, true, false);
 	itup->t_tid = htup->t_self;
 
 	if (buildstate->bufferingMode == GIST_BUFFERING_ACTIVE)
diff --git a/src/backend/access/gist/gistget.c b/src/backend/access/gist/gistget.c
index ad07b9e63c..c75a6a7b53 100644
--- a/src/backend/access/gist/gistget.c
+++ b/src/backend/access/gist/gistget.c
@@ -805,12 +805,14 @@ gistgetbitmap(IndexScanDesc scan, TIDBitmap *tbm)
  *
  * Opclasses that implement a fetch function support index-only scans.
  * Opclasses without compression functions also support index-only scans.
+ * Included attributes can be returned.
  */
 bool
 gistcanreturn(Relation index, int attno)
 {
 	if (OidIsValid(index_getprocid(index, attno, GIST_FETCH_PROC)) ||
-		!OidIsValid(index_getprocid(index, attno, GIST_COMPRESS_PROC)))
+		!OidIsValid(index_getprocid(index, attno, GIST_COMPRESS_PROC))||
+		attno > IndexRelationGetNumberOfKeyAttributes(index))
 		return true;
 	else
 		return false;
diff --git a/src/backend/access/gist/gistscan.c b/src/backend/access/gist/gistscan.c
index 4d97ff1d5d..fb93e8d6a6 100644
--- a/src/backend/access/gist/gistscan.c
+++ b/src/backend/access/gist/gistscan.c
@@ -158,6 +158,7 @@ gistrescan(IndexScanDesc scan, ScanKey key, int nkeys,
 	if (scan->xs_want_itup && !scan->xs_hitupdesc)
 	{
 		int			natts;
+		int			nkeyatts;
 		int			attno;
 
 		/*
@@ -167,13 +168,22 @@ gistrescan(IndexScanDesc scan, ScanKey key, int nkeys,
 		 * types.
 		 */
 		natts = RelationGetNumberOfAttributes(scan->indexRelation);
+		nkeyatts = IndexRelationGetNumberOfKeyAttributes(scan->indexRelation);
 		so->giststate->fetchTupdesc = CreateTemplateTupleDesc(natts, false);
-		for (attno = 1; attno <= natts; attno++)
+		for (attno = 1; attno <= nkeyatts; attno++)
 		{
 			TupleDescInitEntry(so->giststate->fetchTupdesc, attno, NULL,
 							   scan->indexRelation->rd_opcintype[attno - 1],
 							   -1, 0);
 		}
+
+		for (; attno <= natts; attno++)
+		{
+			/* taking opcintype from giststate->tupdesc */
+			TupleDescInitEntry(so->giststate->fetchTupdesc, attno, NULL,
+					TupleDescAttr(so->giststate->tupdesc, attno - 1)->atttypid,
+							   -1, 0);
+		}
 		scan->xs_hitupdesc = so->giststate->fetchTupdesc;
 
 		/* Also create a memory context that will hold the returned tuples */
diff --git a/src/backend/access/gist/gistsplit.c b/src/backend/access/gist/gistsplit.c
index a7038cca67..02918536bf 100644
--- a/src/backend/access/gist/gistsplit.c
+++ b/src/backend/access/gist/gistsplit.c
@@ -207,7 +207,7 @@ placeOne(Relation r, GISTSTATE *giststate, GistSplitVector *v,
 	gistDeCompressAtt(giststate, r, itup, NULL, (OffsetNumber) 0,
 					  identry, isnull);
 
-	for (; attno < giststate->tupdesc->natts; attno++)
+	for (; attno < giststate->truncTupdesc->natts; attno++)
 	{
 		float		lpenalty,
 					rpenalty;
@@ -485,7 +485,7 @@ gistUserPicksplit(Relation r, GistEntryVector *entryvec, int attno, GistSplitVec
 	 */
 	v->spl_dontcare = NULL;
 
-	if (attno + 1 < giststate->tupdesc->natts)
+	if (attno + 1 < giststate->truncTupdesc->natts)
 	{
 		int			NumDontCare;
 
@@ -657,7 +657,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 		 */
 		v->spl_risnull[attno] = v->spl_lisnull[attno] = true;
 
-		if (attno + 1 < giststate->tupdesc->natts)
+		if (attno + 1 < giststate->truncTupdesc->natts)
 			gistSplitByKey(r, page, itup, len, giststate, v, attno + 1);
 		else
 			gistSplitHalf(&v->splitVector, len);
@@ -683,7 +683,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 				v->splitVector.spl_left[v->splitVector.spl_nleft++] = i;
 
 		/* Compute union keys, unless outer recursion level will handle it */
-		if (attno == 0 && giststate->tupdesc->natts == 1)
+		if (attno == 0 && giststate->truncTupdesc->natts == 1)
 		{
 			v->spl_dontcare = NULL;
 			gistunionsubkey(giststate, itup, v);
@@ -700,7 +700,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 			 * Splitting on attno column is not optimal, so consider
 			 * redistributing don't-care tuples according to the next column
 			 */
-			Assert(attno + 1 < giststate->tupdesc->natts);
+			Assert(attno + 1 < giststate->truncTupdesc->natts);
 
 			if (v->spl_dontcare == NULL)
 			{
@@ -771,7 +771,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 	 * that PickSplit (or the special cases above) produced correct union
 	 * datums.
 	 */
-	if (attno == 0 && giststate->tupdesc->natts > 1)
+	if (attno == 0 && giststate->truncTupdesc->natts > 1)
 	{
 		v->spl_dontcare = NULL;
 		gistunionsubkey(giststate, itup, v);
diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c
index dddfe0ae2c..f87e84a5f1 100644
--- a/src/backend/access/gist/gistutil.c
+++ b/src/backend/access/gist/gistutil.c
@@ -159,7 +159,7 @@ gistMakeUnionItVec(GISTSTATE *giststate, IndexTuple *itvec, int len,
 
 	evec = (GistEntryVector *) palloc((len + 2) * sizeof(GISTENTRY) + GEVHDRSZ);
 
-	for (i = 0; i < giststate->tupdesc->natts; i++)
+	for (i = 0; i < giststate->truncTupdesc->natts; i++)
 	{
 		int			j;
 
@@ -220,7 +220,7 @@ gistunion(Relation r, IndexTuple *itvec, int len, GISTSTATE *giststate)
 
 	gistMakeUnionItVec(giststate, itvec, len, attr, isnull);
 
-	return gistFormTuple(giststate, r, attr, isnull, false);
+	return gistFormTuple(giststate, r, attr, isnull, false, true);
 }
 
 /*
@@ -295,7 +295,7 @@ gistDeCompressAtt(GISTSTATE *giststate, Relation r, IndexTuple tuple, Page p,
 {
 	int			i;
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		Datum		datum;
 
@@ -328,7 +328,7 @@ gistgetadjusted(Relation r, IndexTuple oldtup, IndexTuple addtup, GISTSTATE *gis
 	gistDeCompressAtt(giststate, r, addtup, NULL,
 					  (OffsetNumber) 0, addentries, addisnull);
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		gistMakeUnionKey(giststate, i,
 						 oldentries + i, oldisnull[i],
@@ -354,7 +354,7 @@ gistgetadjusted(Relation r, IndexTuple oldtup, IndexTuple addtup, GISTSTATE *gis
 	if (neednew)
 	{
 		/* need to update key */
-		newtup = gistFormTuple(giststate, r, attr, isnull, false);
+		newtup = gistFormTuple(giststate, r, attr, isnull, false, true);
 		newtup->t_tid = oldtup->t_tid;
 	}
 
@@ -441,7 +441,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 		zero_penalty = true;
 
 		/* Loop over index attributes. */
-		for (j = 0; j < r->rd_att->natts; j++)
+		for (j = 0; j < IndexRelationGetNumberOfKeyAttributes(r); j++)
 		{
 			Datum		datum;
 			float		usize;
@@ -469,7 +469,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 				result = i;
 				best_penalty[j] = usize;
 
-				if (j < r->rd_att->natts - 1)
+				if (j < IndexRelationGetNumberOfKeyAttributes(r) - 1)
 					best_penalty[j + 1] = -1;
 
 				/* we have new best, so reset keep-it decision */
@@ -499,7 +499,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 		 * If we looped past the last column, and did not update "result",
 		 * then this tuple is exactly as good as the prior best tuple.
 		 */
-		if (j == r->rd_att->natts && result != i)
+		if (j == IndexRelationGetNumberOfKeyAttributes(r) && result != i)
 		{
 			if (keep_current_best == -1)
 			{
@@ -569,7 +569,7 @@ gistdentryinit(GISTSTATE *giststate, int nkey, GISTENTRY *e,
 
 IndexTuple
 gistFormTuple(GISTSTATE *giststate, Relation r,
-			  Datum attdata[], bool isnull[], bool isleaf)
+			  Datum attdata[], bool isnull[], bool isleaf, bool isTruncated)
 {
 	Datum		compatt[INDEX_MAX_KEYS];
 	int			i;
@@ -578,7 +578,7 @@ gistFormTuple(GISTSTATE *giststate, Relation r,
 	/*
 	 * Call the compress method on each attribute.
 	 */
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		if (isnull[i])
 			compatt[i] = (Datum) 0;
@@ -601,7 +601,23 @@ gistFormTuple(GISTSTATE *giststate, Relation r,
 		}
 	}
 
-	res = index_form_tuple(giststate->tupdesc, compatt, isnull);
+	if (!isTruncated)
+	{
+		/*
+		 * Allocate each included attribute.
+		 */
+		for (; i < r->rd_att->natts; i++)
+		{
+			if (isnull[i])
+				compatt[i] = (Datum) 0;
+			else
+			{
+				compatt[i] = attdata[i];
+			}
+		}
+	}
+
+	res = index_form_tuple(isTruncated?giststate->truncTupdesc:giststate->tupdesc, compatt, isnull);
 
 	/*
 	 * The offset number on tuples on internal pages is unused. For historical
@@ -643,7 +659,7 @@ gistFetchTuple(GISTSTATE *giststate, Relation r, IndexTuple tuple)
 	bool		isnull[INDEX_MAX_KEYS];
 	int			i;
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		Datum		datum;
 
@@ -678,6 +694,14 @@ gistFetchTuple(GISTSTATE *giststate, Relation r, IndexTuple tuple)
 			fetchatt[i] = (Datum) 0;
 		}
 	}
+
+	/*
+	 * Get each included attribute.
+	 */
+	for (; i < r->rd_att->natts; i++)
+	{
+		fetchatt[i] = index_getattr(tuple, i + 1, giststate->tupdesc, &isnull[i]);
+	}
 	MemoryContextSwitchTo(oldcxt);
 
 	return heap_form_tuple(giststate->fetchTupdesc, fetchatt, isnull);
diff --git a/src/include/access/gist_private.h b/src/include/access/gist_private.h
index 36ed7244ba..c1714a6ccc 100644
--- a/src/include/access/gist_private.h
+++ b/src/include/access/gist_private.h
@@ -79,6 +79,8 @@ typedef struct GISTSTATE
 	MemoryContext tempCxt;		/* short-term context for calling functions */
 
 	TupleDesc	tupdesc;		/* index's tuple descriptor */
+	TupleDesc	truncTupdesc;	/* truncated tuple descriptor
+								 * for internal pages */
 	TupleDesc	fetchTupdesc;	/* tuple descriptor for tuples returned in an
 								 * index-only scan */
 
@@ -458,7 +460,8 @@ extern IndexTuple gistgetadjusted(Relation r,
 				IndexTuple addtup,
 				GISTSTATE *giststate);
 extern IndexTuple gistFormTuple(GISTSTATE *giststate,
-			  Relation r, Datum *attdata, bool *isnull, bool isleaf);
+			  Relation r, Datum *attdata, bool *isnull, bool isleaf,
+			  bool isTruncated);
 
 extern OffsetNumber gistchoose(Relation r, Page p,
 		   IndexTuple it,
diff --git a/src/test/regress/expected/amutils.out b/src/test/regress/expected/amutils.out
index 24cd3c5e2e..0c481f94ab 100644
--- a/src/test/regress/expected/amutils.out
+++ b/src/test/regress/expected/amutils.out
@@ -75,7 +75,7 @@ select prop,
  can_unique         | f  |       | 
  can_multi_col      | t  |       | 
  can_exclude        | t  |       | 
- can_include        | f  |       | 
+ can_include        | t  |       | 
  bogus              |    |       | 
 (19 rows)
 
@@ -158,7 +158,7 @@ select amname, prop, pg_indexam_has_property(a.oid, prop) as p
  gist   | can_unique    | f
  gist   | can_multi_col | t
  gist   | can_exclude   | t
- gist   | can_include   | f
+ gist   | can_include   | t
  gist   | bogus         | 
  hash   | can_order     | f
  hash   | can_unique    | f
diff --git a/src/test/regress/expected/index_including.out b/src/test/regress/expected/index_including.out
index 45a1c8d017..3145ef17a9 100644
--- a/src/test/regress/expected/index_including.out
+++ b/src/test/regress/expected/index_including.out
@@ -313,22 +313,20 @@ SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl' ORDER BY indexname;
 
 DROP TABLE tbl;
 /*
- * 7. Check various AMs. All but btree must fail.
+ * 7. Check various AMs. All but btree and gist must fail.
  */
 CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
 CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "brin" does not support included columns
-CREATE INDEX on tbl USING gist(c3) INCLUDE (c4);
-ERROR:  access method "gist" does not support included columns
+CREATE INDEX on tbl USING gist(c4) INCLUDE (c3);
 CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
 ERROR:  access method "spgist" does not support included columns
 CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "gin" does not support included columns
 CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "hash" does not support included columns
-CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
+CREATE INDEX on tbl USING rtree(c4) INCLUDE (c3, c1);
 NOTICE:  substituting access method "gist" for obsolete method "rtree"
-ERROR:  access method "gist" does not support included columns
 CREATE INDEX on tbl USING btree(c1, c2) INCLUDE (c3, c4);
 DROP TABLE tbl;
 /*
diff --git a/src/test/regress/expected/index_including_gist.out b/src/test/regress/expected/index_including_gist.out
new file mode 100644
index 0000000000..2e27acb979
--- /dev/null
+++ b/src/test/regress/expected/index_including_gist.out
@@ -0,0 +1,139 @@
+/*
+ * 1.1. test CREATE INDEX with buffered build
+ */
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+                                  pg_get_indexdef                                  
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+ c1 | c2 | c3 |     c4      
+----+----+----+-------------
+  1 |  2 |  3 | (2,3),(1,2)
+  2 |  4 |  6 | (4,5),(2,3)
+  3 |  6 |  9 | (6,7),(3,4)
+  4 |  8 | 12 | (8,9),(4,5)
+(4 rows)
+
+SET enable_bitmapscan TO off;
+EXPLAIN SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                                     QUERY PLAN                                     
+------------------------------------------------------------------------------------
+ Index Only Scan using tbl_gist_idx on tbl_gist  (cost=0.27..36.41 rows=8 width=44)
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+/*
+ * 1.2. test CREATE INDEX with inserts
+ */
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+                                  pg_get_indexdef                                  
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+ c1 | c2 | c3 |     c4      
+----+----+----+-------------
+  1 |  2 |  3 | (2,3),(1,2)
+  2 |  4 |  6 | (4,5),(2,3)
+  3 |  6 |  9 | (6,7),(3,4)
+  4 |  8 | 12 | (8,9),(4,5)
+(4 rows)
+
+SET enable_bitmapscan TO off;
+EXPLAIN SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                                     QUERY PLAN                                     
+------------------------------------------------------------------------------------
+ Index Only Scan using tbl_gist_idx on tbl_gist  (cost=0.28..36.41 rows=8 width=44)
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+/*
+ * 2. CREATE INDEX CONCURRENTLY
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX CONCURRENTLY tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                     indexdef                                      
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+DROP TABLE tbl_gist;
+/*
+ * 3. REINDEX
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                   indexdef                                    
+-------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c3)
+(1 row)
+
+REINDEX INDEX tbl_gist_idx;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                   indexdef                                    
+-------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c3)
+(1 row)
+
+ALTER TABLE tbl_gist DROP COLUMN c1;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+ indexdef 
+----------
+(0 rows)
+
+DROP TABLE tbl_gist;
+/*
+ * 4. Update, delete values in indexed table.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+UPDATE tbl_gist SET c1 = 100 WHERE c1 = 2;
+UPDATE tbl_gist SET c1 = 1 WHERE c1 = 3;
+DELETE FROM tbl_gist WHERE c1 = 5 OR c3 = 12;
+DROP TABLE tbl_gist;
+/*
+ * 5. Alter column type.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+ALTER TABLE tbl_gist ALTER c1 TYPE bigint;
+ALTER TABLE tbl_gist ALTER c3 TYPE bigint;
+\d tbl_gist
+              Table "public.tbl_gist"
+ Column |  Type   | Collation | Nullable | Default 
+--------+---------+-----------+----------+---------
+ c1     | bigint  |           |          | 
+ c2     | integer |           |          | 
+ c3     | bigint  |           |          | 
+ c4     | box     |           |          | 
+Indexes:
+    "tbl_gist_idx" gist (c4) INCLUDE (c1, c3)
+
+DROP TABLE tbl_gist;
diff --git a/src/test/regress/parallel_schedule b/src/test/regress/parallel_schedule
index 16f979c8d9..a6db2f3abb 100644
--- a/src/test/regress/parallel_schedule
+++ b/src/test/regress/parallel_schedule
@@ -55,7 +55,7 @@ test: copy copyselect copydml
 # ----------
 test: create_misc create_operator create_procedure
 # These depend on the above two
-test: create_index create_view index_including
+test: create_index create_view index_including index_including_gist
 
 # ----------
 # Another group of parallel tests
diff --git a/src/test/regress/serial_schedule b/src/test/regress/serial_schedule
index 42632be675..71ed5efb19 100644
--- a/src/test/regress/serial_schedule
+++ b/src/test/regress/serial_schedule
@@ -66,6 +66,7 @@ test: create_operator
 test: create_procedure
 test: create_index
 test: index_including
+test: index_including_gist
 test: create_view
 test: create_aggregate
 test: create_function_3
diff --git a/src/test/regress/sql/index_including.sql b/src/test/regress/sql/index_including.sql
index ee9d9f6bae..17c3befb90 100644
--- a/src/test/regress/sql/index_including.sql
+++ b/src/test/regress/sql/index_including.sql
@@ -173,15 +173,15 @@ SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl' ORDER BY indexname;
 DROP TABLE tbl;
 
 /*
- * 7. Check various AMs. All but btree must fail.
+ * 7. Check various AMs. All but btree and gist must fail.
  */
 CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
 CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
-CREATE INDEX on tbl USING gist(c3) INCLUDE (c4);
+CREATE INDEX on tbl USING gist(c4) INCLUDE (c3);
 CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
 CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
 CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
-CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
+CREATE INDEX on tbl USING rtree(c4) INCLUDE (c3, c1);
 CREATE INDEX on tbl USING btree(c1, c2) INCLUDE (c3, c4);
 DROP TABLE tbl;
 
diff --git a/src/test/regress/sql/index_including_gist.sql b/src/test/regress/sql/index_including_gist.sql
new file mode 100644
index 0000000000..612d6ca210
--- /dev/null
+++ b/src/test/regress/sql/index_including_gist.sql
@@ -0,0 +1,80 @@
+/*
+ * 1.1. test CREATE INDEX with buffered build
+ */
+
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO off;
+EXPLAIN SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+
+/*
+ * 1.2. test CREATE INDEX with inserts
+ */
+
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO off;
+EXPLAIN SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+
+/*
+ * 2. CREATE INDEX CONCURRENTLY
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX CONCURRENTLY tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+DROP TABLE tbl_gist;
+
+
+/*
+ * 3. REINDEX
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+REINDEX INDEX tbl_gist_idx;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+ALTER TABLE tbl_gist DROP COLUMN c1;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+DROP TABLE tbl_gist;
+
+/*
+ * 4. Update, delete values in indexed table.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+UPDATE tbl_gist SET c1 = 100 WHERE c1 = 2;
+UPDATE tbl_gist SET c1 = 1 WHERE c1 = 3;
+DELETE FROM tbl_gist WHERE c1 = 5 OR c3 = 12;
+DROP TABLE tbl_gist;
+
+/*
+ * 5. Alter column type.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+ALTER TABLE tbl_gist ALTER c1 TYPE bigint;
+ALTER TABLE tbl_gist ALTER c3 TYPE bigint;
+\d tbl_gist
+DROP TABLE tbl_gist;
-- 
2.15.2 (Apple Git-101.1)

#12

Thomas Munro

thomas.munro@enterprisedb.com

over 7 years ago

In reply to: Andrey Borodin (#11)

Re: Covering GiST indexes

On Sun, Jul 29, 2018 at 10:50 PM, Andrey Borodin <x4mmm@yandex-team.ru> wrote:

Thanks! The problem appeared with commit 701fd0b [0] which dropped
validation rules checked in failed test. Here's the patch with fixed tests.

Thanks. I received the attachment.

Just as an FYI, something about the way your mail client encoded it
prevented the mail archives from picking it up (according to someone
who knows much more about these things than me, it's buried in the
"multipart/mixed" part, so not visible to the mail archive if it
extracts only the "text/plain" part, or something like that). That
didn't happen on any of your other patches. I have no opinion of
whether that's a bug in the mail archive or an unhelpful mail client
or a non-ideal way to post patches, but you can see here that we don't
have the patch in the usual place:

/messages/by-id/850AE105-F89A-42E4-AD40-FBC6EA5A5A00@yandex-team.ru

That explains why cfbot didn't automatically test your new patch, so
it still show as broken here:
http://cfbot.cputube.org/andrey-borodin.html

--
Thomas Munro
http://www.enterprisedb.com

#13

x4mmm@yandex-team.ru

over 7 years ago

In reply to: Thomas Munro (#12)

1 attachment(s)

Re: Covering GiST indexes

Thanks, Thomas!

30 июля 2018 г., в 3:58, Thomas Munro <thomas.munro@enterprisedb.com> написал(а):

On Sun, Jul 29, 2018 at 10:50 PM, Andrey Borodin <x4mmm@yandex-team.ru> wrote:

Thanks! The problem appeared with commit 701fd0b [0] which dropped
validation rules checked in failed test. Here's the patch with fixed tests.

Thanks. I received the attachment.

Just as an FYI, something about the way your mail client encoded it
prevented the mail archives from picking it up

I'll try to investigate this.

Here's patch one more: another attempt to put it into archives.

Best regards, Andrey Borodin.

Attachments:

0001-Covering-GiST-v4.patchapplication/octet-stream; name=0001-Covering-GiST-v4.patch; x-unix-mode=0644Download

From ed23a403be4076050a5eb0f9c1c2e34b7f9f36df Mon Sep 17 00:00:00 2001
From: Andrey Borodin <amborodin@acm.org>
Date: Sun, 29 Jul 2018 15:44:12 +0500
Subject: [PATCH] Covering GiST v4 Allow GiST index access method to include
 attributes without opclasses. Included attributes are stored uncompressed and
 cannot be used for index search efficiently, but can participate in Index
 Only Scan.

---
 doc/src/sgml/ref/create_index.sgml                 |   2 +-
 doc/src/sgml/textsearch.sgml                       |   6 +
 src/backend/access/gist/gist.c                     |  34 +++--
 src/backend/access/gist/gistbuild.c                |   2 +-
 src/backend/access/gist/gistget.c                  |   4 +-
 src/backend/access/gist/gistscan.c                 |  12 +-
 src/backend/access/gist/gistsplit.c                |  12 +-
 src/backend/access/gist/gistutil.c                 |  48 +++++--
 src/include/access/gist_private.h                  |   5 +-
 src/test/regress/expected/amutils.out              |   4 +-
 src/test/regress/expected/index_including.out      |   8 +-
 src/test/regress/expected/index_including_gist.out | 139 +++++++++++++++++++++
 src/test/regress/parallel_schedule                 |   2 +-
 src/test/regress/serial_schedule                   |   1 +
 src/test/regress/sql/index_including.sql           |   6 +-
 src/test/regress/sql/index_including_gist.sql      |  80 ++++++++++++
 16 files changed, 324 insertions(+), 41 deletions(-)
 create mode 100644 src/test/regress/expected/index_including_gist.out
 create mode 100644 src/test/regress/sql/index_including_gist.sql

diff --git a/doc/src/sgml/ref/create_index.sgml b/doc/src/sgml/ref/create_index.sgml
index 3c1223b324..7adc8852e8 100644
--- a/doc/src/sgml/ref/create_index.sgml
+++ b/doc/src/sgml/ref/create_index.sgml
@@ -181,7 +181,7 @@ CREATE [ UNIQUE ] INDEX [ CONCURRENTLY ] [ [ IF NOT EXISTS ] <replaceable class=
        </para>
 
        <para>
-        Currently, only the B-tree index access method supports this feature.
+        Currently, B-tree and GiST index access methods supports this feature.
         In B-tree indexes, the values of columns listed in the
         <literal>INCLUDE</literal> clause are included in leaf tuples which
         correspond to heap tuples, but are not included in upper-level
diff --git a/doc/src/sgml/textsearch.sgml b/doc/src/sgml/textsearch.sgml
index 6df424c63e..b3f9c1800b 100644
--- a/doc/src/sgml/textsearch.sgml
+++ b/doc/src/sgml/textsearch.sgml
@@ -3674,6 +3674,12 @@ SELECT plainto_tsquery('supernovae stars');
    retrieved to see if the match is correct.
   </para>
 
+  <para>
+   A GiST index can be covering, i.e. use <literal>INCLUDE</literal> clause.
+   Included columns can have data types without GiST operator class. Included
+   attributes will be stored uncompressed.
+  </para>
+
   <para>
    Lossiness causes performance degradation due to unnecessary fetches of table
    records that turn out to be false matches.  Since random access to table
diff --git a/src/backend/access/gist/gist.c b/src/backend/access/gist/gist.c
index 8a42effdf7..bbaac5de8c 100644
--- a/src/backend/access/gist/gist.c
+++ b/src/backend/access/gist/gist.c
@@ -74,7 +74,7 @@ gisthandler(PG_FUNCTION_ARGS)
 	amroutine->amclusterable = true;
 	amroutine->ampredlocks = true;
 	amroutine->amcanparallel = false;
-	amroutine->amcaninclude = false;
+	amroutine->amcaninclude = true;
 	amroutine->amkeytype = InvalidOid;
 
 	amroutine->ambuild = gistbuild;
@@ -169,7 +169,7 @@ gistinsert(Relation r, Datum *values, bool *isnull,
 	oldCxt = MemoryContextSwitchTo(giststate->tempCxt);
 
 	itup = gistFormTuple(giststate, r,
-						 values, isnull, true /* size is currently bogus */ );
+						 values, isnull, true /* size is currently bogus */, false);
 	itup->t_tid = *ht_ctid;
 
 	gistdoinsert(r, itup, 0, giststate);
@@ -1377,8 +1377,8 @@ gistSplit(Relation r,
 						IndexTupleSize(itup[0]), GiSTPageSize,
 						RelationGetRelationName(r))));
 
-	memset(v.spl_lisnull, true, sizeof(bool) * giststate->tupdesc->natts);
-	memset(v.spl_risnull, true, sizeof(bool) * giststate->tupdesc->natts);
+	memset(v.spl_lisnull, true, sizeof(bool) * giststate->truncTupdesc->natts);
+	memset(v.spl_risnull, true, sizeof(bool) * giststate->truncTupdesc->natts);
 	gistSplitByKey(r, page, itup, len, giststate, &v, 0);
 
 	/* form left and right vector */
@@ -1401,7 +1401,7 @@ gistSplit(Relation r,
 		ROTATEDIST(res);
 		res->block.num = v.splitVector.spl_nright;
 		res->list = gistfillitupvec(rvectup, v.splitVector.spl_nright, &(res->lenlist));
-		res->itup = gistFormTuple(giststate, r, v.spl_rattr, v.spl_risnull, false);
+		res->itup = gistFormTuple(giststate, r, v.spl_rattr, v.spl_risnull, false, true);
 	}
 
 	if (!gistfitpage(lvectup, v.splitVector.spl_nleft))
@@ -1423,7 +1423,7 @@ gistSplit(Relation r,
 		ROTATEDIST(res);
 		res->block.num = v.splitVector.spl_nleft;
 		res->list = gistfillitupvec(lvectup, v.splitVector.spl_nleft, &(res->lenlist));
-		res->itup = gistFormTuple(giststate, r, v.spl_lattr, v.spl_lisnull, false);
+		res->itup = gistFormTuple(giststate, r, v.spl_lattr, v.spl_lisnull, false, true);
 	}
 
 	return res;
@@ -1457,8 +1457,10 @@ initGISTstate(Relation index)
 	giststate->scanCxt = scanCxt;
 	giststate->tempCxt = scanCxt;	/* caller must change this if needed */
 	giststate->tupdesc = index->rd_att;
+	giststate->truncTupdesc = CreateTupleDescCopyConstr(index->rd_att);
+	giststate->truncTupdesc->natts = IndexRelationGetNumberOfKeyAttributes(index);
 
-	for (i = 0; i < index->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(index); i++)
 	{
 		fmgr_info_copy(&(giststate->consistentFn[i]),
 					   index_getprocinfo(index, i + 1, GIST_CONSISTENT_PROC),
@@ -1526,6 +1528,24 @@ initGISTstate(Relation index)
 			giststate->supportCollation[i] = DEFAULT_COLLATION_OID;
 	}
 
+	for (; i < index->rd_att->natts; i++)
+	{
+		giststate->consistentFn[i].fn_oid = InvalidOid;
+		giststate->unionFn[i].fn_oid = InvalidOid;
+		giststate->compressFn[i].fn_oid = InvalidOid;
+		giststate->decompressFn[i].fn_oid = InvalidOid;
+		giststate->penaltyFn[i].fn_oid = InvalidOid;
+		giststate->picksplitFn[i].fn_oid = InvalidOid;
+		giststate->equalFn[i].fn_oid = InvalidOid;
+		giststate->distanceFn[i].fn_oid = InvalidOid;
+		giststate->fetchFn[i].fn_oid = InvalidOid;
+
+		if (OidIsValid(index->rd_indcollation[i]))
+			giststate->supportCollation[i] = index->rd_indcollation[i];
+		else
+			giststate->supportCollation[i] = DEFAULT_COLLATION_OID;
+	}
+
 	MemoryContextSwitchTo(oldCxt);
 
 	return giststate;
diff --git a/src/backend/access/gist/gistbuild.c b/src/backend/access/gist/gistbuild.c
index 434f15f014..f4545f8069 100644
--- a/src/backend/access/gist/gistbuild.c
+++ b/src/backend/access/gist/gistbuild.c
@@ -469,7 +469,7 @@ gistBuildCallback(Relation index,
 	oldCtx = MemoryContextSwitchTo(buildstate->giststate->tempCxt);
 
 	/* form an index tuple and point it at the heap tuple */
-	itup = gistFormTuple(buildstate->giststate, index, values, isnull, true);
+	itup = gistFormTuple(buildstate->giststate, index, values, isnull, true, false);
 	itup->t_tid = htup->t_self;
 
 	if (buildstate->bufferingMode == GIST_BUFFERING_ACTIVE)
diff --git a/src/backend/access/gist/gistget.c b/src/backend/access/gist/gistget.c
index ad07b9e63c..c75a6a7b53 100644
--- a/src/backend/access/gist/gistget.c
+++ b/src/backend/access/gist/gistget.c
@@ -805,12 +805,14 @@ gistgetbitmap(IndexScanDesc scan, TIDBitmap *tbm)
  *
  * Opclasses that implement a fetch function support index-only scans.
  * Opclasses without compression functions also support index-only scans.
+ * Included attributes can be returned.
  */
 bool
 gistcanreturn(Relation index, int attno)
 {
 	if (OidIsValid(index_getprocid(index, attno, GIST_FETCH_PROC)) ||
-		!OidIsValid(index_getprocid(index, attno, GIST_COMPRESS_PROC)))
+		!OidIsValid(index_getprocid(index, attno, GIST_COMPRESS_PROC))||
+		attno > IndexRelationGetNumberOfKeyAttributes(index))
 		return true;
 	else
 		return false;
diff --git a/src/backend/access/gist/gistscan.c b/src/backend/access/gist/gistscan.c
index 4d97ff1d5d..fb93e8d6a6 100644
--- a/src/backend/access/gist/gistscan.c
+++ b/src/backend/access/gist/gistscan.c
@@ -158,6 +158,7 @@ gistrescan(IndexScanDesc scan, ScanKey key, int nkeys,
 	if (scan->xs_want_itup && !scan->xs_hitupdesc)
 	{
 		int			natts;
+		int			nkeyatts;
 		int			attno;
 
 		/*
@@ -167,13 +168,22 @@ gistrescan(IndexScanDesc scan, ScanKey key, int nkeys,
 		 * types.
 		 */
 		natts = RelationGetNumberOfAttributes(scan->indexRelation);
+		nkeyatts = IndexRelationGetNumberOfKeyAttributes(scan->indexRelation);
 		so->giststate->fetchTupdesc = CreateTemplateTupleDesc(natts, false);
-		for (attno = 1; attno <= natts; attno++)
+		for (attno = 1; attno <= nkeyatts; attno++)
 		{
 			TupleDescInitEntry(so->giststate->fetchTupdesc, attno, NULL,
 							   scan->indexRelation->rd_opcintype[attno - 1],
 							   -1, 0);
 		}
+
+		for (; attno <= natts; attno++)
+		{
+			/* taking opcintype from giststate->tupdesc */
+			TupleDescInitEntry(so->giststate->fetchTupdesc, attno, NULL,
+					TupleDescAttr(so->giststate->tupdesc, attno - 1)->atttypid,
+							   -1, 0);
+		}
 		scan->xs_hitupdesc = so->giststate->fetchTupdesc;
 
 		/* Also create a memory context that will hold the returned tuples */
diff --git a/src/backend/access/gist/gistsplit.c b/src/backend/access/gist/gistsplit.c
index a7038cca67..02918536bf 100644
--- a/src/backend/access/gist/gistsplit.c
+++ b/src/backend/access/gist/gistsplit.c
@@ -207,7 +207,7 @@ placeOne(Relation r, GISTSTATE *giststate, GistSplitVector *v,
 	gistDeCompressAtt(giststate, r, itup, NULL, (OffsetNumber) 0,
 					  identry, isnull);
 
-	for (; attno < giststate->tupdesc->natts; attno++)
+	for (; attno < giststate->truncTupdesc->natts; attno++)
 	{
 		float		lpenalty,
 					rpenalty;
@@ -485,7 +485,7 @@ gistUserPicksplit(Relation r, GistEntryVector *entryvec, int attno, GistSplitVec
 	 */
 	v->spl_dontcare = NULL;
 
-	if (attno + 1 < giststate->tupdesc->natts)
+	if (attno + 1 < giststate->truncTupdesc->natts)
 	{
 		int			NumDontCare;
 
@@ -657,7 +657,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 		 */
 		v->spl_risnull[attno] = v->spl_lisnull[attno] = true;
 
-		if (attno + 1 < giststate->tupdesc->natts)
+		if (attno + 1 < giststate->truncTupdesc->natts)
 			gistSplitByKey(r, page, itup, len, giststate, v, attno + 1);
 		else
 			gistSplitHalf(&v->splitVector, len);
@@ -683,7 +683,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 				v->splitVector.spl_left[v->splitVector.spl_nleft++] = i;
 
 		/* Compute union keys, unless outer recursion level will handle it */
-		if (attno == 0 && giststate->tupdesc->natts == 1)
+		if (attno == 0 && giststate->truncTupdesc->natts == 1)
 		{
 			v->spl_dontcare = NULL;
 			gistunionsubkey(giststate, itup, v);
@@ -700,7 +700,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 			 * Splitting on attno column is not optimal, so consider
 			 * redistributing don't-care tuples according to the next column
 			 */
-			Assert(attno + 1 < giststate->tupdesc->natts);
+			Assert(attno + 1 < giststate->truncTupdesc->natts);
 
 			if (v->spl_dontcare == NULL)
 			{
@@ -771,7 +771,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 	 * that PickSplit (or the special cases above) produced correct union
 	 * datums.
 	 */
-	if (attno == 0 && giststate->tupdesc->natts > 1)
+	if (attno == 0 && giststate->truncTupdesc->natts > 1)
 	{
 		v->spl_dontcare = NULL;
 		gistunionsubkey(giststate, itup, v);
diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c
index dddfe0ae2c..f87e84a5f1 100644
--- a/src/backend/access/gist/gistutil.c
+++ b/src/backend/access/gist/gistutil.c
@@ -159,7 +159,7 @@ gistMakeUnionItVec(GISTSTATE *giststate, IndexTuple *itvec, int len,
 
 	evec = (GistEntryVector *) palloc((len + 2) * sizeof(GISTENTRY) + GEVHDRSZ);
 
-	for (i = 0; i < giststate->tupdesc->natts; i++)
+	for (i = 0; i < giststate->truncTupdesc->natts; i++)
 	{
 		int			j;
 
@@ -220,7 +220,7 @@ gistunion(Relation r, IndexTuple *itvec, int len, GISTSTATE *giststate)
 
 	gistMakeUnionItVec(giststate, itvec, len, attr, isnull);
 
-	return gistFormTuple(giststate, r, attr, isnull, false);
+	return gistFormTuple(giststate, r, attr, isnull, false, true);
 }
 
 /*
@@ -295,7 +295,7 @@ gistDeCompressAtt(GISTSTATE *giststate, Relation r, IndexTuple tuple, Page p,
 {
 	int			i;
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		Datum		datum;
 
@@ -328,7 +328,7 @@ gistgetadjusted(Relation r, IndexTuple oldtup, IndexTuple addtup, GISTSTATE *gis
 	gistDeCompressAtt(giststate, r, addtup, NULL,
 					  (OffsetNumber) 0, addentries, addisnull);
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		gistMakeUnionKey(giststate, i,
 						 oldentries + i, oldisnull[i],
@@ -354,7 +354,7 @@ gistgetadjusted(Relation r, IndexTuple oldtup, IndexTuple addtup, GISTSTATE *gis
 	if (neednew)
 	{
 		/* need to update key */
-		newtup = gistFormTuple(giststate, r, attr, isnull, false);
+		newtup = gistFormTuple(giststate, r, attr, isnull, false, true);
 		newtup->t_tid = oldtup->t_tid;
 	}
 
@@ -441,7 +441,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 		zero_penalty = true;
 
 		/* Loop over index attributes. */
-		for (j = 0; j < r->rd_att->natts; j++)
+		for (j = 0; j < IndexRelationGetNumberOfKeyAttributes(r); j++)
 		{
 			Datum		datum;
 			float		usize;
@@ -469,7 +469,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 				result = i;
 				best_penalty[j] = usize;
 
-				if (j < r->rd_att->natts - 1)
+				if (j < IndexRelationGetNumberOfKeyAttributes(r) - 1)
 					best_penalty[j + 1] = -1;
 
 				/* we have new best, so reset keep-it decision */
@@ -499,7 +499,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 		 * If we looped past the last column, and did not update "result",
 		 * then this tuple is exactly as good as the prior best tuple.
 		 */
-		if (j == r->rd_att->natts && result != i)
+		if (j == IndexRelationGetNumberOfKeyAttributes(r) && result != i)
 		{
 			if (keep_current_best == -1)
 			{
@@ -569,7 +569,7 @@ gistdentryinit(GISTSTATE *giststate, int nkey, GISTENTRY *e,
 
 IndexTuple
 gistFormTuple(GISTSTATE *giststate, Relation r,
-			  Datum attdata[], bool isnull[], bool isleaf)
+			  Datum attdata[], bool isnull[], bool isleaf, bool isTruncated)
 {
 	Datum		compatt[INDEX_MAX_KEYS];
 	int			i;
@@ -578,7 +578,7 @@ gistFormTuple(GISTSTATE *giststate, Relation r,
 	/*
 	 * Call the compress method on each attribute.
 	 */
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		if (isnull[i])
 			compatt[i] = (Datum) 0;
@@ -601,7 +601,23 @@ gistFormTuple(GISTSTATE *giststate, Relation r,
 		}
 	}
 
-	res = index_form_tuple(giststate->tupdesc, compatt, isnull);
+	if (!isTruncated)
+	{
+		/*
+		 * Allocate each included attribute.
+		 */
+		for (; i < r->rd_att->natts; i++)
+		{
+			if (isnull[i])
+				compatt[i] = (Datum) 0;
+			else
+			{
+				compatt[i] = attdata[i];
+			}
+		}
+	}
+
+	res = index_form_tuple(isTruncated?giststate->truncTupdesc:giststate->tupdesc, compatt, isnull);
 
 	/*
 	 * The offset number on tuples on internal pages is unused. For historical
@@ -643,7 +659,7 @@ gistFetchTuple(GISTSTATE *giststate, Relation r, IndexTuple tuple)
 	bool		isnull[INDEX_MAX_KEYS];
 	int			i;
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		Datum		datum;
 
@@ -678,6 +694,14 @@ gistFetchTuple(GISTSTATE *giststate, Relation r, IndexTuple tuple)
 			fetchatt[i] = (Datum) 0;
 		}
 	}
+
+	/*
+	 * Get each included attribute.
+	 */
+	for (; i < r->rd_att->natts; i++)
+	{
+		fetchatt[i] = index_getattr(tuple, i + 1, giststate->tupdesc, &isnull[i]);
+	}
 	MemoryContextSwitchTo(oldcxt);
 
 	return heap_form_tuple(giststate->fetchTupdesc, fetchatt, isnull);
diff --git a/src/include/access/gist_private.h b/src/include/access/gist_private.h
index 36ed7244ba..c1714a6ccc 100644
--- a/src/include/access/gist_private.h
+++ b/src/include/access/gist_private.h
@@ -79,6 +79,8 @@ typedef struct GISTSTATE
 	MemoryContext tempCxt;		/* short-term context for calling functions */
 
 	TupleDesc	tupdesc;		/* index's tuple descriptor */
+	TupleDesc	truncTupdesc;	/* truncated tuple descriptor
+								 * for internal pages */
 	TupleDesc	fetchTupdesc;	/* tuple descriptor for tuples returned in an
 								 * index-only scan */
 
@@ -458,7 +460,8 @@ extern IndexTuple gistgetadjusted(Relation r,
 				IndexTuple addtup,
 				GISTSTATE *giststate);
 extern IndexTuple gistFormTuple(GISTSTATE *giststate,
-			  Relation r, Datum *attdata, bool *isnull, bool isleaf);
+			  Relation r, Datum *attdata, bool *isnull, bool isleaf,
+			  bool isTruncated);
 
 extern OffsetNumber gistchoose(Relation r, Page p,
 		   IndexTuple it,
diff --git a/src/test/regress/expected/amutils.out b/src/test/regress/expected/amutils.out
index 24cd3c5e2e..0c481f94ab 100644
--- a/src/test/regress/expected/amutils.out
+++ b/src/test/regress/expected/amutils.out
@@ -75,7 +75,7 @@ select prop,
  can_unique         | f  |       | 
  can_multi_col      | t  |       | 
  can_exclude        | t  |       | 
- can_include        | f  |       | 
+ can_include        | t  |       | 
  bogus              |    |       | 
 (19 rows)
 
@@ -158,7 +158,7 @@ select amname, prop, pg_indexam_has_property(a.oid, prop) as p
  gist   | can_unique    | f
  gist   | can_multi_col | t
  gist   | can_exclude   | t
- gist   | can_include   | f
+ gist   | can_include   | t
  gist   | bogus         | 
  hash   | can_order     | f
  hash   | can_unique    | f
diff --git a/src/test/regress/expected/index_including.out b/src/test/regress/expected/index_including.out
index 45a1c8d017..3145ef17a9 100644
--- a/src/test/regress/expected/index_including.out
+++ b/src/test/regress/expected/index_including.out
@@ -313,22 +313,20 @@ SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl' ORDER BY indexname;
 
 DROP TABLE tbl;
 /*
- * 7. Check various AMs. All but btree must fail.
+ * 7. Check various AMs. All but btree and gist must fail.
  */
 CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
 CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "brin" does not support included columns
-CREATE INDEX on tbl USING gist(c3) INCLUDE (c4);
-ERROR:  access method "gist" does not support included columns
+CREATE INDEX on tbl USING gist(c4) INCLUDE (c3);
 CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
 ERROR:  access method "spgist" does not support included columns
 CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "gin" does not support included columns
 CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "hash" does not support included columns
-CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
+CREATE INDEX on tbl USING rtree(c4) INCLUDE (c3, c1);
 NOTICE:  substituting access method "gist" for obsolete method "rtree"
-ERROR:  access method "gist" does not support included columns
 CREATE INDEX on tbl USING btree(c1, c2) INCLUDE (c3, c4);
 DROP TABLE tbl;
 /*
diff --git a/src/test/regress/expected/index_including_gist.out b/src/test/regress/expected/index_including_gist.out
new file mode 100644
index 0000000000..2e27acb979
--- /dev/null
+++ b/src/test/regress/expected/index_including_gist.out
@@ -0,0 +1,139 @@
+/*
+ * 1.1. test CREATE INDEX with buffered build
+ */
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+                                  pg_get_indexdef                                  
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+ c1 | c2 | c3 |     c4      
+----+----+----+-------------
+  1 |  2 |  3 | (2,3),(1,2)
+  2 |  4 |  6 | (4,5),(2,3)
+  3 |  6 |  9 | (6,7),(3,4)
+  4 |  8 | 12 | (8,9),(4,5)
+(4 rows)
+
+SET enable_bitmapscan TO off;
+EXPLAIN SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                                     QUERY PLAN                                     
+------------------------------------------------------------------------------------
+ Index Only Scan using tbl_gist_idx on tbl_gist  (cost=0.27..36.41 rows=8 width=44)
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+/*
+ * 1.2. test CREATE INDEX with inserts
+ */
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+                                  pg_get_indexdef                                  
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+ c1 | c2 | c3 |     c4      
+----+----+----+-------------
+  1 |  2 |  3 | (2,3),(1,2)
+  2 |  4 |  6 | (4,5),(2,3)
+  3 |  6 |  9 | (6,7),(3,4)
+  4 |  8 | 12 | (8,9),(4,5)
+(4 rows)
+
+SET enable_bitmapscan TO off;
+EXPLAIN SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                                     QUERY PLAN                                     
+------------------------------------------------------------------------------------
+ Index Only Scan using tbl_gist_idx on tbl_gist  (cost=0.28..36.41 rows=8 width=44)
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+/*
+ * 2. CREATE INDEX CONCURRENTLY
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX CONCURRENTLY tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                     indexdef                                      
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+DROP TABLE tbl_gist;
+/*
+ * 3. REINDEX
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                   indexdef                                    
+-------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c3)
+(1 row)
+
+REINDEX INDEX tbl_gist_idx;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                   indexdef                                    
+-------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c3)
+(1 row)
+
+ALTER TABLE tbl_gist DROP COLUMN c1;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+ indexdef 
+----------
+(0 rows)
+
+DROP TABLE tbl_gist;
+/*
+ * 4. Update, delete values in indexed table.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+UPDATE tbl_gist SET c1 = 100 WHERE c1 = 2;
+UPDATE tbl_gist SET c1 = 1 WHERE c1 = 3;
+DELETE FROM tbl_gist WHERE c1 = 5 OR c3 = 12;
+DROP TABLE tbl_gist;
+/*
+ * 5. Alter column type.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+ALTER TABLE tbl_gist ALTER c1 TYPE bigint;
+ALTER TABLE tbl_gist ALTER c3 TYPE bigint;
+\d tbl_gist
+              Table "public.tbl_gist"
+ Column |  Type   | Collation | Nullable | Default 
+--------+---------+-----------+----------+---------
+ c1     | bigint  |           |          | 
+ c2     | integer |           |          | 
+ c3     | bigint  |           |          | 
+ c4     | box     |           |          | 
+Indexes:
+    "tbl_gist_idx" gist (c4) INCLUDE (c1, c3)
+
+DROP TABLE tbl_gist;
diff --git a/src/test/regress/parallel_schedule b/src/test/regress/parallel_schedule
index 16f979c8d9..a6db2f3abb 100644
--- a/src/test/regress/parallel_schedule
+++ b/src/test/regress/parallel_schedule
@@ -55,7 +55,7 @@ test: copy copyselect copydml
 # ----------
 test: create_misc create_operator create_procedure
 # These depend on the above two
-test: create_index create_view index_including
+test: create_index create_view index_including index_including_gist
 
 # ----------
 # Another group of parallel tests
diff --git a/src/test/regress/serial_schedule b/src/test/regress/serial_schedule
index 42632be675..71ed5efb19 100644
--- a/src/test/regress/serial_schedule
+++ b/src/test/regress/serial_schedule
@@ -66,6 +66,7 @@ test: create_operator
 test: create_procedure
 test: create_index
 test: index_including
+test: index_including_gist
 test: create_view
 test: create_aggregate
 test: create_function_3
diff --git a/src/test/regress/sql/index_including.sql b/src/test/regress/sql/index_including.sql
index ee9d9f6bae..17c3befb90 100644
--- a/src/test/regress/sql/index_including.sql
+++ b/src/test/regress/sql/index_including.sql
@@ -173,15 +173,15 @@ SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl' ORDER BY indexname;
 DROP TABLE tbl;
 
 /*
- * 7. Check various AMs. All but btree must fail.
+ * 7. Check various AMs. All but btree and gist must fail.
  */
 CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
 CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
-CREATE INDEX on tbl USING gist(c3) INCLUDE (c4);
+CREATE INDEX on tbl USING gist(c4) INCLUDE (c3);
 CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
 CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
 CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
-CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
+CREATE INDEX on tbl USING rtree(c4) INCLUDE (c3, c1);
 CREATE INDEX on tbl USING btree(c1, c2) INCLUDE (c3, c4);
 DROP TABLE tbl;
 
diff --git a/src/test/regress/sql/index_including_gist.sql b/src/test/regress/sql/index_including_gist.sql
new file mode 100644
index 0000000000..612d6ca210
--- /dev/null
+++ b/src/test/regress/sql/index_including_gist.sql
@@ -0,0 +1,80 @@
+/*
+ * 1.1. test CREATE INDEX with buffered build
+ */
+
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO off;
+EXPLAIN SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+
+/*
+ * 1.2. test CREATE INDEX with inserts
+ */
+
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO off;
+EXPLAIN SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+
+/*
+ * 2. CREATE INDEX CONCURRENTLY
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX CONCURRENTLY tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+DROP TABLE tbl_gist;
+
+
+/*
+ * 3. REINDEX
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+REINDEX INDEX tbl_gist_idx;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+ALTER TABLE tbl_gist DROP COLUMN c1;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+DROP TABLE tbl_gist;
+
+/*
+ * 4. Update, delete values in indexed table.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+UPDATE tbl_gist SET c1 = 100 WHERE c1 = 2;
+UPDATE tbl_gist SET c1 = 1 WHERE c1 = 3;
+DELETE FROM tbl_gist WHERE c1 = 5 OR c3 = 12;
+DROP TABLE tbl_gist;
+
+/*
+ * 5. Alter column type.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+ALTER TABLE tbl_gist ALTER c1 TYPE bigint;
+ALTER TABLE tbl_gist ALTER c3 TYPE bigint;
+\d tbl_gist
+DROP TABLE tbl_gist;
-- 
2.15.2 (Apple Git-101.1)

#14

Dmitry Dolgov

9erthalion6@gmail.com

about 7 years ago

In reply to: Andrey Borodin (#13)

Re: Covering GiST indexes

On Mon, Jul 30, 2018 at 7:50 AM Andrey Borodin <x4mmm@yandex-team.ru> wrote:

Thanks, Thomas!

30 июля 2018 г., в 3:58, Thomas Munro <thomas.munro@enterprisedb.com> написал(а):

On Sun, Jul 29, 2018 at 10:50 PM, Andrey Borodin <x4mmm@yandex-team.ru> wrote:

Thanks! The problem appeared with commit 701fd0b [0] which dropped
validation rules checked in failed test. Here's the patch with fixed tests.

Thanks. I received the attachment.

Just as an FYI, something about the way your mail client encoded it
prevented the mail archives from picking it up

I'll try to investigate this.

Here's patch one more: another attempt to put it into archives.

Hi,

Looks like this time the patch was picked up correctly, but there are some
conflicts with the current master branch:
http://cfbot.cputube.org/patch_20_1615.log
Could you please rebase it one more time?

#15

x4mmm@yandex-team.ru

about 7 years ago

In reply to: Dmitry Dolgov (#14)

1 attachment(s)

Re: Covering GiST indexes

Hi, Dmitry!

26 нояб. 2018 г., в 21:20, Dmitry Dolgov <9erthalion6@gmail.com> написал(а):

Looks like this time the patch was picked up correctly, but there are some
conflicts with the current master branch:
http://cfbot.cputube.org/patch_20_1615.log
Could you please rebase it one more time?

Here's rebased version. Thanks!

Best regards, Andrey Borodin.

Attachments:

0001-Covering-GiST-v5.patchapplication/octet-stream; name=0001-Covering-GiST-v5.patch; x-unix-mode=0644Download

From e6495b8a83eec7d77a9e9b5bf014346f50f71b5a Mon Sep 17 00:00:00 2001
From: Andrey <amborodin@acm.org>
Date: Mon, 26 Nov 2018 21:55:45 +0500
Subject: [PATCH] Covering-GiST-v5

---
 doc/src/sgml/ref/create_index.sgml            |   2 +-
 doc/src/sgml/textsearch.sgml                  |   6 +
 src/backend/access/gist/gist.c                |  34 ++++-
 src/backend/access/gist/gistbuild.c           |   2 +-
 src/backend/access/gist/gistget.c             |   4 +-
 src/backend/access/gist/gistscan.c            |  12 +-
 src/backend/access/gist/gistsplit.c           |  12 +-
 src/backend/access/gist/gistutil.c            |  48 ++++--
 src/include/access/gist_private.h             |   5 +-
 src/test/regress/expected/amutils.out         |   4 +-
 src/test/regress/expected/index_including.out |   8 +-
 .../regress/expected/index_including_gist.out | 139 ++++++++++++++++++
 src/test/regress/parallel_schedule            |   2 +-
 src/test/regress/serial_schedule              |   1 +
 src/test/regress/sql/index_including.sql      |   6 +-
 src/test/regress/sql/index_including_gist.sql |  80 ++++++++++
 16 files changed, 324 insertions(+), 41 deletions(-)
 create mode 100644 src/test/regress/expected/index_including_gist.out
 create mode 100644 src/test/regress/sql/index_including_gist.sql

diff --git a/doc/src/sgml/ref/create_index.sgml b/doc/src/sgml/ref/create_index.sgml
index ad619cdcfe..cd3409eb76 100644
--- a/doc/src/sgml/ref/create_index.sgml
+++ b/doc/src/sgml/ref/create_index.sgml
@@ -181,7 +181,7 @@ CREATE [ UNIQUE ] INDEX [ CONCURRENTLY ] [ [ IF NOT EXISTS ] <replaceable class=
        </para>
 
        <para>
-        Currently, only the B-tree index access method supports this feature.
+        Currently, B-tree and GiST index access methods supports this feature.
         In B-tree indexes, the values of columns listed in the
         <literal>INCLUDE</literal> clause are included in leaf tuples which
         correspond to heap tuples, but are not included in upper-level
diff --git a/doc/src/sgml/textsearch.sgml b/doc/src/sgml/textsearch.sgml
index ecebade767..0305a5ad38 100644
--- a/doc/src/sgml/textsearch.sgml
+++ b/doc/src/sgml/textsearch.sgml
@@ -3674,6 +3674,12 @@ SELECT plainto_tsquery('supernovae stars');
    retrieved to see if the match is correct.
   </para>
 
+  <para>
+   A GiST index can be covering, i.e. use <literal>INCLUDE</literal> clause.
+   Included columns can have data types without GiST operator class. Included
+   attributes will be stored uncompressed.
+  </para>
+
   <para>
    Lossiness causes performance degradation due to unnecessary fetches of table
    records that turn out to be false matches.  Since random access to table
diff --git a/src/backend/access/gist/gist.c b/src/backend/access/gist/gist.c
index 8a42effdf7..bbaac5de8c 100644
--- a/src/backend/access/gist/gist.c
+++ b/src/backend/access/gist/gist.c
@@ -74,7 +74,7 @@ gisthandler(PG_FUNCTION_ARGS)
 	amroutine->amclusterable = true;
 	amroutine->ampredlocks = true;
 	amroutine->amcanparallel = false;
-	amroutine->amcaninclude = false;
+	amroutine->amcaninclude = true;
 	amroutine->amkeytype = InvalidOid;
 
 	amroutine->ambuild = gistbuild;
@@ -169,7 +169,7 @@ gistinsert(Relation r, Datum *values, bool *isnull,
 	oldCxt = MemoryContextSwitchTo(giststate->tempCxt);
 
 	itup = gistFormTuple(giststate, r,
-						 values, isnull, true /* size is currently bogus */ );
+						 values, isnull, true /* size is currently bogus */, false);
 	itup->t_tid = *ht_ctid;
 
 	gistdoinsert(r, itup, 0, giststate);
@@ -1377,8 +1377,8 @@ gistSplit(Relation r,
 						IndexTupleSize(itup[0]), GiSTPageSize,
 						RelationGetRelationName(r))));
 
-	memset(v.spl_lisnull, true, sizeof(bool) * giststate->tupdesc->natts);
-	memset(v.spl_risnull, true, sizeof(bool) * giststate->tupdesc->natts);
+	memset(v.spl_lisnull, true, sizeof(bool) * giststate->truncTupdesc->natts);
+	memset(v.spl_risnull, true, sizeof(bool) * giststate->truncTupdesc->natts);
 	gistSplitByKey(r, page, itup, len, giststate, &v, 0);
 
 	/* form left and right vector */
@@ -1401,7 +1401,7 @@ gistSplit(Relation r,
 		ROTATEDIST(res);
 		res->block.num = v.splitVector.spl_nright;
 		res->list = gistfillitupvec(rvectup, v.splitVector.spl_nright, &(res->lenlist));
-		res->itup = gistFormTuple(giststate, r, v.spl_rattr, v.spl_risnull, false);
+		res->itup = gistFormTuple(giststate, r, v.spl_rattr, v.spl_risnull, false, true);
 	}
 
 	if (!gistfitpage(lvectup, v.splitVector.spl_nleft))
@@ -1423,7 +1423,7 @@ gistSplit(Relation r,
 		ROTATEDIST(res);
 		res->block.num = v.splitVector.spl_nleft;
 		res->list = gistfillitupvec(lvectup, v.splitVector.spl_nleft, &(res->lenlist));
-		res->itup = gistFormTuple(giststate, r, v.spl_lattr, v.spl_lisnull, false);
+		res->itup = gistFormTuple(giststate, r, v.spl_lattr, v.spl_lisnull, false, true);
 	}
 
 	return res;
@@ -1457,8 +1457,10 @@ initGISTstate(Relation index)
 	giststate->scanCxt = scanCxt;
 	giststate->tempCxt = scanCxt;	/* caller must change this if needed */
 	giststate->tupdesc = index->rd_att;
+	giststate->truncTupdesc = CreateTupleDescCopyConstr(index->rd_att);
+	giststate->truncTupdesc->natts = IndexRelationGetNumberOfKeyAttributes(index);
 
-	for (i = 0; i < index->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(index); i++)
 	{
 		fmgr_info_copy(&(giststate->consistentFn[i]),
 					   index_getprocinfo(index, i + 1, GIST_CONSISTENT_PROC),
@@ -1526,6 +1528,24 @@ initGISTstate(Relation index)
 			giststate->supportCollation[i] = DEFAULT_COLLATION_OID;
 	}
 
+	for (; i < index->rd_att->natts; i++)
+	{
+		giststate->consistentFn[i].fn_oid = InvalidOid;
+		giststate->unionFn[i].fn_oid = InvalidOid;
+		giststate->compressFn[i].fn_oid = InvalidOid;
+		giststate->decompressFn[i].fn_oid = InvalidOid;
+		giststate->penaltyFn[i].fn_oid = InvalidOid;
+		giststate->picksplitFn[i].fn_oid = InvalidOid;
+		giststate->equalFn[i].fn_oid = InvalidOid;
+		giststate->distanceFn[i].fn_oid = InvalidOid;
+		giststate->fetchFn[i].fn_oid = InvalidOid;
+
+		if (OidIsValid(index->rd_indcollation[i]))
+			giststate->supportCollation[i] = index->rd_indcollation[i];
+		else
+			giststate->supportCollation[i] = DEFAULT_COLLATION_OID;
+	}
+
 	MemoryContextSwitchTo(oldCxt);
 
 	return giststate;
diff --git a/src/backend/access/gist/gistbuild.c b/src/backend/access/gist/gistbuild.c
index 434f15f014..f4545f8069 100644
--- a/src/backend/access/gist/gistbuild.c
+++ b/src/backend/access/gist/gistbuild.c
@@ -469,7 +469,7 @@ gistBuildCallback(Relation index,
 	oldCtx = MemoryContextSwitchTo(buildstate->giststate->tempCxt);
 
 	/* form an index tuple and point it at the heap tuple */
-	itup = gistFormTuple(buildstate->giststate, index, values, isnull, true);
+	itup = gistFormTuple(buildstate->giststate, index, values, isnull, true, false);
 	itup->t_tid = htup->t_self;
 
 	if (buildstate->bufferingMode == GIST_BUFFERING_ACTIVE)
diff --git a/src/backend/access/gist/gistget.c b/src/backend/access/gist/gistget.c
index e4a3786be0..30ab522f9b 100644
--- a/src/backend/access/gist/gistget.c
+++ b/src/backend/access/gist/gistget.c
@@ -769,12 +769,14 @@ gistgetbitmap(IndexScanDesc scan, TIDBitmap *tbm)
  *
  * Opclasses that implement a fetch function support index-only scans.
  * Opclasses without compression functions also support index-only scans.
+ * Included attributes can be returned.
  */
 bool
 gistcanreturn(Relation index, int attno)
 {
 	if (OidIsValid(index_getprocid(index, attno, GIST_FETCH_PROC)) ||
-		!OidIsValid(index_getprocid(index, attno, GIST_COMPRESS_PROC)))
+		!OidIsValid(index_getprocid(index, attno, GIST_COMPRESS_PROC))||
+		attno > IndexRelationGetNumberOfKeyAttributes(index))
 		return true;
 	else
 		return false;
diff --git a/src/backend/access/gist/gistscan.c b/src/backend/access/gist/gistscan.c
index 6b18bd3afb..abd04c9dee 100644
--- a/src/backend/access/gist/gistscan.c
+++ b/src/backend/access/gist/gistscan.c
@@ -158,6 +158,7 @@ gistrescan(IndexScanDesc scan, ScanKey key, int nkeys,
 	if (scan->xs_want_itup && !scan->xs_hitupdesc)
 	{
 		int			natts;
+		int			nkeyatts;
 		int			attno;
 
 		/*
@@ -167,13 +168,22 @@ gistrescan(IndexScanDesc scan, ScanKey key, int nkeys,
 		 * types.
 		 */
 		natts = RelationGetNumberOfAttributes(scan->indexRelation);
+		nkeyatts = IndexRelationGetNumberOfKeyAttributes(scan->indexRelation);
 		so->giststate->fetchTupdesc = CreateTemplateTupleDesc(natts);
-		for (attno = 1; attno <= natts; attno++)
+		for (attno = 1; attno <= nkeyatts; attno++)
 		{
 			TupleDescInitEntry(so->giststate->fetchTupdesc, attno, NULL,
 							   scan->indexRelation->rd_opcintype[attno - 1],
 							   -1, 0);
 		}
+
+		for (; attno <= natts; attno++)
+		{
+			/* taking opcintype from giststate->tupdesc */
+			TupleDescInitEntry(so->giststate->fetchTupdesc, attno, NULL,
+					TupleDescAttr(so->giststate->tupdesc, attno - 1)->atttypid,
+							   -1, 0);
+		}
 		scan->xs_hitupdesc = so->giststate->fetchTupdesc;
 
 		/* Also create a memory context that will hold the returned tuples */
diff --git a/src/backend/access/gist/gistsplit.c b/src/backend/access/gist/gistsplit.c
index a7038cca67..02918536bf 100644
--- a/src/backend/access/gist/gistsplit.c
+++ b/src/backend/access/gist/gistsplit.c
@@ -207,7 +207,7 @@ placeOne(Relation r, GISTSTATE *giststate, GistSplitVector *v,
 	gistDeCompressAtt(giststate, r, itup, NULL, (OffsetNumber) 0,
 					  identry, isnull);
 
-	for (; attno < giststate->tupdesc->natts; attno++)
+	for (; attno < giststate->truncTupdesc->natts; attno++)
 	{
 		float		lpenalty,
 					rpenalty;
@@ -485,7 +485,7 @@ gistUserPicksplit(Relation r, GistEntryVector *entryvec, int attno, GistSplitVec
 	 */
 	v->spl_dontcare = NULL;
 
-	if (attno + 1 < giststate->tupdesc->natts)
+	if (attno + 1 < giststate->truncTupdesc->natts)
 	{
 		int			NumDontCare;
 
@@ -657,7 +657,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 		 */
 		v->spl_risnull[attno] = v->spl_lisnull[attno] = true;
 
-		if (attno + 1 < giststate->tupdesc->natts)
+		if (attno + 1 < giststate->truncTupdesc->natts)
 			gistSplitByKey(r, page, itup, len, giststate, v, attno + 1);
 		else
 			gistSplitHalf(&v->splitVector, len);
@@ -683,7 +683,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 				v->splitVector.spl_left[v->splitVector.spl_nleft++] = i;
 
 		/* Compute union keys, unless outer recursion level will handle it */
-		if (attno == 0 && giststate->tupdesc->natts == 1)
+		if (attno == 0 && giststate->truncTupdesc->natts == 1)
 		{
 			v->spl_dontcare = NULL;
 			gistunionsubkey(giststate, itup, v);
@@ -700,7 +700,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 			 * Splitting on attno column is not optimal, so consider
 			 * redistributing don't-care tuples according to the next column
 			 */
-			Assert(attno + 1 < giststate->tupdesc->natts);
+			Assert(attno + 1 < giststate->truncTupdesc->natts);
 
 			if (v->spl_dontcare == NULL)
 			{
@@ -771,7 +771,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 	 * that PickSplit (or the special cases above) produced correct union
 	 * datums.
 	 */
-	if (attno == 0 && giststate->tupdesc->natts > 1)
+	if (attno == 0 && giststate->truncTupdesc->natts > 1)
 	{
 		v->spl_dontcare = NULL;
 		gistunionsubkey(giststate, itup, v);
diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c
index 70627e5df6..771b37f625 100644
--- a/src/backend/access/gist/gistutil.c
+++ b/src/backend/access/gist/gistutil.c
@@ -160,7 +160,7 @@ gistMakeUnionItVec(GISTSTATE *giststate, IndexTuple *itvec, int len,
 
 	evec = (GistEntryVector *) palloc((len + 2) * sizeof(GISTENTRY) + GEVHDRSZ);
 
-	for (i = 0; i < giststate->tupdesc->natts; i++)
+	for (i = 0; i < giststate->truncTupdesc->natts; i++)
 	{
 		int			j;
 
@@ -221,7 +221,7 @@ gistunion(Relation r, IndexTuple *itvec, int len, GISTSTATE *giststate)
 
 	gistMakeUnionItVec(giststate, itvec, len, attr, isnull);
 
-	return gistFormTuple(giststate, r, attr, isnull, false);
+	return gistFormTuple(giststate, r, attr, isnull, false, true);
 }
 
 /*
@@ -296,7 +296,7 @@ gistDeCompressAtt(GISTSTATE *giststate, Relation r, IndexTuple tuple, Page p,
 {
 	int			i;
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		Datum		datum;
 
@@ -329,7 +329,7 @@ gistgetadjusted(Relation r, IndexTuple oldtup, IndexTuple addtup, GISTSTATE *gis
 	gistDeCompressAtt(giststate, r, addtup, NULL,
 					  (OffsetNumber) 0, addentries, addisnull);
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		gistMakeUnionKey(giststate, i,
 						 oldentries + i, oldisnull[i],
@@ -355,7 +355,7 @@ gistgetadjusted(Relation r, IndexTuple oldtup, IndexTuple addtup, GISTSTATE *gis
 	if (neednew)
 	{
 		/* need to update key */
-		newtup = gistFormTuple(giststate, r, attr, isnull, false);
+		newtup = gistFormTuple(giststate, r, attr, isnull, false, true);
 		newtup->t_tid = oldtup->t_tid;
 	}
 
@@ -442,7 +442,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 		zero_penalty = true;
 
 		/* Loop over index attributes. */
-		for (j = 0; j < r->rd_att->natts; j++)
+		for (j = 0; j < IndexRelationGetNumberOfKeyAttributes(r); j++)
 		{
 			Datum		datum;
 			float		usize;
@@ -470,7 +470,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 				result = i;
 				best_penalty[j] = usize;
 
-				if (j < r->rd_att->natts - 1)
+				if (j < IndexRelationGetNumberOfKeyAttributes(r) - 1)
 					best_penalty[j + 1] = -1;
 
 				/* we have new best, so reset keep-it decision */
@@ -500,7 +500,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 		 * If we looped past the last column, and did not update "result",
 		 * then this tuple is exactly as good as the prior best tuple.
 		 */
-		if (j == r->rd_att->natts && result != i)
+		if (j == IndexRelationGetNumberOfKeyAttributes(r) && result != i)
 		{
 			if (keep_current_best == -1)
 			{
@@ -570,7 +570,7 @@ gistdentryinit(GISTSTATE *giststate, int nkey, GISTENTRY *e,
 
 IndexTuple
 gistFormTuple(GISTSTATE *giststate, Relation r,
-			  Datum attdata[], bool isnull[], bool isleaf)
+			  Datum attdata[], bool isnull[], bool isleaf, bool isTruncated)
 {
 	Datum		compatt[INDEX_MAX_KEYS];
 	int			i;
@@ -579,7 +579,7 @@ gistFormTuple(GISTSTATE *giststate, Relation r,
 	/*
 	 * Call the compress method on each attribute.
 	 */
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		if (isnull[i])
 			compatt[i] = (Datum) 0;
@@ -602,7 +602,23 @@ gistFormTuple(GISTSTATE *giststate, Relation r,
 		}
 	}
 
-	res = index_form_tuple(giststate->tupdesc, compatt, isnull);
+	if (!isTruncated)
+	{
+		/*
+		 * Allocate each included attribute.
+		 */
+		for (; i < r->rd_att->natts; i++)
+		{
+			if (isnull[i])
+				compatt[i] = (Datum) 0;
+			else
+			{
+				compatt[i] = attdata[i];
+			}
+		}
+	}
+
+	res = index_form_tuple(isTruncated?giststate->truncTupdesc:giststate->tupdesc, compatt, isnull);
 
 	/*
 	 * The offset number on tuples on internal pages is unused. For historical
@@ -644,7 +660,7 @@ gistFetchTuple(GISTSTATE *giststate, Relation r, IndexTuple tuple)
 	bool		isnull[INDEX_MAX_KEYS];
 	int			i;
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		Datum		datum;
 
@@ -679,6 +695,14 @@ gistFetchTuple(GISTSTATE *giststate, Relation r, IndexTuple tuple)
 			fetchatt[i] = (Datum) 0;
 		}
 	}
+
+	/*
+	 * Get each included attribute.
+	 */
+	for (; i < r->rd_att->natts; i++)
+	{
+		fetchatt[i] = index_getattr(tuple, i + 1, giststate->tupdesc, &isnull[i]);
+	}
 	MemoryContextSwitchTo(oldcxt);
 
 	return heap_form_tuple(giststate->fetchTupdesc, fetchatt, isnull);
diff --git a/src/include/access/gist_private.h b/src/include/access/gist_private.h
index 36ed7244ba..c1714a6ccc 100644
--- a/src/include/access/gist_private.h
+++ b/src/include/access/gist_private.h
@@ -79,6 +79,8 @@ typedef struct GISTSTATE
 	MemoryContext tempCxt;		/* short-term context for calling functions */
 
 	TupleDesc	tupdesc;		/* index's tuple descriptor */
+	TupleDesc	truncTupdesc;	/* truncated tuple descriptor
+								 * for internal pages */
 	TupleDesc	fetchTupdesc;	/* tuple descriptor for tuples returned in an
 								 * index-only scan */
 
@@ -458,7 +460,8 @@ extern IndexTuple gistgetadjusted(Relation r,
 				IndexTuple addtup,
 				GISTSTATE *giststate);
 extern IndexTuple gistFormTuple(GISTSTATE *giststate,
-			  Relation r, Datum *attdata, bool *isnull, bool isleaf);
+			  Relation r, Datum *attdata, bool *isnull, bool isleaf,
+			  bool isTruncated);
 
 extern OffsetNumber gistchoose(Relation r, Page p,
 		   IndexTuple it,
diff --git a/src/test/regress/expected/amutils.out b/src/test/regress/expected/amutils.out
index 4570a39b05..d92a6d12c6 100644
--- a/src/test/regress/expected/amutils.out
+++ b/src/test/regress/expected/amutils.out
@@ -75,7 +75,7 @@ select prop,
  can_unique         | f  |       | 
  can_multi_col      | t  |       | 
  can_exclude        | t  |       | 
- can_include        | f  |       | 
+ can_include        | t  |       | 
  bogus              |    |       | 
 (19 rows)
 
@@ -159,7 +159,7 @@ select amname, prop, pg_indexam_has_property(a.oid, prop) as p
  gist   | can_unique    | f
  gist   | can_multi_col | t
  gist   | can_exclude   | t
- gist   | can_include   | f
+ gist   | can_include   | t
  gist   | bogus         | 
  hash   | can_order     | f
  hash   | can_unique    | f
diff --git a/src/test/regress/expected/index_including.out b/src/test/regress/expected/index_including.out
index 16b4be34de..0def30c881 100644
--- a/src/test/regress/expected/index_including.out
+++ b/src/test/regress/expected/index_including.out
@@ -313,22 +313,20 @@ SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl' ORDER BY indexname;
 
 DROP TABLE tbl;
 /*
- * 7. Check various AMs. All but btree must fail.
+ * 7. Check various AMs. All but btree and gist must fail.
  */
 CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
 CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "brin" does not support included columns
-CREATE INDEX on tbl USING gist(c3) INCLUDE (c4);
-ERROR:  access method "gist" does not support included columns
+CREATE INDEX on tbl USING gist(c4) INCLUDE (c3);
 CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
 ERROR:  access method "spgist" does not support included columns
 CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "gin" does not support included columns
 CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "hash" does not support included columns
-CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
+CREATE INDEX on tbl USING rtree(c4) INCLUDE (c3, c1);
 NOTICE:  substituting access method "gist" for obsolete method "rtree"
-ERROR:  access method "gist" does not support included columns
 CREATE INDEX on tbl USING btree(c1, c2) INCLUDE (c3, c4);
 DROP TABLE tbl;
 /*
diff --git a/src/test/regress/expected/index_including_gist.out b/src/test/regress/expected/index_including_gist.out
new file mode 100644
index 0000000000..2e27acb979
--- /dev/null
+++ b/src/test/regress/expected/index_including_gist.out
@@ -0,0 +1,139 @@
+/*
+ * 1.1. test CREATE INDEX with buffered build
+ */
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+                                  pg_get_indexdef                                  
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+ c1 | c2 | c3 |     c4      
+----+----+----+-------------
+  1 |  2 |  3 | (2,3),(1,2)
+  2 |  4 |  6 | (4,5),(2,3)
+  3 |  6 |  9 | (6,7),(3,4)
+  4 |  8 | 12 | (8,9),(4,5)
+(4 rows)
+
+SET enable_bitmapscan TO off;
+EXPLAIN SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                                     QUERY PLAN                                     
+------------------------------------------------------------------------------------
+ Index Only Scan using tbl_gist_idx on tbl_gist  (cost=0.27..36.41 rows=8 width=44)
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+/*
+ * 1.2. test CREATE INDEX with inserts
+ */
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+                                  pg_get_indexdef                                  
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+ c1 | c2 | c3 |     c4      
+----+----+----+-------------
+  1 |  2 |  3 | (2,3),(1,2)
+  2 |  4 |  6 | (4,5),(2,3)
+  3 |  6 |  9 | (6,7),(3,4)
+  4 |  8 | 12 | (8,9),(4,5)
+(4 rows)
+
+SET enable_bitmapscan TO off;
+EXPLAIN SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                                     QUERY PLAN                                     
+------------------------------------------------------------------------------------
+ Index Only Scan using tbl_gist_idx on tbl_gist  (cost=0.28..36.41 rows=8 width=44)
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+/*
+ * 2. CREATE INDEX CONCURRENTLY
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX CONCURRENTLY tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                     indexdef                                      
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+DROP TABLE tbl_gist;
+/*
+ * 3. REINDEX
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                   indexdef                                    
+-------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c3)
+(1 row)
+
+REINDEX INDEX tbl_gist_idx;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                   indexdef                                    
+-------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c3)
+(1 row)
+
+ALTER TABLE tbl_gist DROP COLUMN c1;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+ indexdef 
+----------
+(0 rows)
+
+DROP TABLE tbl_gist;
+/*
+ * 4. Update, delete values in indexed table.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+UPDATE tbl_gist SET c1 = 100 WHERE c1 = 2;
+UPDATE tbl_gist SET c1 = 1 WHERE c1 = 3;
+DELETE FROM tbl_gist WHERE c1 = 5 OR c3 = 12;
+DROP TABLE tbl_gist;
+/*
+ * 5. Alter column type.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+ALTER TABLE tbl_gist ALTER c1 TYPE bigint;
+ALTER TABLE tbl_gist ALTER c3 TYPE bigint;
+\d tbl_gist
+              Table "public.tbl_gist"
+ Column |  Type   | Collation | Nullable | Default 
+--------+---------+-----------+----------+---------
+ c1     | bigint  |           |          | 
+ c2     | integer |           |          | 
+ c3     | bigint  |           |          | 
+ c4     | box     |           |          | 
+Indexes:
+    "tbl_gist_idx" gist (c4) INCLUDE (c1, c3)
+
+DROP TABLE tbl_gist;
diff --git a/src/test/regress/parallel_schedule b/src/test/regress/parallel_schedule
index cc0bbf5db9..5684291ab0 100644
--- a/src/test/regress/parallel_schedule
+++ b/src/test/regress/parallel_schedule
@@ -55,7 +55,7 @@ test: copy copyselect copydml
 # ----------
 test: create_misc create_operator create_procedure
 # These depend on the above two
-test: create_index create_view index_including
+test: create_index create_view index_including index_including_gist
 
 # ----------
 # Another group of parallel tests
diff --git a/src/test/regress/serial_schedule b/src/test/regress/serial_schedule
index 0c10c7100c..309ccec81a 100644
--- a/src/test/regress/serial_schedule
+++ b/src/test/regress/serial_schedule
@@ -63,6 +63,7 @@ test: create_operator
 test: create_procedure
 test: create_index
 test: index_including
+test: index_including_gist
 test: create_view
 test: create_aggregate
 test: create_function_3
diff --git a/src/test/regress/sql/index_including.sql b/src/test/regress/sql/index_including.sql
index ef5fd882f5..53cffb15fc 100644
--- a/src/test/regress/sql/index_including.sql
+++ b/src/test/regress/sql/index_including.sql
@@ -173,15 +173,15 @@ SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl' ORDER BY indexname;
 DROP TABLE tbl;
 
 /*
- * 7. Check various AMs. All but btree must fail.
+ * 7. Check various AMs. All but btree and gist must fail.
  */
 CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
 CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
-CREATE INDEX on tbl USING gist(c3) INCLUDE (c4);
+CREATE INDEX on tbl USING gist(c4) INCLUDE (c3);
 CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
 CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
 CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
-CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
+CREATE INDEX on tbl USING rtree(c4) INCLUDE (c3, c1);
 CREATE INDEX on tbl USING btree(c1, c2) INCLUDE (c3, c4);
 DROP TABLE tbl;
 
diff --git a/src/test/regress/sql/index_including_gist.sql b/src/test/regress/sql/index_including_gist.sql
new file mode 100644
index 0000000000..612d6ca210
--- /dev/null
+++ b/src/test/regress/sql/index_including_gist.sql
@@ -0,0 +1,80 @@
+/*
+ * 1.1. test CREATE INDEX with buffered build
+ */
+
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO off;
+EXPLAIN SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+
+/*
+ * 1.2. test CREATE INDEX with inserts
+ */
+
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO off;
+EXPLAIN SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+
+/*
+ * 2. CREATE INDEX CONCURRENTLY
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX CONCURRENTLY tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+DROP TABLE tbl_gist;
+
+
+/*
+ * 3. REINDEX
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+REINDEX INDEX tbl_gist_idx;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+ALTER TABLE tbl_gist DROP COLUMN c1;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+DROP TABLE tbl_gist;
+
+/*
+ * 4. Update, delete values in indexed table.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+UPDATE tbl_gist SET c1 = 100 WHERE c1 = 2;
+UPDATE tbl_gist SET c1 = 1 WHERE c1 = 3;
+DELETE FROM tbl_gist WHERE c1 = 5 OR c3 = 12;
+DROP TABLE tbl_gist;
+
+/*
+ * 5. Alter column type.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+ALTER TABLE tbl_gist ALTER c1 TYPE bigint;
+ALTER TABLE tbl_gist ALTER c3 TYPE bigint;
+\d tbl_gist
+DROP TABLE tbl_gist;
-- 
2.17.2 (Apple Git-113)

#16

andreas@proxel.se

almost 7 years ago

In reply to: Andrey Borodin (#15)

Re: Covering GiST indexes

On 11/26/18 5:56 PM, Andrey Borodin wrote:

Here's rebased version. Thanks!

Here is my review.

= General

The features seems useful and an obvious extension of covering B-trees,
and a potentially useful feature together with exclusion constraints.

The patch still applies (with git apply -3), compiles and passes the
test suite.

From some testing it seems like it works as expected.

= Code

* Have some minor style issues like that there should be spaces around
|| (in gistcanreturn()) and ? and : (in gistFormTuple()).

* I do not see any need for adding the new parameter to gistFormTuple.
As far as I can tell isTruncated is always equal to !isleaf.

* The comment below from gistFormTuple() does not look correct. No
allocation is taking place.

/*
* Allocate each included attribute.
*/

* Why do you set a supportCollation for the included columns? As far as
I can tell the collation is only used by the various GiST support
functions. Also in theory we should be able to skip initializing these
array entires, but it is probably for the best that we do.

* I think you should check for the attno first in gistcanreturn() to
take advantage of the short circuiting.

* I am no fan of the tupdesc vs truncTupdesc separation and think that
it is a potential hazard, but I do not have any better suggestion right now.

* There is no test case for exclusion constraints, and I feel since that
is one of the more important uses we should probably have at least one
such test case.

= Minor comments

* I think that "the" should have been kept in the text below.

-        Currently, only the B-tree index access method supports this 
feature.
+        Currently, B-tree and GiST index access methods supports this 
feature.

* I am not a native speaker but I think it should be "use the INCLUDE
clause" in the diff below, and I think it also should be "without any
GiST operator class".

+   A GiST index can be covering, i.e. use <literal>INCLUDE</literal> 
clause.
+   Included columns can have data types without GiST operator class. 
Included
+   attributes will be stored uncompressed.

* I think you should write something like "Included attributes always
support index-only scans." rather than "Included attributes can be
returned." in the comment for gistcanreturn().

* Why the random noise in the diff below? I think using "(c3) INCLUDE
(c4)" for both gist and rtree results in a cleaner patch.

  CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
  CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
-CREATE INDEX on tbl USING gist(c3) INCLUDE (c4);
+CREATE INDEX on tbl USING gist(c4) INCLUDE (c3);
  CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
  CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
  CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
-CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
+CREATE INDEX on tbl USING rtree(c4) INCLUDE (c3, c1);
  CREATE INDEX on tbl USING btree(c1, c2) INCLUDE (c3, c4);

#17

x4mmm@yandex-team.ru

almost 7 years ago

In reply to: Andreas Karlsson (#16)

1 attachment(s)

Re: Covering GiST indexes

Thank you very much for reviewing the patch!

28 янв. 2019 г., в 12:15, Andreas Karlsson <andreas@proxel.se> написал(а):

= Code

* Have some minor style issues like that there should be spaces around || (in gistcanreturn()) and ? and : (in gistFormTuple()).

Fixed.

* I do not see any need for adding the new parameter to gistFormTuple. As far as I can tell isTruncated is always equal to !isleaf.

You are right. I've removed isTruncated parameter.

* The comment below from gistFormTuple() does not look correct. No allocation is taking place.

/*
* Allocate each included attribute.
*/

Fixed.

* Why do you set a supportCollation for the included columns? As far as I can tell the collation is only used by the various GiST support functions. Also in theory we should be able to skip initializing these array entires, but it is probably for the best that we do.

Removed supportCollation.

* I think you should check for the attno first in gistcanreturn() to take advantage of the short circuiting.

Done.

* I am no fan of the tupdesc vs truncTupdesc separation and think that it is a potential hazard, but I do not have any better suggestion right now.

B-tree is copying tupdesc every time they truncate tuple. We need tuple truncation a little more often: when we are doing page split, we have to form all page tuples, truncated.
Initially, I've optimized only this case, but this led to prepared tupledesc for truncated tuples.

* There is no test case for exclusion constraints, and I feel since that is one of the more important uses we should probably have at least one such test case.

Actually, I did not understand this test case. Can you elaborate more on this? How included attributes should participate in exclude index? What for?

= Minor comments

* I think that "the" should have been kept in the text below.
-        Currently, only the B-tree index access method supports this feature.
+        Currently, B-tree and GiST index access methods supports this feature.

Fixed.

* I am not a native speaker but I think it should be "use the INCLUDE clause" in the diff below, and I think it also should be "without any GiST operator class".
+   A GiST index can be covering, i.e. use <literal>INCLUDE</literal> clause.
+   Included columns can have data types without GiST operator class. Included
+   attributes will be stored uncompressed.

Fixed.

* I think you should write something like "Included attributes always support index-only scans." rather than "Included attributes can be returned." in the comment for gistcanreturn().

Fixed, but slightly reworded.

* Why the random noise in the diff below? I think using "(c3) INCLUDE (c4)" for both gist and rtree results in a cleaner patch.

I've used columns with and without opclass in INCLUDE. This led to these seemingly random changes.

PFA v6. Thanks for reviewing!

Best regards, Andrey Borodin.

Attachments:

0001-Covering-GiST-v6.patchapplication/octet-stream; name=0001-Covering-GiST-v6.patch; x-unix-mode=0644Download

From 4c822f5daa33b7ea9e3cc2897035f1921a2a6ce0 Mon Sep 17 00:00:00 2001
From: Andrey <amborodin@acm.org>
Date: Mon, 26 Nov 2018 21:55:45 +0500
Subject: [PATCH] Covering GiST v6

This patch allow adding INCLUDE colums to GiST index.
These colums do not participate in GiST tree build,
are not used for chosing subtree for inserts or splits.
But they can be fetched during IndexOnlyScan.
---
 doc/src/sgml/ref/create_index.sgml            |   2 +-
 doc/src/sgml/textsearch.sgml                  |   6 +
 src/backend/access/gist/gist.c                |  26 +++-
 src/backend/access/gist/gistget.c             |   6 +-
 src/backend/access/gist/gistscan.c            |  12 +-
 src/backend/access/gist/gistsplit.c           |  12 +-
 src/backend/access/gist/gistutil.c            |  44 ++++--
 src/include/access/gist_private.h             |   2 +
 src/test/regress/expected/amutils.out         |   4 +-
 src/test/regress/expected/index_including.out |   8 +-
 .../regress/expected/index_including_gist.out | 139 ++++++++++++++++++
 src/test/regress/parallel_schedule            |   2 +-
 src/test/regress/serial_schedule              |   1 +
 src/test/regress/sql/index_including.sql      |   6 +-
 src/test/regress/sql/index_including_gist.sql |  80 ++++++++++
 15 files changed, 315 insertions(+), 35 deletions(-)
 create mode 100644 src/test/regress/expected/index_including_gist.out
 create mode 100644 src/test/regress/sql/index_including_gist.sql

diff --git a/doc/src/sgml/ref/create_index.sgml b/doc/src/sgml/ref/create_index.sgml
index ad619cdcfe..cd3409eb76 100644
--- a/doc/src/sgml/ref/create_index.sgml
+++ b/doc/src/sgml/ref/create_index.sgml
@@ -181,7 +181,7 @@ CREATE [ UNIQUE ] INDEX [ CONCURRENTLY ] [ [ IF NOT EXISTS ] <replaceable class=
        </para>
 
        <para>
-        Currently, only the B-tree index access method supports this feature.
+        Currently, B-tree and GiST index access methods supports this feature.
         In B-tree indexes, the values of columns listed in the
         <literal>INCLUDE</literal> clause are included in leaf tuples which
         correspond to heap tuples, but are not included in upper-level
diff --git a/doc/src/sgml/textsearch.sgml b/doc/src/sgml/textsearch.sgml
index ecebade767..0305a5ad38 100644
--- a/doc/src/sgml/textsearch.sgml
+++ b/doc/src/sgml/textsearch.sgml
@@ -3674,6 +3674,12 @@ SELECT plainto_tsquery('supernovae stars');
    retrieved to see if the match is correct.
   </para>
 
+  <para>
+   A GiST index can be covering, i.e. use <literal>INCLUDE</literal> clause.
+   Included columns can have data types without GiST operator class. Included
+   attributes will be stored uncompressed.
+  </para>
+
   <para>
    Lossiness causes performance degradation due to unnecessary fetches of table
    records that turn out to be false matches.  Since random access to table
diff --git a/src/backend/access/gist/gist.c b/src/backend/access/gist/gist.c
index b75b3a8dac..99647bf709 100644
--- a/src/backend/access/gist/gist.c
+++ b/src/backend/access/gist/gist.c
@@ -75,7 +75,7 @@ gisthandler(PG_FUNCTION_ARGS)
 	amroutine->amclusterable = true;
 	amroutine->ampredlocks = true;
 	amroutine->amcanparallel = false;
-	amroutine->amcaninclude = false;
+	amroutine->amcaninclude = true;
 	amroutine->amkeytype = InvalidOid;
 
 	amroutine->ambuild = gistbuild;
@@ -170,7 +170,7 @@ gistinsert(Relation r, Datum *values, bool *isnull,
 	oldCxt = MemoryContextSwitchTo(giststate->tempCxt);
 
 	itup = gistFormTuple(giststate, r,
-						 values, isnull, true /* size is currently bogus */ );
+						 values, isnull, true /* size is currently bogus */);
 	itup->t_tid = *ht_ctid;
 
 	gistdoinsert(r, itup, 0, giststate, heapRel);
@@ -1382,8 +1382,8 @@ gistSplit(Relation r,
 						IndexTupleSize(itup[0]), GiSTPageSize,
 						RelationGetRelationName(r))));
 
-	memset(v.spl_lisnull, true, sizeof(bool) * giststate->tupdesc->natts);
-	memset(v.spl_risnull, true, sizeof(bool) * giststate->tupdesc->natts);
+	memset(v.spl_lisnull, true, sizeof(bool) * giststate->truncTupdesc->natts);
+	memset(v.spl_risnull, true, sizeof(bool) * giststate->truncTupdesc->natts);
 	gistSplitByKey(r, page, itup, len, giststate, &v, 0);
 
 	/* form left and right vector */
@@ -1462,8 +1462,10 @@ initGISTstate(Relation index)
 	giststate->scanCxt = scanCxt;
 	giststate->tempCxt = scanCxt;	/* caller must change this if needed */
 	giststate->tupdesc = index->rd_att;
+	giststate->truncTupdesc = CreateTupleDescCopyConstr(index->rd_att);
+	giststate->truncTupdesc->natts = IndexRelationGetNumberOfKeyAttributes(index);
 
-	for (i = 0; i < index->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(index); i++)
 	{
 		fmgr_info_copy(&(giststate->consistentFn[i]),
 					   index_getprocinfo(index, i + 1, GIST_CONSISTENT_PROC),
@@ -1531,6 +1533,20 @@ initGISTstate(Relation index)
 			giststate->supportCollation[i] = DEFAULT_COLLATION_OID;
 	}
 
+	for (; i < index->rd_att->natts; i++)
+	{
+		giststate->consistentFn[i].fn_oid = InvalidOid;
+		giststate->unionFn[i].fn_oid = InvalidOid;
+		giststate->compressFn[i].fn_oid = InvalidOid;
+		giststate->decompressFn[i].fn_oid = InvalidOid;
+		giststate->penaltyFn[i].fn_oid = InvalidOid;
+		giststate->picksplitFn[i].fn_oid = InvalidOid;
+		giststate->equalFn[i].fn_oid = InvalidOid;
+		giststate->distanceFn[i].fn_oid = InvalidOid;
+		giststate->fetchFn[i].fn_oid = InvalidOid;
+		giststate->supportCollation[i] = InvalidOid;
+	}
+
 	MemoryContextSwitchTo(oldCxt);
 
 	return giststate;
diff --git a/src/backend/access/gist/gistget.c b/src/backend/access/gist/gistget.c
index a96ef5c3ac..61821da831 100644
--- a/src/backend/access/gist/gistget.c
+++ b/src/backend/access/gist/gistget.c
@@ -769,12 +769,14 @@ gistgetbitmap(IndexScanDesc scan, TIDBitmap *tbm)
  *
  * Opclasses that implement a fetch function support index-only scans.
  * Opclasses without compression functions also support index-only scans.
+ * Included attributes always can be fetched for index-only scans.
  */
 bool
 gistcanreturn(Relation index, int attno)
 {
-	if (OidIsValid(index_getprocid(index, attno, GIST_FETCH_PROC)) ||
-		!OidIsValid(index_getprocid(index, attno, GIST_COMPRESS_PROC)))
+	if (attno > IndexRelationGetNumberOfKeyAttributes(index) ||
+		!OidIsValid(index_getprocid(index, attno, GIST_FETCH_PROC)) ||
+		OidIsValid(index_getprocid(index, attno, GIST_COMPRESS_PROC)))
 		return true;
 	else
 		return false;
diff --git a/src/backend/access/gist/gistscan.c b/src/backend/access/gist/gistscan.c
index 78a8ede794..4fffc8b33a 100644
--- a/src/backend/access/gist/gistscan.c
+++ b/src/backend/access/gist/gistscan.c
@@ -158,6 +158,7 @@ gistrescan(IndexScanDesc scan, ScanKey key, int nkeys,
 	if (scan->xs_want_itup && !scan->xs_hitupdesc)
 	{
 		int			natts;
+		int			nkeyatts;
 		int			attno;
 
 		/*
@@ -167,13 +168,22 @@ gistrescan(IndexScanDesc scan, ScanKey key, int nkeys,
 		 * types.
 		 */
 		natts = RelationGetNumberOfAttributes(scan->indexRelation);
+		nkeyatts = IndexRelationGetNumberOfKeyAttributes(scan->indexRelation);
 		so->giststate->fetchTupdesc = CreateTemplateTupleDesc(natts);
-		for (attno = 1; attno <= natts; attno++)
+		for (attno = 1; attno <= nkeyatts; attno++)
 		{
 			TupleDescInitEntry(so->giststate->fetchTupdesc, attno, NULL,
 							   scan->indexRelation->rd_opcintype[attno - 1],
 							   -1, 0);
 		}
+
+		for (; attno <= natts; attno++)
+		{
+			/* taking opcintype from giststate->tupdesc */
+			TupleDescInitEntry(so->giststate->fetchTupdesc, attno, NULL,
+					TupleDescAttr(so->giststate->tupdesc, attno - 1)->atttypid,
+							   -1, 0);
+		}
 		scan->xs_hitupdesc = so->giststate->fetchTupdesc;
 
 		/* Also create a memory context that will hold the returned tuples */
diff --git a/src/backend/access/gist/gistsplit.c b/src/backend/access/gist/gistsplit.c
index f210e0c39f..cf3af2821a 100644
--- a/src/backend/access/gist/gistsplit.c
+++ b/src/backend/access/gist/gistsplit.c
@@ -207,7 +207,7 @@ placeOne(Relation r, GISTSTATE *giststate, GistSplitVector *v,
 	gistDeCompressAtt(giststate, r, itup, NULL, (OffsetNumber) 0,
 					  identry, isnull);
 
-	for (; attno < giststate->tupdesc->natts; attno++)
+	for (; attno < giststate->truncTupdesc->natts; attno++)
 	{
 		float		lpenalty,
 					rpenalty;
@@ -485,7 +485,7 @@ gistUserPicksplit(Relation r, GistEntryVector *entryvec, int attno, GistSplitVec
 	 */
 	v->spl_dontcare = NULL;
 
-	if (attno + 1 < giststate->tupdesc->natts)
+	if (attno + 1 < giststate->truncTupdesc->natts)
 	{
 		int			NumDontCare;
 
@@ -657,7 +657,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 		 */
 		v->spl_risnull[attno] = v->spl_lisnull[attno] = true;
 
-		if (attno + 1 < giststate->tupdesc->natts)
+		if (attno + 1 < giststate->truncTupdesc->natts)
 			gistSplitByKey(r, page, itup, len, giststate, v, attno + 1);
 		else
 			gistSplitHalf(&v->splitVector, len);
@@ -683,7 +683,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 				v->splitVector.spl_left[v->splitVector.spl_nleft++] = i;
 
 		/* Compute union keys, unless outer recursion level will handle it */
-		if (attno == 0 && giststate->tupdesc->natts == 1)
+		if (attno == 0 && giststate->truncTupdesc->natts == 1)
 		{
 			v->spl_dontcare = NULL;
 			gistunionsubkey(giststate, itup, v);
@@ -700,7 +700,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 			 * Splitting on attno column is not optimal, so consider
 			 * redistributing don't-care tuples according to the next column
 			 */
-			Assert(attno + 1 < giststate->tupdesc->natts);
+			Assert(attno + 1 < giststate->truncTupdesc->natts);
 
 			if (v->spl_dontcare == NULL)
 			{
@@ -771,7 +771,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 	 * that PickSplit (or the special cases above) produced correct union
 	 * datums.
 	 */
-	if (attno == 0 && giststate->tupdesc->natts > 1)
+	if (attno == 0 && giststate->truncTupdesc->natts > 1)
 	{
 		v->spl_dontcare = NULL;
 		gistunionsubkey(giststate, itup, v);
diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c
index 8d3dfad27b..79b1adc656 100644
--- a/src/backend/access/gist/gistutil.c
+++ b/src/backend/access/gist/gistutil.c
@@ -160,7 +160,7 @@ gistMakeUnionItVec(GISTSTATE *giststate, IndexTuple *itvec, int len,
 
 	evec = (GistEntryVector *) palloc((len + 2) * sizeof(GISTENTRY) + GEVHDRSZ);
 
-	for (i = 0; i < giststate->tupdesc->natts; i++)
+	for (i = 0; i < giststate->truncTupdesc->natts; i++)
 	{
 		int			j;
 
@@ -296,7 +296,7 @@ gistDeCompressAtt(GISTSTATE *giststate, Relation r, IndexTuple tuple, Page p,
 {
 	int			i;
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		Datum		datum;
 
@@ -329,7 +329,7 @@ gistgetadjusted(Relation r, IndexTuple oldtup, IndexTuple addtup, GISTSTATE *gis
 	gistDeCompressAtt(giststate, r, addtup, NULL,
 					  (OffsetNumber) 0, addentries, addisnull);
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		gistMakeUnionKey(giststate, i,
 						 oldentries + i, oldisnull[i],
@@ -442,7 +442,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 		zero_penalty = true;
 
 		/* Loop over index attributes. */
-		for (j = 0; j < r->rd_att->natts; j++)
+		for (j = 0; j < IndexRelationGetNumberOfKeyAttributes(r); j++)
 		{
 			Datum		datum;
 			float		usize;
@@ -470,7 +470,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 				result = i;
 				best_penalty[j] = usize;
 
-				if (j < r->rd_att->natts - 1)
+				if (j < IndexRelationGetNumberOfKeyAttributes(r) - 1)
 					best_penalty[j + 1] = -1;
 
 				/* we have new best, so reset keep-it decision */
@@ -500,7 +500,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 		 * If we looped past the last column, and did not update "result",
 		 * then this tuple is exactly as good as the prior best tuple.
 		 */
-		if (j == r->rd_att->natts && result != i)
+		if (j == IndexRelationGetNumberOfKeyAttributes(r) && result != i)
 		{
 			if (keep_current_best == -1)
 			{
@@ -575,11 +575,13 @@ gistFormTuple(GISTSTATE *giststate, Relation r,
 	Datum		compatt[INDEX_MAX_KEYS];
 	int			i;
 	IndexTuple	res;
+	/* currently we truncate tuples iif they are on internal pages */
+	bool isTruncated = !isleaf;
 
 	/*
 	 * Call the compress method on each attribute.
 	 */
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		if (isnull[i])
 			compatt[i] = (Datum) 0;
@@ -602,7 +604,23 @@ gistFormTuple(GISTSTATE *giststate, Relation r,
 		}
 	}
 
-	res = index_form_tuple(giststate->tupdesc, compatt, isnull);
+	if (!isTruncated)
+	{
+		/*
+		 * Emplace each included attribute.
+		 */
+		for (; i < r->rd_att->natts; i++)
+		{
+			if (isnull[i])
+				compatt[i] = (Datum) 0;
+			else
+			{
+				compatt[i] = attdata[i];
+			}
+		}
+	}
+
+	res = index_form_tuple(isTruncated ? giststate->truncTupdesc : giststate->tupdesc, compatt, isnull);
 
 	/*
 	 * The offset number on tuples on internal pages is unused. For historical
@@ -644,7 +662,7 @@ gistFetchTuple(GISTSTATE *giststate, Relation r, IndexTuple tuple)
 	bool		isnull[INDEX_MAX_KEYS];
 	int			i;
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		Datum		datum;
 
@@ -679,6 +697,14 @@ gistFetchTuple(GISTSTATE *giststate, Relation r, IndexTuple tuple)
 			fetchatt[i] = (Datum) 0;
 		}
 	}
+
+	/*
+	 * Get each included attribute.
+	 */
+	for (; i < r->rd_att->natts; i++)
+	{
+		fetchatt[i] = index_getattr(tuple, i + 1, giststate->tupdesc, &isnull[i]);
+	}
 	MemoryContextSwitchTo(oldcxt);
 
 	return heap_form_tuple(giststate->fetchTupdesc, fetchatt, isnull);
diff --git a/src/include/access/gist_private.h b/src/include/access/gist_private.h
index 3698942f9d..ccf06e676c 100644
--- a/src/include/access/gist_private.h
+++ b/src/include/access/gist_private.h
@@ -79,6 +79,8 @@ typedef struct GISTSTATE
 	MemoryContext tempCxt;		/* short-term context for calling functions */
 
 	TupleDesc	tupdesc;		/* index's tuple descriptor */
+	TupleDesc	truncTupdesc;	/* truncated tuple descriptor
+								 * for internal pages */
 	TupleDesc	fetchTupdesc;	/* tuple descriptor for tuples returned in an
 								 * index-only scan */
 
diff --git a/src/test/regress/expected/amutils.out b/src/test/regress/expected/amutils.out
index 4570a39b05..d92a6d12c6 100644
--- a/src/test/regress/expected/amutils.out
+++ b/src/test/regress/expected/amutils.out
@@ -75,7 +75,7 @@ select prop,
  can_unique         | f  |       | 
  can_multi_col      | t  |       | 
  can_exclude        | t  |       | 
- can_include        | f  |       | 
+ can_include        | t  |       | 
  bogus              |    |       | 
 (19 rows)
 
@@ -159,7 +159,7 @@ select amname, prop, pg_indexam_has_property(a.oid, prop) as p
  gist   | can_unique    | f
  gist   | can_multi_col | t
  gist   | can_exclude   | t
- gist   | can_include   | f
+ gist   | can_include   | t
  gist   | bogus         | 
  hash   | can_order     | f
  hash   | can_unique    | f
diff --git a/src/test/regress/expected/index_including.out b/src/test/regress/expected/index_including.out
index 16b4be34de..0def30c881 100644
--- a/src/test/regress/expected/index_including.out
+++ b/src/test/regress/expected/index_including.out
@@ -313,22 +313,20 @@ SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl' ORDER BY indexname;
 
 DROP TABLE tbl;
 /*
- * 7. Check various AMs. All but btree must fail.
+ * 7. Check various AMs. All but btree and gist must fail.
  */
 CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
 CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "brin" does not support included columns
-CREATE INDEX on tbl USING gist(c3) INCLUDE (c4);
-ERROR:  access method "gist" does not support included columns
+CREATE INDEX on tbl USING gist(c4) INCLUDE (c3);
 CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
 ERROR:  access method "spgist" does not support included columns
 CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "gin" does not support included columns
 CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "hash" does not support included columns
-CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
+CREATE INDEX on tbl USING rtree(c4) INCLUDE (c3, c1);
 NOTICE:  substituting access method "gist" for obsolete method "rtree"
-ERROR:  access method "gist" does not support included columns
 CREATE INDEX on tbl USING btree(c1, c2) INCLUDE (c3, c4);
 DROP TABLE tbl;
 /*
diff --git a/src/test/regress/expected/index_including_gist.out b/src/test/regress/expected/index_including_gist.out
new file mode 100644
index 0000000000..2e27acb979
--- /dev/null
+++ b/src/test/regress/expected/index_including_gist.out
@@ -0,0 +1,139 @@
+/*
+ * 1.1. test CREATE INDEX with buffered build
+ */
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+                                  pg_get_indexdef                                  
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+ c1 | c2 | c3 |     c4      
+----+----+----+-------------
+  1 |  2 |  3 | (2,3),(1,2)
+  2 |  4 |  6 | (4,5),(2,3)
+  3 |  6 |  9 | (6,7),(3,4)
+  4 |  8 | 12 | (8,9),(4,5)
+(4 rows)
+
+SET enable_bitmapscan TO off;
+EXPLAIN SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                                     QUERY PLAN                                     
+------------------------------------------------------------------------------------
+ Index Only Scan using tbl_gist_idx on tbl_gist  (cost=0.27..36.41 rows=8 width=44)
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+/*
+ * 1.2. test CREATE INDEX with inserts
+ */
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+                                  pg_get_indexdef                                  
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+ c1 | c2 | c3 |     c4      
+----+----+----+-------------
+  1 |  2 |  3 | (2,3),(1,2)
+  2 |  4 |  6 | (4,5),(2,3)
+  3 |  6 |  9 | (6,7),(3,4)
+  4 |  8 | 12 | (8,9),(4,5)
+(4 rows)
+
+SET enable_bitmapscan TO off;
+EXPLAIN SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                                     QUERY PLAN                                     
+------------------------------------------------------------------------------------
+ Index Only Scan using tbl_gist_idx on tbl_gist  (cost=0.28..36.41 rows=8 width=44)
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+/*
+ * 2. CREATE INDEX CONCURRENTLY
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX CONCURRENTLY tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                     indexdef                                      
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+DROP TABLE tbl_gist;
+/*
+ * 3. REINDEX
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                   indexdef                                    
+-------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c3)
+(1 row)
+
+REINDEX INDEX tbl_gist_idx;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                   indexdef                                    
+-------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c3)
+(1 row)
+
+ALTER TABLE tbl_gist DROP COLUMN c1;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+ indexdef 
+----------
+(0 rows)
+
+DROP TABLE tbl_gist;
+/*
+ * 4. Update, delete values in indexed table.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+UPDATE tbl_gist SET c1 = 100 WHERE c1 = 2;
+UPDATE tbl_gist SET c1 = 1 WHERE c1 = 3;
+DELETE FROM tbl_gist WHERE c1 = 5 OR c3 = 12;
+DROP TABLE tbl_gist;
+/*
+ * 5. Alter column type.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+ALTER TABLE tbl_gist ALTER c1 TYPE bigint;
+ALTER TABLE tbl_gist ALTER c3 TYPE bigint;
+\d tbl_gist
+              Table "public.tbl_gist"
+ Column |  Type   | Collation | Nullable | Default 
+--------+---------+-----------+----------+---------
+ c1     | bigint  |           |          | 
+ c2     | integer |           |          | 
+ c3     | bigint  |           |          | 
+ c4     | box     |           |          | 
+Indexes:
+    "tbl_gist_idx" gist (c4) INCLUDE (c1, c3)
+
+DROP TABLE tbl_gist;
diff --git a/src/test/regress/parallel_schedule b/src/test/regress/parallel_schedule
index cc0bbf5db9..5684291ab0 100644
--- a/src/test/regress/parallel_schedule
+++ b/src/test/regress/parallel_schedule
@@ -55,7 +55,7 @@ test: copy copyselect copydml
 # ----------
 test: create_misc create_operator create_procedure
 # These depend on the above two
-test: create_index create_view index_including
+test: create_index create_view index_including index_including_gist
 
 # ----------
 # Another group of parallel tests
diff --git a/src/test/regress/serial_schedule b/src/test/regress/serial_schedule
index 0c10c7100c..309ccec81a 100644
--- a/src/test/regress/serial_schedule
+++ b/src/test/regress/serial_schedule
@@ -63,6 +63,7 @@ test: create_operator
 test: create_procedure
 test: create_index
 test: index_including
+test: index_including_gist
 test: create_view
 test: create_aggregate
 test: create_function_3
diff --git a/src/test/regress/sql/index_including.sql b/src/test/regress/sql/index_including.sql
index ef5fd882f5..53cffb15fc 100644
--- a/src/test/regress/sql/index_including.sql
+++ b/src/test/regress/sql/index_including.sql
@@ -173,15 +173,15 @@ SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl' ORDER BY indexname;
 DROP TABLE tbl;
 
 /*
- * 7. Check various AMs. All but btree must fail.
+ * 7. Check various AMs. All but btree and gist must fail.
  */
 CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
 CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
-CREATE INDEX on tbl USING gist(c3) INCLUDE (c4);
+CREATE INDEX on tbl USING gist(c4) INCLUDE (c3);
 CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
 CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
 CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
-CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
+CREATE INDEX on tbl USING rtree(c4) INCLUDE (c3, c1);
 CREATE INDEX on tbl USING btree(c1, c2) INCLUDE (c3, c4);
 DROP TABLE tbl;
 
diff --git a/src/test/regress/sql/index_including_gist.sql b/src/test/regress/sql/index_including_gist.sql
new file mode 100644
index 0000000000..612d6ca210
--- /dev/null
+++ b/src/test/regress/sql/index_including_gist.sql
@@ -0,0 +1,80 @@
+/*
+ * 1.1. test CREATE INDEX with buffered build
+ */
+
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO off;
+EXPLAIN SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+
+/*
+ * 1.2. test CREATE INDEX with inserts
+ */
+
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO off;
+EXPLAIN SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+
+/*
+ * 2. CREATE INDEX CONCURRENTLY
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX CONCURRENTLY tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+DROP TABLE tbl_gist;
+
+
+/*
+ * 3. REINDEX
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+REINDEX INDEX tbl_gist_idx;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+ALTER TABLE tbl_gist DROP COLUMN c1;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+DROP TABLE tbl_gist;
+
+/*
+ * 4. Update, delete values in indexed table.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+UPDATE tbl_gist SET c1 = 100 WHERE c1 = 2;
+UPDATE tbl_gist SET c1 = 1 WHERE c1 = 3;
+DELETE FROM tbl_gist WHERE c1 = 5 OR c3 = 12;
+DROP TABLE tbl_gist;
+
+/*
+ * 5. Alter column type.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+ALTER TABLE tbl_gist ALTER c1 TYPE bigint;
+ALTER TABLE tbl_gist ALTER c3 TYPE bigint;
+\d tbl_gist
+DROP TABLE tbl_gist;
-- 
2.20.1

#18

andreas@proxel.se

almost 7 years ago

In reply to: Andrey Borodin (#17)

Re: Covering GiST indexes

On 1/28/19 7:26 PM, Andrey Borodin wrote:

* I am no fan of the tupdesc vs truncTupdesc separation and think that it is a potential hazard, but I do not have any better suggestion right now.

B-tree is copying tupdesc every time they truncate tuple. We need tuple truncation a little more often: when we are doing page split, we have to form all page tuples, truncated.
Initially, I've optimized only this case, but this led to prepared tupledesc for truncated tuples.

* There is no test case for exclusion constraints, and I feel since that is one of the more important uses we should probably have at least one such test case.

Actually, I did not understand this test case. Can you elaborate more on this? How included attributes should participate in exclude index? What for?

I mean include a table like below among the tests. I feel like this is a
main use case for INCLUDE.

CREATE TABLE t2 (
x int4range,
y int,

EXCLUDE USING gist (x WITH &&) INCLUDE (y)
);

* Why the random noise in the diff below? I think using "(c3) INCLUDE (c4)" for both gist and rtree results in a cleaner patch.

I've used columns with and without opclass in INCLUDE. This led to these seemingly random changes.

I mean the diff would be smaller as the below. It also may make sense to
make both lines "(c3) INCLUDE (c1, c4)".

  CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
  CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
  CREATE INDEX on tbl USING gist(c3) INCLUDE (c4);
  CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
  CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
  CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
-CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
+CREATE INDEX on tbl USING rtree(c3) INCLUDE (c4);
  CREATE INDEX on tbl USING btree(c1, c2) INCLUDE (c3, c4);

Andreas

#19

x4mmm@yandex-team.ru

almost 7 years ago

In reply to: Andreas Karlsson (#18)

1 attachment(s)

Re: Covering GiST indexes

29 янв. 2019 г., в 7:32, Andreas Karlsson <andreas@proxel.se> написал(а):

On 1/28/19 7:26 PM, Andrey Borodin wrote:

* I am no fan of the tupdesc vs truncTupdesc separation and think that it is a potential hazard, but I do not have any better suggestion right now.

B-tree is copying tupdesc every time they truncate tuple. We need tuple truncation a little more often: when we are doing page split, we have to form all page tuples, truncated.
Initially, I've optimized only this case, but this led to prepared tupledesc for truncated tuples.

* There is no test case for exclusion constraints, and I feel since that is one of the more important uses we should probably have at least one such test case.

Actually, I did not understand this test case. Can you elaborate more on this? How included attributes should participate in exclude index? What for?

I mean include a table like below among the tests. I feel like this is a main use case for INCLUDE.

CREATE TABLE t2 (
x int4range,
y int,

EXCLUDE USING gist (x WITH &&) INCLUDE (y)
);

Thanks for the explanation. Added this as case 6 to index_including_gist.

* Why the random noise in the diff below? I think using "(c3) INCLUDE (c4)" for both gist and rtree results in a cleaner patch.

I've used columns with and without opclass in INCLUDE. This led to these seemingly random changes.

I mean the diff would be smaller as the below. It also may make sense to make both lines "(c3) INCLUDE (c1, c4)".
CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
CREATE INDEX on tbl USING gist(c3) INCLUDE (c4);
CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
-CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
+CREATE INDEX on tbl USING rtree(c3) INCLUDE (c4);
CREATE INDEX on tbl USING btree(c1, c2) INCLUDE (c3, c4);

I've took your version of this test and added all variations of included attributes.

PFA v7.

Thanks!

Best regards, Andrey Borodin.

Attachments:

0001-Covering-GiST-v7.patchapplication/octet-stream; name=0001-Covering-GiST-v7.patch; x-unix-mode=0644Download

From d96b8cb03edc11a29d650f21d1495f85ae0f2fd0 Mon Sep 17 00:00:00 2001
From: Andrey <amborodin@acm.org>
Date: Mon, 26 Nov 2018 21:55:45 +0500
Subject: [PATCH] Covering GiST v7

This patch allow adding INCLUDE colums to GiST index.
These colums do not participate in GiST tree build,
are not used for chosing subtree for inserts or splits.
But they can be fetched during IndexOnlyScan.
---
 doc/src/sgml/ref/create_index.sgml            |   4 +-
 doc/src/sgml/textsearch.sgml                  |   6 +
 src/backend/access/gist/gist.c                |  26 ++-
 src/backend/access/gist/gistget.c             |   4 +-
 src/backend/access/gist/gistscan.c            |  12 +-
 src/backend/access/gist/gistsplit.c           |  12 +-
 src/backend/access/gist/gistutil.c            |  44 ++++-
 src/include/access/gist_private.h             |   2 +
 src/test/regress/expected/amutils.out         |   4 +-
 src/test/regress/expected/index_including.out |  10 +-
 .../regress/expected/index_including_gist.out | 166 ++++++++++++++++++
 src/test/regress/parallel_schedule            |   2 +-
 src/test/regress/serial_schedule              |   1 +
 src/test/regress/sql/index_including.sql      |   5 +-
 src/test/regress/sql/index_including_gist.sql |  90 ++++++++++
 15 files changed, 357 insertions(+), 31 deletions(-)
 create mode 100644 src/test/regress/expected/index_including_gist.out
 create mode 100644 src/test/regress/sql/index_including_gist.sql

diff --git a/doc/src/sgml/ref/create_index.sgml b/doc/src/sgml/ref/create_index.sgml
index ad619cdcfe..f8657c6ae2 100644
--- a/doc/src/sgml/ref/create_index.sgml
+++ b/doc/src/sgml/ref/create_index.sgml
@@ -181,8 +181,8 @@ CREATE [ UNIQUE ] INDEX [ CONCURRENTLY ] [ [ IF NOT EXISTS ] <replaceable class=
        </para>
 
        <para>
-        Currently, only the B-tree index access method supports this feature.
-        In B-tree indexes, the values of columns listed in the
+        Currently, the B-tree and the GiST index access methods supports this
+        feature. In B-tree indexes, the values of columns listed in the
         <literal>INCLUDE</literal> clause are included in leaf tuples which
         correspond to heap tuples, but are not included in upper-level
         index entries used for tree navigation.
diff --git a/doc/src/sgml/textsearch.sgml b/doc/src/sgml/textsearch.sgml
index ecebade767..e4c75ebf6f 100644
--- a/doc/src/sgml/textsearch.sgml
+++ b/doc/src/sgml/textsearch.sgml
@@ -3674,6 +3674,12 @@ SELECT plainto_tsquery('supernovae stars');
    retrieved to see if the match is correct.
   </para>
 
+  <para>
+   A GiST index can be covering, i.e. use the <literal>INCLUDE</literal>
+   clause. Included columns can have data types without any GiST operator
+   class. Included attributes will be stored uncompressed.
+  </para>
+
   <para>
    Lossiness causes performance degradation due to unnecessary fetches of table
    records that turn out to be false matches.  Since random access to table
diff --git a/src/backend/access/gist/gist.c b/src/backend/access/gist/gist.c
index b75b3a8dac..99647bf709 100644
--- a/src/backend/access/gist/gist.c
+++ b/src/backend/access/gist/gist.c
@@ -75,7 +75,7 @@ gisthandler(PG_FUNCTION_ARGS)
 	amroutine->amclusterable = true;
 	amroutine->ampredlocks = true;
 	amroutine->amcanparallel = false;
-	amroutine->amcaninclude = false;
+	amroutine->amcaninclude = true;
 	amroutine->amkeytype = InvalidOid;
 
 	amroutine->ambuild = gistbuild;
@@ -170,7 +170,7 @@ gistinsert(Relation r, Datum *values, bool *isnull,
 	oldCxt = MemoryContextSwitchTo(giststate->tempCxt);
 
 	itup = gistFormTuple(giststate, r,
-						 values, isnull, true /* size is currently bogus */ );
+						 values, isnull, true /* size is currently bogus */);
 	itup->t_tid = *ht_ctid;
 
 	gistdoinsert(r, itup, 0, giststate, heapRel);
@@ -1382,8 +1382,8 @@ gistSplit(Relation r,
 						IndexTupleSize(itup[0]), GiSTPageSize,
 						RelationGetRelationName(r))));
 
-	memset(v.spl_lisnull, true, sizeof(bool) * giststate->tupdesc->natts);
-	memset(v.spl_risnull, true, sizeof(bool) * giststate->tupdesc->natts);
+	memset(v.spl_lisnull, true, sizeof(bool) * giststate->truncTupdesc->natts);
+	memset(v.spl_risnull, true, sizeof(bool) * giststate->truncTupdesc->natts);
 	gistSplitByKey(r, page, itup, len, giststate, &v, 0);
 
 	/* form left and right vector */
@@ -1462,8 +1462,10 @@ initGISTstate(Relation index)
 	giststate->scanCxt = scanCxt;
 	giststate->tempCxt = scanCxt;	/* caller must change this if needed */
 	giststate->tupdesc = index->rd_att;
+	giststate->truncTupdesc = CreateTupleDescCopyConstr(index->rd_att);
+	giststate->truncTupdesc->natts = IndexRelationGetNumberOfKeyAttributes(index);
 
-	for (i = 0; i < index->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(index); i++)
 	{
 		fmgr_info_copy(&(giststate->consistentFn[i]),
 					   index_getprocinfo(index, i + 1, GIST_CONSISTENT_PROC),
@@ -1531,6 +1533,20 @@ initGISTstate(Relation index)
 			giststate->supportCollation[i] = DEFAULT_COLLATION_OID;
 	}
 
+	for (; i < index->rd_att->natts; i++)
+	{
+		giststate->consistentFn[i].fn_oid = InvalidOid;
+		giststate->unionFn[i].fn_oid = InvalidOid;
+		giststate->compressFn[i].fn_oid = InvalidOid;
+		giststate->decompressFn[i].fn_oid = InvalidOid;
+		giststate->penaltyFn[i].fn_oid = InvalidOid;
+		giststate->picksplitFn[i].fn_oid = InvalidOid;
+		giststate->equalFn[i].fn_oid = InvalidOid;
+		giststate->distanceFn[i].fn_oid = InvalidOid;
+		giststate->fetchFn[i].fn_oid = InvalidOid;
+		giststate->supportCollation[i] = InvalidOid;
+	}
+
 	MemoryContextSwitchTo(oldCxt);
 
 	return giststate;
diff --git a/src/backend/access/gist/gistget.c b/src/backend/access/gist/gistget.c
index a96ef5c3ac..a4ee8763f6 100644
--- a/src/backend/access/gist/gistget.c
+++ b/src/backend/access/gist/gistget.c
@@ -769,11 +769,13 @@ gistgetbitmap(IndexScanDesc scan, TIDBitmap *tbm)
  *
  * Opclasses that implement a fetch function support index-only scans.
  * Opclasses without compression functions also support index-only scans.
+ * Included attributes always can be fetched for index-only scans.
  */
 bool
 gistcanreturn(Relation index, int attno)
 {
-	if (OidIsValid(index_getprocid(index, attno, GIST_FETCH_PROC)) ||
+	if (attno > IndexRelationGetNumberOfKeyAttributes(index) ||
+		OidIsValid(index_getprocid(index, attno, GIST_FETCH_PROC)) ||
 		!OidIsValid(index_getprocid(index, attno, GIST_COMPRESS_PROC)))
 		return true;
 	else
diff --git a/src/backend/access/gist/gistscan.c b/src/backend/access/gist/gistscan.c
index 78a8ede794..4fffc8b33a 100644
--- a/src/backend/access/gist/gistscan.c
+++ b/src/backend/access/gist/gistscan.c
@@ -158,6 +158,7 @@ gistrescan(IndexScanDesc scan, ScanKey key, int nkeys,
 	if (scan->xs_want_itup && !scan->xs_hitupdesc)
 	{
 		int			natts;
+		int			nkeyatts;
 		int			attno;
 
 		/*
@@ -167,13 +168,22 @@ gistrescan(IndexScanDesc scan, ScanKey key, int nkeys,
 		 * types.
 		 */
 		natts = RelationGetNumberOfAttributes(scan->indexRelation);
+		nkeyatts = IndexRelationGetNumberOfKeyAttributes(scan->indexRelation);
 		so->giststate->fetchTupdesc = CreateTemplateTupleDesc(natts);
-		for (attno = 1; attno <= natts; attno++)
+		for (attno = 1; attno <= nkeyatts; attno++)
 		{
 			TupleDescInitEntry(so->giststate->fetchTupdesc, attno, NULL,
 							   scan->indexRelation->rd_opcintype[attno - 1],
 							   -1, 0);
 		}
+
+		for (; attno <= natts; attno++)
+		{
+			/* taking opcintype from giststate->tupdesc */
+			TupleDescInitEntry(so->giststate->fetchTupdesc, attno, NULL,
+					TupleDescAttr(so->giststate->tupdesc, attno - 1)->atttypid,
+							   -1, 0);
+		}
 		scan->xs_hitupdesc = so->giststate->fetchTupdesc;
 
 		/* Also create a memory context that will hold the returned tuples */
diff --git a/src/backend/access/gist/gistsplit.c b/src/backend/access/gist/gistsplit.c
index f210e0c39f..cf3af2821a 100644
--- a/src/backend/access/gist/gistsplit.c
+++ b/src/backend/access/gist/gistsplit.c
@@ -207,7 +207,7 @@ placeOne(Relation r, GISTSTATE *giststate, GistSplitVector *v,
 	gistDeCompressAtt(giststate, r, itup, NULL, (OffsetNumber) 0,
 					  identry, isnull);
 
-	for (; attno < giststate->tupdesc->natts; attno++)
+	for (; attno < giststate->truncTupdesc->natts; attno++)
 	{
 		float		lpenalty,
 					rpenalty;
@@ -485,7 +485,7 @@ gistUserPicksplit(Relation r, GistEntryVector *entryvec, int attno, GistSplitVec
 	 */
 	v->spl_dontcare = NULL;
 
-	if (attno + 1 < giststate->tupdesc->natts)
+	if (attno + 1 < giststate->truncTupdesc->natts)
 	{
 		int			NumDontCare;
 
@@ -657,7 +657,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 		 */
 		v->spl_risnull[attno] = v->spl_lisnull[attno] = true;
 
-		if (attno + 1 < giststate->tupdesc->natts)
+		if (attno + 1 < giststate->truncTupdesc->natts)
 			gistSplitByKey(r, page, itup, len, giststate, v, attno + 1);
 		else
 			gistSplitHalf(&v->splitVector, len);
@@ -683,7 +683,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 				v->splitVector.spl_left[v->splitVector.spl_nleft++] = i;
 
 		/* Compute union keys, unless outer recursion level will handle it */
-		if (attno == 0 && giststate->tupdesc->natts == 1)
+		if (attno == 0 && giststate->truncTupdesc->natts == 1)
 		{
 			v->spl_dontcare = NULL;
 			gistunionsubkey(giststate, itup, v);
@@ -700,7 +700,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 			 * Splitting on attno column is not optimal, so consider
 			 * redistributing don't-care tuples according to the next column
 			 */
-			Assert(attno + 1 < giststate->tupdesc->natts);
+			Assert(attno + 1 < giststate->truncTupdesc->natts);
 
 			if (v->spl_dontcare == NULL)
 			{
@@ -771,7 +771,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 	 * that PickSplit (or the special cases above) produced correct union
 	 * datums.
 	 */
-	if (attno == 0 && giststate->tupdesc->natts > 1)
+	if (attno == 0 && giststate->truncTupdesc->natts > 1)
 	{
 		v->spl_dontcare = NULL;
 		gistunionsubkey(giststate, itup, v);
diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c
index 8d3dfad27b..79b1adc656 100644
--- a/src/backend/access/gist/gistutil.c
+++ b/src/backend/access/gist/gistutil.c
@@ -160,7 +160,7 @@ gistMakeUnionItVec(GISTSTATE *giststate, IndexTuple *itvec, int len,
 
 	evec = (GistEntryVector *) palloc((len + 2) * sizeof(GISTENTRY) + GEVHDRSZ);
 
-	for (i = 0; i < giststate->tupdesc->natts; i++)
+	for (i = 0; i < giststate->truncTupdesc->natts; i++)
 	{
 		int			j;
 
@@ -296,7 +296,7 @@ gistDeCompressAtt(GISTSTATE *giststate, Relation r, IndexTuple tuple, Page p,
 {
 	int			i;
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		Datum		datum;
 
@@ -329,7 +329,7 @@ gistgetadjusted(Relation r, IndexTuple oldtup, IndexTuple addtup, GISTSTATE *gis
 	gistDeCompressAtt(giststate, r, addtup, NULL,
 					  (OffsetNumber) 0, addentries, addisnull);
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		gistMakeUnionKey(giststate, i,
 						 oldentries + i, oldisnull[i],
@@ -442,7 +442,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 		zero_penalty = true;
 
 		/* Loop over index attributes. */
-		for (j = 0; j < r->rd_att->natts; j++)
+		for (j = 0; j < IndexRelationGetNumberOfKeyAttributes(r); j++)
 		{
 			Datum		datum;
 			float		usize;
@@ -470,7 +470,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 				result = i;
 				best_penalty[j] = usize;
 
-				if (j < r->rd_att->natts - 1)
+				if (j < IndexRelationGetNumberOfKeyAttributes(r) - 1)
 					best_penalty[j + 1] = -1;
 
 				/* we have new best, so reset keep-it decision */
@@ -500,7 +500,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 		 * If we looped past the last column, and did not update "result",
 		 * then this tuple is exactly as good as the prior best tuple.
 		 */
-		if (j == r->rd_att->natts && result != i)
+		if (j == IndexRelationGetNumberOfKeyAttributes(r) && result != i)
 		{
 			if (keep_current_best == -1)
 			{
@@ -575,11 +575,13 @@ gistFormTuple(GISTSTATE *giststate, Relation r,
 	Datum		compatt[INDEX_MAX_KEYS];
 	int			i;
 	IndexTuple	res;
+	/* currently we truncate tuples iif they are on internal pages */
+	bool isTruncated = !isleaf;
 
 	/*
 	 * Call the compress method on each attribute.
 	 */
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		if (isnull[i])
 			compatt[i] = (Datum) 0;
@@ -602,7 +604,23 @@ gistFormTuple(GISTSTATE *giststate, Relation r,
 		}
 	}
 
-	res = index_form_tuple(giststate->tupdesc, compatt, isnull);
+	if (!isTruncated)
+	{
+		/*
+		 * Emplace each included attribute.
+		 */
+		for (; i < r->rd_att->natts; i++)
+		{
+			if (isnull[i])
+				compatt[i] = (Datum) 0;
+			else
+			{
+				compatt[i] = attdata[i];
+			}
+		}
+	}
+
+	res = index_form_tuple(isTruncated ? giststate->truncTupdesc : giststate->tupdesc, compatt, isnull);
 
 	/*
 	 * The offset number on tuples on internal pages is unused. For historical
@@ -644,7 +662,7 @@ gistFetchTuple(GISTSTATE *giststate, Relation r, IndexTuple tuple)
 	bool		isnull[INDEX_MAX_KEYS];
 	int			i;
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		Datum		datum;
 
@@ -679,6 +697,14 @@ gistFetchTuple(GISTSTATE *giststate, Relation r, IndexTuple tuple)
 			fetchatt[i] = (Datum) 0;
 		}
 	}
+
+	/*
+	 * Get each included attribute.
+	 */
+	for (; i < r->rd_att->natts; i++)
+	{
+		fetchatt[i] = index_getattr(tuple, i + 1, giststate->tupdesc, &isnull[i]);
+	}
 	MemoryContextSwitchTo(oldcxt);
 
 	return heap_form_tuple(giststate->fetchTupdesc, fetchatt, isnull);
diff --git a/src/include/access/gist_private.h b/src/include/access/gist_private.h
index 3698942f9d..ccf06e676c 100644
--- a/src/include/access/gist_private.h
+++ b/src/include/access/gist_private.h
@@ -79,6 +79,8 @@ typedef struct GISTSTATE
 	MemoryContext tempCxt;		/* short-term context for calling functions */
 
 	TupleDesc	tupdesc;		/* index's tuple descriptor */
+	TupleDesc	truncTupdesc;	/* truncated tuple descriptor
+								 * for internal pages */
 	TupleDesc	fetchTupdesc;	/* tuple descriptor for tuples returned in an
 								 * index-only scan */
 
diff --git a/src/test/regress/expected/amutils.out b/src/test/regress/expected/amutils.out
index 4570a39b05..d92a6d12c6 100644
--- a/src/test/regress/expected/amutils.out
+++ b/src/test/regress/expected/amutils.out
@@ -75,7 +75,7 @@ select prop,
  can_unique         | f  |       | 
  can_multi_col      | t  |       | 
  can_exclude        | t  |       | 
- can_include        | f  |       | 
+ can_include        | t  |       | 
  bogus              |    |       | 
 (19 rows)
 
@@ -159,7 +159,7 @@ select amname, prop, pg_indexam_has_property(a.oid, prop) as p
  gist   | can_unique    | f
  gist   | can_multi_col | t
  gist   | can_exclude   | t
- gist   | can_include   | f
+ gist   | can_include   | t
  gist   | bogus         | 
  hash   | can_order     | f
  hash   | can_unique    | f
diff --git a/src/test/regress/expected/index_including.out b/src/test/regress/expected/index_including.out
index 16b4be34de..0f1fb88843 100644
--- a/src/test/regress/expected/index_including.out
+++ b/src/test/regress/expected/index_including.out
@@ -313,13 +313,14 @@ SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl' ORDER BY indexname;
 
 DROP TABLE tbl;
 /*
- * 7. Check various AMs. All but btree must fail.
+ * 7. Check various AMs. All but btree and gist must fail.
  */
 CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
 CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "brin" does not support included columns
 CREATE INDEX on tbl USING gist(c3) INCLUDE (c4);
-ERROR:  access method "gist" does not support included columns
+CREATE INDEX on tbl USING gist(c3) INCLUDE (c1);
+CREATE INDEX on tbl USING gist(c3) INCLUDE (c1, c4);
 CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
 ERROR:  access method "spgist" does not support included columns
 CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
@@ -328,7 +329,10 @@ CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "hash" does not support included columns
 CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
 NOTICE:  substituting access method "gist" for obsolete method "rtree"
-ERROR:  access method "gist" does not support included columns
+ERROR:  data type integer has no default operator class for access method "gist"
+HINT:  You must specify an operator class for the index or define a default operator class for the data type.
+CREATE INDEX on tbl USING rtree(c4) INCLUDE (c1, c4);
+NOTICE:  substituting access method "gist" for obsolete method "rtree"
 CREATE INDEX on tbl USING btree(c1, c2) INCLUDE (c3, c4);
 DROP TABLE tbl;
 /*
diff --git a/src/test/regress/expected/index_including_gist.out b/src/test/regress/expected/index_including_gist.out
new file mode 100644
index 0000000000..c6d567798b
--- /dev/null
+++ b/src/test/regress/expected/index_including_gist.out
@@ -0,0 +1,166 @@
+/*
+ * 1.1. test CREATE INDEX with buffered build
+ */
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+                                  pg_get_indexdef                                  
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+ c1 | c2 | c3 |     c4      
+----+----+----+-------------
+  1 |  2 |  3 | (2,3),(1,2)
+  2 |  4 |  6 | (4,5),(2,3)
+  3 |  6 |  9 | (6,7),(3,4)
+  4 |  8 | 12 | (8,9),(4,5)
+(4 rows)
+
+SET enable_bitmapscan TO off;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                   QUERY PLAN                   
+------------------------------------------------
+ Index Only Scan using tbl_gist_idx on tbl_gist
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+/*
+ * 1.2. test CREATE INDEX with inserts
+ */
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+                                  pg_get_indexdef                                  
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+ c1 | c2 | c3 |     c4      
+----+----+----+-------------
+  1 |  2 |  3 | (2,3),(1,2)
+  2 |  4 |  6 | (4,5),(2,3)
+  3 |  6 |  9 | (6,7),(3,4)
+  4 |  8 | 12 | (8,9),(4,5)
+(4 rows)
+
+SET enable_bitmapscan TO off;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                   QUERY PLAN                   
+------------------------------------------------
+ Index Only Scan using tbl_gist_idx on tbl_gist
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+/*
+ * 2. CREATE INDEX CONCURRENTLY
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX CONCURRENTLY tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                     indexdef                                      
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+DROP TABLE tbl_gist;
+/*
+ * 3. REINDEX
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                   indexdef                                    
+-------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c3)
+(1 row)
+
+REINDEX INDEX tbl_gist_idx;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                   indexdef                                    
+-------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c3)
+(1 row)
+
+ALTER TABLE tbl_gist DROP COLUMN c1;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+ indexdef 
+----------
+(0 rows)
+
+DROP TABLE tbl_gist;
+/*
+ * 4. Update, delete values in indexed table.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+UPDATE tbl_gist SET c1 = 100 WHERE c1 = 2;
+UPDATE tbl_gist SET c1 = 1 WHERE c1 = 3;
+DELETE FROM tbl_gist WHERE c1 = 5 OR c3 = 12;
+DROP TABLE tbl_gist;
+/*
+ * 5. Alter column type.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+ALTER TABLE tbl_gist ALTER c1 TYPE bigint;
+ALTER TABLE tbl_gist ALTER c3 TYPE bigint;
+\d tbl_gist
+              Table "public.tbl_gist"
+ Column |  Type   | Collation | Nullable | Default 
+--------+---------+-----------+----------+---------
+ c1     | bigint  |           |          | 
+ c2     | integer |           |          | 
+ c3     | bigint  |           |          | 
+ c4     | box     |           |          | 
+Indexes:
+    "tbl_gist_idx" gist (c4) INCLUDE (c1, c3)
+
+DROP TABLE tbl_gist;
+/*
+ * 5. EXCLUDE constraint.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box, EXCLUDE USING gist (c4 WITH &&) INCLUDE (c1, c2, c3));
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+ERROR:  conflicting key value violates exclusion constraint "tbl_gist_c4_c1_c2_c3_excl"
+DETAIL:  Key (c4)=((4,5),(2,3)) conflicts with existing key (c4)=((2,3),(1,2)).
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(3*x,2*x),point(3*x+1,2*x+1)) FROM generate_series(1,10) AS x;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                         QUERY PLAN                          
+-------------------------------------------------------------
+ Index Only Scan using tbl_gist_c4_c1_c2_c3_excl on tbl_gist
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+\d tbl_gist
+              Table "public.tbl_gist"
+ Column |  Type   | Collation | Nullable | Default 
+--------+---------+-----------+----------+---------
+ c1     | integer |           |          | 
+ c2     | integer |           |          | 
+ c3     | integer |           |          | 
+ c4     | box     |           |          | 
+Indexes:
+    "tbl_gist_c4_c1_c2_c3_excl" EXCLUDE USING gist (c4 WITH &&) INCLUDE (c1, c2, c3)
+
+DROP TABLE tbl_gist;
diff --git a/src/test/regress/parallel_schedule b/src/test/regress/parallel_schedule
index cc0bbf5db9..5684291ab0 100644
--- a/src/test/regress/parallel_schedule
+++ b/src/test/regress/parallel_schedule
@@ -55,7 +55,7 @@ test: copy copyselect copydml
 # ----------
 test: create_misc create_operator create_procedure
 # These depend on the above two
-test: create_index create_view index_including
+test: create_index create_view index_including index_including_gist
 
 # ----------
 # Another group of parallel tests
diff --git a/src/test/regress/serial_schedule b/src/test/regress/serial_schedule
index 0c10c7100c..309ccec81a 100644
--- a/src/test/regress/serial_schedule
+++ b/src/test/regress/serial_schedule
@@ -63,6 +63,7 @@ test: create_operator
 test: create_procedure
 test: create_index
 test: index_including
+test: index_including_gist
 test: create_view
 test: create_aggregate
 test: create_function_3
diff --git a/src/test/regress/sql/index_including.sql b/src/test/regress/sql/index_including.sql
index ef5fd882f5..4fd015d065 100644
--- a/src/test/regress/sql/index_including.sql
+++ b/src/test/regress/sql/index_including.sql
@@ -173,15 +173,18 @@ SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl' ORDER BY indexname;
 DROP TABLE tbl;
 
 /*
- * 7. Check various AMs. All but btree must fail.
+ * 7. Check various AMs. All but btree and gist must fail.
  */
 CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
 CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
 CREATE INDEX on tbl USING gist(c3) INCLUDE (c4);
+CREATE INDEX on tbl USING gist(c3) INCLUDE (c1);
+CREATE INDEX on tbl USING gist(c3) INCLUDE (c1, c4);
 CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
 CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
 CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
 CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
+CREATE INDEX on tbl USING rtree(c4) INCLUDE (c1, c4);
 CREATE INDEX on tbl USING btree(c1, c2) INCLUDE (c3, c4);
 DROP TABLE tbl;
 
diff --git a/src/test/regress/sql/index_including_gist.sql b/src/test/regress/sql/index_including_gist.sql
new file mode 100644
index 0000000000..c517d51d76
--- /dev/null
+++ b/src/test/regress/sql/index_including_gist.sql
@@ -0,0 +1,90 @@
+/*
+ * 1.1. test CREATE INDEX with buffered build
+ */
+
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO off;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+
+/*
+ * 1.2. test CREATE INDEX with inserts
+ */
+
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO off;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+
+/*
+ * 2. CREATE INDEX CONCURRENTLY
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX CONCURRENTLY tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+DROP TABLE tbl_gist;
+
+
+/*
+ * 3. REINDEX
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+REINDEX INDEX tbl_gist_idx;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+ALTER TABLE tbl_gist DROP COLUMN c1;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+DROP TABLE tbl_gist;
+
+/*
+ * 4. Update, delete values in indexed table.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+UPDATE tbl_gist SET c1 = 100 WHERE c1 = 2;
+UPDATE tbl_gist SET c1 = 1 WHERE c1 = 3;
+DELETE FROM tbl_gist WHERE c1 = 5 OR c3 = 12;
+DROP TABLE tbl_gist;
+
+/*
+ * 5. Alter column type.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+ALTER TABLE tbl_gist ALTER c1 TYPE bigint;
+ALTER TABLE tbl_gist ALTER c3 TYPE bigint;
+\d tbl_gist
+DROP TABLE tbl_gist;
+
+/*
+ * 5. EXCLUDE constraint.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box, EXCLUDE USING gist (c4 WITH &&) INCLUDE (c1, c2, c3));
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(3*x,2*x),point(3*x+1,2*x+1)) FROM generate_series(1,10) AS x;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+\d tbl_gist
+DROP TABLE tbl_gist;
-- 
2.20.1

#20

andreas@proxel.se

almost 7 years ago

In reply to: Andrey Borodin (#19)

Re: Covering GiST indexes

Thanks for the new version of the patch. Based on my knowledge of PG
this is starting to look good, and I have only three small comments below.

I am not 100% a fan of truncTupdesc, but as long as it is well commented
I think that it is fine.

= Review

* I think it is worth writing a short comment when you create
truncTupdesc about why this is done.

* Very minor thing: the diff below is pointless churn on a line not
touched by the patch.

-                        values, isnull, true /* size is currently bogus 
*/ );
+                        values, isnull, true /* size is currently bogus 
*/);

* Another very minor thing: The diff below from gistFormTuple() should
probably be consistent about brackets.

+           if (isnull[i])
+               compatt[i] = (Datum) 0;
+           else
+           {
+               compatt[i] = attdata[i];
+           }

Andreas

#21

andreas@proxel.se

almost 7 years ago

In reply to: Andreas Karlsson (#20)

Re: Covering GiST indexes

On 29/01/2019 18.00, Andreas Karlsson wrote:

Thanks for the new version of the patch. Based on my knowledge of PG
this is starting to look good, and I have only three small comments below.

Sorry, I missed retree in the tests. Can you fix the below to match the
gist example?

  CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
  NOTICE:  substituting access method "gist" for obsolete method "rtree"
-ERROR:  access method "gist" does not support included columns
+ERROR:  data type integer has no default operator class for access 
method "gist"
+HINT:  You must specify an operator class for the index or define a 
default operator class for the data type.
+CREATE INDEX on tbl USING rtree(c4) INCLUDE (c1, c4);

Andreas

#22

x4mmm@yandex-team.ru

almost 7 years ago

In reply to: Andreas Karlsson (#20)

1 attachment(s)

Re: Covering GiST indexes

Thanks! Here's new version 8

29 янв. 2019 г., в 22:00, Andreas Karlsson <andreas@proxel.se> написал(а):

* I think it is worth writing a short comment when you create truncTupdesc about why this is done.

It resulted in not so short comment, but I think it is OK.

* Very minor thing: the diff below is pointless churn on a line not touched by the patch.
-                        values, isnull, true /* size is currently bogus */ );
+                        values, isnull, true /* size is currently bogus */);

Fixed.

* Another very minor thing: The diff below from gistFormTuple() should probably be consistent about brackets.
+           if (isnull[i])
+               compatt[i] = (Datum) 0;
+           else
+           {
+               compatt[i] = attdata[i];
+           }

Fixed.

Also, I've unified gist and r-tree syntax tests for INCLUDE.

Best regards, Andrey Borodin.

Attachments:

0001-Covering-GiST-v8.patchapplication/octet-stream; name=0001-Covering-GiST-v8.patch; x-unix-mode=0644Download

From 019e536ea566f733087eb5037a41b8bbcf8aa40d Mon Sep 17 00:00:00 2001
From: Andrey <amborodin@acm.org>
Date: Mon, 26 Nov 2018 21:55:45 +0500
Subject: [PATCH] Covering GiST v8

This patch allow adding INCLUDE colums to GiST index.
These colums do not participate in GiST tree build,
are not used for chosing subtree for inserts or splits.
But they can be fetched during IndexOnlyScan.
---
 doc/src/sgml/ref/create_index.sgml            |   4 +-
 doc/src/sgml/textsearch.sgml                  |   6 +
 src/backend/access/gist/gist.c                |  33 +++-
 src/backend/access/gist/gistget.c             |   4 +-
 src/backend/access/gist/gistscan.c            |  12 +-
 src/backend/access/gist/gistsplit.c           |  12 +-
 src/backend/access/gist/gistutil.c            |  42 ++++-
 src/include/access/gist_private.h             |   2 +
 src/test/regress/expected/amutils.out         |   4 +-
 src/test/regress/expected/index_including.out |  17 +-
 .../regress/expected/index_including_gist.out | 166 ++++++++++++++++++
 src/test/regress/parallel_schedule            |   2 +-
 src/test/regress/serial_schedule              |   1 +
 src/test/regress/sql/index_including.sql      |   8 +-
 src/test/regress/sql/index_including_gist.sql |  90 ++++++++++
 15 files changed, 373 insertions(+), 30 deletions(-)
 create mode 100644 src/test/regress/expected/index_including_gist.out
 create mode 100644 src/test/regress/sql/index_including_gist.sql

diff --git a/doc/src/sgml/ref/create_index.sgml b/doc/src/sgml/ref/create_index.sgml
index ad619cdcfe..f8657c6ae2 100644
--- a/doc/src/sgml/ref/create_index.sgml
+++ b/doc/src/sgml/ref/create_index.sgml
@@ -181,8 +181,8 @@ CREATE [ UNIQUE ] INDEX [ CONCURRENTLY ] [ [ IF NOT EXISTS ] <replaceable class=
        </para>
 
        <para>
-        Currently, only the B-tree index access method supports this feature.
-        In B-tree indexes, the values of columns listed in the
+        Currently, the B-tree and the GiST index access methods supports this
+        feature. In B-tree indexes, the values of columns listed in the
         <literal>INCLUDE</literal> clause are included in leaf tuples which
         correspond to heap tuples, but are not included in upper-level
         index entries used for tree navigation.
diff --git a/doc/src/sgml/textsearch.sgml b/doc/src/sgml/textsearch.sgml
index ecebade767..e4c75ebf6f 100644
--- a/doc/src/sgml/textsearch.sgml
+++ b/doc/src/sgml/textsearch.sgml
@@ -3674,6 +3674,12 @@ SELECT plainto_tsquery('supernovae stars');
    retrieved to see if the match is correct.
   </para>
 
+  <para>
+   A GiST index can be covering, i.e. use the <literal>INCLUDE</literal>
+   clause. Included columns can have data types without any GiST operator
+   class. Included attributes will be stored uncompressed.
+  </para>
+
   <para>
    Lossiness causes performance degradation due to unnecessary fetches of table
    records that turn out to be false matches.  Since random access to table
diff --git a/src/backend/access/gist/gist.c b/src/backend/access/gist/gist.c
index b75b3a8dac..d6ee44dc9c 100644
--- a/src/backend/access/gist/gist.c
+++ b/src/backend/access/gist/gist.c
@@ -75,7 +75,7 @@ gisthandler(PG_FUNCTION_ARGS)
 	amroutine->amclusterable = true;
 	amroutine->ampredlocks = true;
 	amroutine->amcanparallel = false;
-	amroutine->amcaninclude = false;
+	amroutine->amcaninclude = true;
 	amroutine->amkeytype = InvalidOid;
 
 	amroutine->ambuild = gistbuild;
@@ -1382,8 +1382,8 @@ gistSplit(Relation r,
 						IndexTupleSize(itup[0]), GiSTPageSize,
 						RelationGetRelationName(r))));
 
-	memset(v.spl_lisnull, true, sizeof(bool) * giststate->tupdesc->natts);
-	memset(v.spl_risnull, true, sizeof(bool) * giststate->tupdesc->natts);
+	memset(v.spl_lisnull, true, sizeof(bool) * giststate->truncTupdesc->natts);
+	memset(v.spl_risnull, true, sizeof(bool) * giststate->truncTupdesc->natts);
 	gistSplitByKey(r, page, itup, len, giststate, &v, 0);
 
 	/* form left and right vector */
@@ -1462,8 +1462,19 @@ initGISTstate(Relation index)
 	giststate->scanCxt = scanCxt;
 	giststate->tempCxt = scanCxt;	/* caller must change this if needed */
 	giststate->tupdesc = index->rd_att;
+	/*
+	 * truncTupdesc for truncated tuples may contain less attributes, because
+	 * it does not store INCLUDE attributes.
+	 * It is used to form tuples during tuple adjustement and page split.
+	 * B-tree creates shortened tuple descriptor for every truncated tuple,
+	 * because it is doing this less often: it does not have to form
+	 * truncated tuples during page split. Also, B-tree is not ajdusting
+	 * tuples on internal pages the way GiST does.
+	 */
+	giststate->truncTupdesc = CreateTupleDescCopyConstr(index->rd_att);
+	giststate->truncTupdesc->natts = IndexRelationGetNumberOfKeyAttributes(index);
 
-	for (i = 0; i < index->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(index); i++)
 	{
 		fmgr_info_copy(&(giststate->consistentFn[i]),
 					   index_getprocinfo(index, i + 1, GIST_CONSISTENT_PROC),
@@ -1531,6 +1542,20 @@ initGISTstate(Relation index)
 			giststate->supportCollation[i] = DEFAULT_COLLATION_OID;
 	}
 
+	for (; i < index->rd_att->natts; i++)
+	{
+		giststate->consistentFn[i].fn_oid = InvalidOid;
+		giststate->unionFn[i].fn_oid = InvalidOid;
+		giststate->compressFn[i].fn_oid = InvalidOid;
+		giststate->decompressFn[i].fn_oid = InvalidOid;
+		giststate->penaltyFn[i].fn_oid = InvalidOid;
+		giststate->picksplitFn[i].fn_oid = InvalidOid;
+		giststate->equalFn[i].fn_oid = InvalidOid;
+		giststate->distanceFn[i].fn_oid = InvalidOid;
+		giststate->fetchFn[i].fn_oid = InvalidOid;
+		giststate->supportCollation[i] = InvalidOid;
+	}
+
 	MemoryContextSwitchTo(oldCxt);
 
 	return giststate;
diff --git a/src/backend/access/gist/gistget.c b/src/backend/access/gist/gistget.c
index a96ef5c3ac..a4ee8763f6 100644
--- a/src/backend/access/gist/gistget.c
+++ b/src/backend/access/gist/gistget.c
@@ -769,11 +769,13 @@ gistgetbitmap(IndexScanDesc scan, TIDBitmap *tbm)
  *
  * Opclasses that implement a fetch function support index-only scans.
  * Opclasses without compression functions also support index-only scans.
+ * Included attributes always can be fetched for index-only scans.
  */
 bool
 gistcanreturn(Relation index, int attno)
 {
-	if (OidIsValid(index_getprocid(index, attno, GIST_FETCH_PROC)) ||
+	if (attno > IndexRelationGetNumberOfKeyAttributes(index) ||
+		OidIsValid(index_getprocid(index, attno, GIST_FETCH_PROC)) ||
 		!OidIsValid(index_getprocid(index, attno, GIST_COMPRESS_PROC)))
 		return true;
 	else
diff --git a/src/backend/access/gist/gistscan.c b/src/backend/access/gist/gistscan.c
index 78a8ede794..4fffc8b33a 100644
--- a/src/backend/access/gist/gistscan.c
+++ b/src/backend/access/gist/gistscan.c
@@ -158,6 +158,7 @@ gistrescan(IndexScanDesc scan, ScanKey key, int nkeys,
 	if (scan->xs_want_itup && !scan->xs_hitupdesc)
 	{
 		int			natts;
+		int			nkeyatts;
 		int			attno;
 
 		/*
@@ -167,13 +168,22 @@ gistrescan(IndexScanDesc scan, ScanKey key, int nkeys,
 		 * types.
 		 */
 		natts = RelationGetNumberOfAttributes(scan->indexRelation);
+		nkeyatts = IndexRelationGetNumberOfKeyAttributes(scan->indexRelation);
 		so->giststate->fetchTupdesc = CreateTemplateTupleDesc(natts);
-		for (attno = 1; attno <= natts; attno++)
+		for (attno = 1; attno <= nkeyatts; attno++)
 		{
 			TupleDescInitEntry(so->giststate->fetchTupdesc, attno, NULL,
 							   scan->indexRelation->rd_opcintype[attno - 1],
 							   -1, 0);
 		}
+
+		for (; attno <= natts; attno++)
+		{
+			/* taking opcintype from giststate->tupdesc */
+			TupleDescInitEntry(so->giststate->fetchTupdesc, attno, NULL,
+					TupleDescAttr(so->giststate->tupdesc, attno - 1)->atttypid,
+							   -1, 0);
+		}
 		scan->xs_hitupdesc = so->giststate->fetchTupdesc;
 
 		/* Also create a memory context that will hold the returned tuples */
diff --git a/src/backend/access/gist/gistsplit.c b/src/backend/access/gist/gistsplit.c
index f210e0c39f..cf3af2821a 100644
--- a/src/backend/access/gist/gistsplit.c
+++ b/src/backend/access/gist/gistsplit.c
@@ -207,7 +207,7 @@ placeOne(Relation r, GISTSTATE *giststate, GistSplitVector *v,
 	gistDeCompressAtt(giststate, r, itup, NULL, (OffsetNumber) 0,
 					  identry, isnull);
 
-	for (; attno < giststate->tupdesc->natts; attno++)
+	for (; attno < giststate->truncTupdesc->natts; attno++)
 	{
 		float		lpenalty,
 					rpenalty;
@@ -485,7 +485,7 @@ gistUserPicksplit(Relation r, GistEntryVector *entryvec, int attno, GistSplitVec
 	 */
 	v->spl_dontcare = NULL;
 
-	if (attno + 1 < giststate->tupdesc->natts)
+	if (attno + 1 < giststate->truncTupdesc->natts)
 	{
 		int			NumDontCare;
 
@@ -657,7 +657,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 		 */
 		v->spl_risnull[attno] = v->spl_lisnull[attno] = true;
 
-		if (attno + 1 < giststate->tupdesc->natts)
+		if (attno + 1 < giststate->truncTupdesc->natts)
 			gistSplitByKey(r, page, itup, len, giststate, v, attno + 1);
 		else
 			gistSplitHalf(&v->splitVector, len);
@@ -683,7 +683,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 				v->splitVector.spl_left[v->splitVector.spl_nleft++] = i;
 
 		/* Compute union keys, unless outer recursion level will handle it */
-		if (attno == 0 && giststate->tupdesc->natts == 1)
+		if (attno == 0 && giststate->truncTupdesc->natts == 1)
 		{
 			v->spl_dontcare = NULL;
 			gistunionsubkey(giststate, itup, v);
@@ -700,7 +700,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 			 * Splitting on attno column is not optimal, so consider
 			 * redistributing don't-care tuples according to the next column
 			 */
-			Assert(attno + 1 < giststate->tupdesc->natts);
+			Assert(attno + 1 < giststate->truncTupdesc->natts);
 
 			if (v->spl_dontcare == NULL)
 			{
@@ -771,7 +771,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 	 * that PickSplit (or the special cases above) produced correct union
 	 * datums.
 	 */
-	if (attno == 0 && giststate->tupdesc->natts > 1)
+	if (attno == 0 && giststate->truncTupdesc->natts > 1)
 	{
 		v->spl_dontcare = NULL;
 		gistunionsubkey(giststate, itup, v);
diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c
index 8d3dfad27b..f7024eda12 100644
--- a/src/backend/access/gist/gistutil.c
+++ b/src/backend/access/gist/gistutil.c
@@ -160,7 +160,7 @@ gistMakeUnionItVec(GISTSTATE *giststate, IndexTuple *itvec, int len,
 
 	evec = (GistEntryVector *) palloc((len + 2) * sizeof(GISTENTRY) + GEVHDRSZ);
 
-	for (i = 0; i < giststate->tupdesc->natts; i++)
+	for (i = 0; i < giststate->truncTupdesc->natts; i++)
 	{
 		int			j;
 
@@ -296,7 +296,7 @@ gistDeCompressAtt(GISTSTATE *giststate, Relation r, IndexTuple tuple, Page p,
 {
 	int			i;
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		Datum		datum;
 
@@ -329,7 +329,7 @@ gistgetadjusted(Relation r, IndexTuple oldtup, IndexTuple addtup, GISTSTATE *gis
 	gistDeCompressAtt(giststate, r, addtup, NULL,
 					  (OffsetNumber) 0, addentries, addisnull);
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		gistMakeUnionKey(giststate, i,
 						 oldentries + i, oldisnull[i],
@@ -442,7 +442,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 		zero_penalty = true;
 
 		/* Loop over index attributes. */
-		for (j = 0; j < r->rd_att->natts; j++)
+		for (j = 0; j < IndexRelationGetNumberOfKeyAttributes(r); j++)
 		{
 			Datum		datum;
 			float		usize;
@@ -470,7 +470,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 				result = i;
 				best_penalty[j] = usize;
 
-				if (j < r->rd_att->natts - 1)
+				if (j < IndexRelationGetNumberOfKeyAttributes(r) - 1)
 					best_penalty[j + 1] = -1;
 
 				/* we have new best, so reset keep-it decision */
@@ -500,7 +500,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 		 * If we looped past the last column, and did not update "result",
 		 * then this tuple is exactly as good as the prior best tuple.
 		 */
-		if (j == r->rd_att->natts && result != i)
+		if (j == IndexRelationGetNumberOfKeyAttributes(r) && result != i)
 		{
 			if (keep_current_best == -1)
 			{
@@ -575,11 +575,13 @@ gistFormTuple(GISTSTATE *giststate, Relation r,
 	Datum		compatt[INDEX_MAX_KEYS];
 	int			i;
 	IndexTuple	res;
+	/* currently we truncate tuples iif they are on internal pages */
+	bool isTruncated = !isleaf;
 
 	/*
 	 * Call the compress method on each attribute.
 	 */
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		if (isnull[i])
 			compatt[i] = (Datum) 0;
@@ -602,7 +604,21 @@ gistFormTuple(GISTSTATE *giststate, Relation r,
 		}
 	}
 
-	res = index_form_tuple(giststate->tupdesc, compatt, isnull);
+	if (!isTruncated)
+	{
+		/*
+		 * Emplace each included attribute.
+		 */
+		for (; i < r->rd_att->natts; i++)
+		{
+			if (isnull[i])
+				compatt[i] = (Datum) 0;
+			else
+				compatt[i] = attdata[i];
+		}
+	}
+
+	res = index_form_tuple(isTruncated ? giststate->truncTupdesc : giststate->tupdesc, compatt, isnull);
 
 	/*
 	 * The offset number on tuples on internal pages is unused. For historical
@@ -644,7 +660,7 @@ gistFetchTuple(GISTSTATE *giststate, Relation r, IndexTuple tuple)
 	bool		isnull[INDEX_MAX_KEYS];
 	int			i;
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		Datum		datum;
 
@@ -679,6 +695,14 @@ gistFetchTuple(GISTSTATE *giststate, Relation r, IndexTuple tuple)
 			fetchatt[i] = (Datum) 0;
 		}
 	}
+
+	/*
+	 * Get each included attribute.
+	 */
+	for (; i < r->rd_att->natts; i++)
+	{
+		fetchatt[i] = index_getattr(tuple, i + 1, giststate->tupdesc, &isnull[i]);
+	}
 	MemoryContextSwitchTo(oldcxt);
 
 	return heap_form_tuple(giststate->fetchTupdesc, fetchatt, isnull);
diff --git a/src/include/access/gist_private.h b/src/include/access/gist_private.h
index 3698942f9d..ccf06e676c 100644
--- a/src/include/access/gist_private.h
+++ b/src/include/access/gist_private.h
@@ -79,6 +79,8 @@ typedef struct GISTSTATE
 	MemoryContext tempCxt;		/* short-term context for calling functions */
 
 	TupleDesc	tupdesc;		/* index's tuple descriptor */
+	TupleDesc	truncTupdesc;	/* truncated tuple descriptor
+								 * for internal pages */
 	TupleDesc	fetchTupdesc;	/* tuple descriptor for tuples returned in an
 								 * index-only scan */
 
diff --git a/src/test/regress/expected/amutils.out b/src/test/regress/expected/amutils.out
index 4570a39b05..d92a6d12c6 100644
--- a/src/test/regress/expected/amutils.out
+++ b/src/test/regress/expected/amutils.out
@@ -75,7 +75,7 @@ select prop,
  can_unique         | f  |       | 
  can_multi_col      | t  |       | 
  can_exclude        | t  |       | 
- can_include        | f  |       | 
+ can_include        | t  |       | 
  bogus              |    |       | 
 (19 rows)
 
@@ -159,7 +159,7 @@ select amname, prop, pg_indexam_has_property(a.oid, prop) as p
  gist   | can_unique    | f
  gist   | can_multi_col | t
  gist   | can_exclude   | t
- gist   | can_include   | f
+ gist   | can_include   | t
  gist   | bogus         | 
  hash   | can_order     | f
  hash   | can_unique    | f
diff --git a/src/test/regress/expected/index_including.out b/src/test/regress/expected/index_including.out
index 16b4be34de..d424a85e69 100644
--- a/src/test/regress/expected/index_including.out
+++ b/src/test/regress/expected/index_including.out
@@ -313,13 +313,17 @@ SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl' ORDER BY indexname;
 
 DROP TABLE tbl;
 /*
- * 7. Check various AMs. All but btree must fail.
+ * 7. Check various AMs. All but btree and gist must fail.
  */
 CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
 CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "brin" does not support included columns
+CREATE INDEX on tbl USING gist(c1, c2) INCLUDE (c3, c4);
+ERROR:  data type integer has no default operator class for access method "gist"
+HINT:  You must specify an operator class for the index or define a default operator class for the data type.
 CREATE INDEX on tbl USING gist(c3) INCLUDE (c4);
-ERROR:  access method "gist" does not support included columns
+CREATE INDEX on tbl USING gist(c3) INCLUDE (c1);
+CREATE INDEX on tbl USING gist(c3) INCLUDE (c1, c4);
 CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
 ERROR:  access method "spgist" does not support included columns
 CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
@@ -328,7 +332,14 @@ CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "hash" does not support included columns
 CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
 NOTICE:  substituting access method "gist" for obsolete method "rtree"
-ERROR:  access method "gist" does not support included columns
+ERROR:  data type integer has no default operator class for access method "gist"
+HINT:  You must specify an operator class for the index or define a default operator class for the data type.
+CREATE INDEX on tbl USING rtree(c3) INCLUDE (c4);
+NOTICE:  substituting access method "gist" for obsolete method "rtree"
+CREATE INDEX on tbl USING rtree(c3) INCLUDE (c1);
+NOTICE:  substituting access method "gist" for obsolete method "rtree"
+CREATE INDEX on tbl USING rtree(c3) INCLUDE (c1, c4);
+NOTICE:  substituting access method "gist" for obsolete method "rtree"
 CREATE INDEX on tbl USING btree(c1, c2) INCLUDE (c3, c4);
 DROP TABLE tbl;
 /*
diff --git a/src/test/regress/expected/index_including_gist.out b/src/test/regress/expected/index_including_gist.out
new file mode 100644
index 0000000000..ed9906da66
--- /dev/null
+++ b/src/test/regress/expected/index_including_gist.out
@@ -0,0 +1,166 @@
+/*
+ * 1.1. test CREATE INDEX with buffered build
+ */
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+                                  pg_get_indexdef                                  
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+ c1 | c2 | c3 |     c4      
+----+----+----+-------------
+  1 |  2 |  3 | (2,3),(1,2)
+  2 |  4 |  6 | (4,5),(2,3)
+  3 |  6 |  9 | (6,7),(3,4)
+  4 |  8 | 12 | (8,9),(4,5)
+(4 rows)
+
+SET enable_bitmapscan TO off;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                   QUERY PLAN                   
+------------------------------------------------
+ Index Only Scan using tbl_gist_idx on tbl_gist
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+/*
+ * 1.2. test CREATE INDEX with inserts
+ */
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+                                  pg_get_indexdef                                  
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+ c1 | c2 | c3 |     c4      
+----+----+----+-------------
+  1 |  2 |  3 | (2,3),(1,2)
+  2 |  4 |  6 | (4,5),(2,3)
+  3 |  6 |  9 | (6,7),(3,4)
+  4 |  8 | 12 | (8,9),(4,5)
+(4 rows)
+
+SET enable_bitmapscan TO off;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                   QUERY PLAN                   
+------------------------------------------------
+ Index Only Scan using tbl_gist_idx on tbl_gist
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+/*
+ * 2. CREATE INDEX CONCURRENTLY
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX CONCURRENTLY tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                     indexdef                                      
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+DROP TABLE tbl_gist;
+/*
+ * 3. REINDEX
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                   indexdef                                    
+-------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c3)
+(1 row)
+
+REINDEX INDEX tbl_gist_idx;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                   indexdef                                    
+-------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c3)
+(1 row)
+
+ALTER TABLE tbl_gist DROP COLUMN c1;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+ indexdef 
+----------
+(0 rows)
+
+DROP TABLE tbl_gist;
+/*
+ * 4. Update, delete values in indexed table.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+UPDATE tbl_gist SET c1 = 100 WHERE c1 = 2;
+UPDATE tbl_gist SET c1 = 1 WHERE c1 = 3;
+DELETE FROM tbl_gist WHERE c1 = 5 OR c3 = 12;
+DROP TABLE tbl_gist;
+/*
+ * 5. Alter column type.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+ALTER TABLE tbl_gist ALTER c1 TYPE bigint;
+ALTER TABLE tbl_gist ALTER c3 TYPE bigint;
+\d tbl_gist
+              Table "public.tbl_gist"
+ Column |  Type   | Collation | Nullable | Default 
+--------+---------+-----------+----------+---------
+ c1     | bigint  |           |          | 
+ c2     | integer |           |          | 
+ c3     | bigint  |           |          | 
+ c4     | box     |           |          | 
+Indexes:
+    "tbl_gist_idx" gist (c4) INCLUDE (c1, c3)
+
+DROP TABLE tbl_gist;
+/*
+ * 6. EXCLUDE constraint.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box, EXCLUDE USING gist (c4 WITH &&) INCLUDE (c1, c2, c3));
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+ERROR:  conflicting key value violates exclusion constraint "tbl_gist_c4_c1_c2_c3_excl"
+DETAIL:  Key (c4)=((4,5),(2,3)) conflicts with existing key (c4)=((2,3),(1,2)).
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(3*x,2*x),point(3*x+1,2*x+1)) FROM generate_series(1,10) AS x;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                         QUERY PLAN                          
+-------------------------------------------------------------
+ Index Only Scan using tbl_gist_c4_c1_c2_c3_excl on tbl_gist
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+\d tbl_gist
+              Table "public.tbl_gist"
+ Column |  Type   | Collation | Nullable | Default 
+--------+---------+-----------+----------+---------
+ c1     | integer |           |          | 
+ c2     | integer |           |          | 
+ c3     | integer |           |          | 
+ c4     | box     |           |          | 
+Indexes:
+    "tbl_gist_c4_c1_c2_c3_excl" EXCLUDE USING gist (c4 WITH &&) INCLUDE (c1, c2, c3)
+
+DROP TABLE tbl_gist;
diff --git a/src/test/regress/parallel_schedule b/src/test/regress/parallel_schedule
index cc0bbf5db9..5684291ab0 100644
--- a/src/test/regress/parallel_schedule
+++ b/src/test/regress/parallel_schedule
@@ -55,7 +55,7 @@ test: copy copyselect copydml
 # ----------
 test: create_misc create_operator create_procedure
 # These depend on the above two
-test: create_index create_view index_including
+test: create_index create_view index_including index_including_gist
 
 # ----------
 # Another group of parallel tests
diff --git a/src/test/regress/serial_schedule b/src/test/regress/serial_schedule
index 0c10c7100c..309ccec81a 100644
--- a/src/test/regress/serial_schedule
+++ b/src/test/regress/serial_schedule
@@ -63,6 +63,7 @@ test: create_operator
 test: create_procedure
 test: create_index
 test: index_including
+test: index_including_gist
 test: create_view
 test: create_aggregate
 test: create_function_3
diff --git a/src/test/regress/sql/index_including.sql b/src/test/regress/sql/index_including.sql
index ef5fd882f5..b307d4975c 100644
--- a/src/test/regress/sql/index_including.sql
+++ b/src/test/regress/sql/index_including.sql
@@ -173,15 +173,21 @@ SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl' ORDER BY indexname;
 DROP TABLE tbl;
 
 /*
- * 7. Check various AMs. All but btree must fail.
+ * 7. Check various AMs. All but btree and gist must fail.
  */
 CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
 CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
+CREATE INDEX on tbl USING gist(c1, c2) INCLUDE (c3, c4);
 CREATE INDEX on tbl USING gist(c3) INCLUDE (c4);
+CREATE INDEX on tbl USING gist(c3) INCLUDE (c1);
+CREATE INDEX on tbl USING gist(c3) INCLUDE (c1, c4);
 CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
 CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
 CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
 CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
+CREATE INDEX on tbl USING rtree(c3) INCLUDE (c4);
+CREATE INDEX on tbl USING rtree(c3) INCLUDE (c1);
+CREATE INDEX on tbl USING rtree(c3) INCLUDE (c1, c4);
 CREATE INDEX on tbl USING btree(c1, c2) INCLUDE (c3, c4);
 DROP TABLE tbl;
 
diff --git a/src/test/regress/sql/index_including_gist.sql b/src/test/regress/sql/index_including_gist.sql
new file mode 100644
index 0000000000..7d5c99b2e7
--- /dev/null
+++ b/src/test/regress/sql/index_including_gist.sql
@@ -0,0 +1,90 @@
+/*
+ * 1.1. test CREATE INDEX with buffered build
+ */
+
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO off;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+
+/*
+ * 1.2. test CREATE INDEX with inserts
+ */
+
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO off;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+
+/*
+ * 2. CREATE INDEX CONCURRENTLY
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX CONCURRENTLY tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+DROP TABLE tbl_gist;
+
+
+/*
+ * 3. REINDEX
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+REINDEX INDEX tbl_gist_idx;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+ALTER TABLE tbl_gist DROP COLUMN c1;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+DROP TABLE tbl_gist;
+
+/*
+ * 4. Update, delete values in indexed table.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+UPDATE tbl_gist SET c1 = 100 WHERE c1 = 2;
+UPDATE tbl_gist SET c1 = 1 WHERE c1 = 3;
+DELETE FROM tbl_gist WHERE c1 = 5 OR c3 = 12;
+DROP TABLE tbl_gist;
+
+/*
+ * 5. Alter column type.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+ALTER TABLE tbl_gist ALTER c1 TYPE bigint;
+ALTER TABLE tbl_gist ALTER c3 TYPE bigint;
+\d tbl_gist
+DROP TABLE tbl_gist;
+
+/*
+ * 6. EXCLUDE constraint.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box, EXCLUDE USING gist (c4 WITH &&) INCLUDE (c1, c2, c3));
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(3*x,2*x),point(3*x+1,2*x+1)) FROM generate_series(1,10) AS x;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+\d tbl_gist
+DROP TABLE tbl_gist;
-- 
2.20.1

#23

andreas@proxel.se

almost 7 years ago

In reply to: Andrey Borodin (#22)

Re: Covering GiST indexes

On 29/01/2019 18.45, Andrey Borodin wrote:

Also, I've unified gist and r-tree syntax tests for INCLUDE.

Why not just keep it simple? The purpose is not to test the GiST
indexes, for that we have index_including_gist.sql.

I propose the following.

/*
* 7. Check various AMs. All but btree and gist must fail.
*/
CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
CREATE INDEX on tbl USING gist(c3) INCLUDE (c1, c4);
CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
CREATE INDEX on tbl USING rtree(c3) INCLUDE (c1, c4);
DROP TABLE tbl;

and

/*
* 7. Check various AMs. All but btree and gist must fail.
*/
CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
ERROR: access method "brin" does not support included columns
CREATE INDEX on tbl USING gist(c3) INCLUDE (c1, c4);
CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
ERROR: access method "spgist" does not support included columns
CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
ERROR: access method "gin" does not support included columns
CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
ERROR: access method "hash" does not support included columns
CREATE INDEX on tbl USING rtree(c3) INCLUDE (c1, c4);
NOTICE: substituting access method "gist" for obsolete method "rtree"
DROP TABLE tbl;

#24

x4mmm@yandex-team.ru

almost 7 years ago

In reply to: Andreas Karlsson (#23)

1 attachment(s)

Re: Covering GiST indexes

29 янв. 2019 г., в 23:43, Andreas Karlsson <andreas@proxel.se> написал(а):

On 29/01/2019 18.45, Andrey Borodin wrote:

Also, I've unified gist and r-tree syntax tests for INCLUDE.

Why not just keep it simple? The purpose is not to test the GiST indexes, for that we have index_including_gist.sql.

I propose the following.

Indeed, it's quite strange to check that GiST cannot be built without GiST opclass.

PFA patch without redundant checks.

Best regards, Andrey Borodin.

Attachments:

0001-Covering-GiST-v9.patchapplication/octet-stream; name=0001-Covering-GiST-v9.patch; x-unix-mode=0644Download

From d8a12b796449076197bf9a256b323cb28e4689e9 Mon Sep 17 00:00:00 2001
From: Andrey <amborodin@acm.org>
Date: Mon, 26 Nov 2018 21:55:45 +0500
Subject: [PATCH] Covering GiST v9

This patch allow adding INCLUDE colums to GiST index.
These colums do not participate in GiST tree build,
are not used for chosing subtree for inserts or splits.
But they can be fetched during IndexOnlyScan.
---
 doc/src/sgml/ref/create_index.sgml            |   4 +-
 doc/src/sgml/textsearch.sgml                  |   6 +
 src/backend/access/gist/gist.c                |  33 +++-
 src/backend/access/gist/gistget.c             |   4 +-
 src/backend/access/gist/gistscan.c            |  12 +-
 src/backend/access/gist/gistsplit.c           |  12 +-
 src/backend/access/gist/gistutil.c            |  42 ++++-
 src/include/access/gist_private.h             |   2 +
 src/test/regress/expected/amutils.out         |   4 +-
 src/test/regress/expected/index_including.out |   9 +-
 .../regress/expected/index_including_gist.out | 166 ++++++++++++++++++
 src/test/regress/parallel_schedule            |   2 +-
 src/test/regress/serial_schedule              |   1 +
 src/test/regress/sql/index_including.sql      |   7 +-
 src/test/regress/sql/index_including_gist.sql |  90 ++++++++++
 15 files changed, 358 insertions(+), 36 deletions(-)
 create mode 100644 src/test/regress/expected/index_including_gist.out
 create mode 100644 src/test/regress/sql/index_including_gist.sql

diff --git a/doc/src/sgml/ref/create_index.sgml b/doc/src/sgml/ref/create_index.sgml
index ad619cdcfe..f8657c6ae2 100644
--- a/doc/src/sgml/ref/create_index.sgml
+++ b/doc/src/sgml/ref/create_index.sgml
@@ -181,8 +181,8 @@ CREATE [ UNIQUE ] INDEX [ CONCURRENTLY ] [ [ IF NOT EXISTS ] <replaceable class=
        </para>
 
        <para>
-        Currently, only the B-tree index access method supports this feature.
-        In B-tree indexes, the values of columns listed in the
+        Currently, the B-tree and the GiST index access methods supports this
+        feature. In B-tree indexes, the values of columns listed in the
         <literal>INCLUDE</literal> clause are included in leaf tuples which
         correspond to heap tuples, but are not included in upper-level
         index entries used for tree navigation.
diff --git a/doc/src/sgml/textsearch.sgml b/doc/src/sgml/textsearch.sgml
index ecebade767..e4c75ebf6f 100644
--- a/doc/src/sgml/textsearch.sgml
+++ b/doc/src/sgml/textsearch.sgml
@@ -3674,6 +3674,12 @@ SELECT plainto_tsquery('supernovae stars');
    retrieved to see if the match is correct.
   </para>
 
+  <para>
+   A GiST index can be covering, i.e. use the <literal>INCLUDE</literal>
+   clause. Included columns can have data types without any GiST operator
+   class. Included attributes will be stored uncompressed.
+  </para>
+
   <para>
    Lossiness causes performance degradation due to unnecessary fetches of table
    records that turn out to be false matches.  Since random access to table
diff --git a/src/backend/access/gist/gist.c b/src/backend/access/gist/gist.c
index b75b3a8dac..d6ee44dc9c 100644
--- a/src/backend/access/gist/gist.c
+++ b/src/backend/access/gist/gist.c
@@ -75,7 +75,7 @@ gisthandler(PG_FUNCTION_ARGS)
 	amroutine->amclusterable = true;
 	amroutine->ampredlocks = true;
 	amroutine->amcanparallel = false;
-	amroutine->amcaninclude = false;
+	amroutine->amcaninclude = true;
 	amroutine->amkeytype = InvalidOid;
 
 	amroutine->ambuild = gistbuild;
@@ -1382,8 +1382,8 @@ gistSplit(Relation r,
 						IndexTupleSize(itup[0]), GiSTPageSize,
 						RelationGetRelationName(r))));
 
-	memset(v.spl_lisnull, true, sizeof(bool) * giststate->tupdesc->natts);
-	memset(v.spl_risnull, true, sizeof(bool) * giststate->tupdesc->natts);
+	memset(v.spl_lisnull, true, sizeof(bool) * giststate->truncTupdesc->natts);
+	memset(v.spl_risnull, true, sizeof(bool) * giststate->truncTupdesc->natts);
 	gistSplitByKey(r, page, itup, len, giststate, &v, 0);
 
 	/* form left and right vector */
@@ -1462,8 +1462,19 @@ initGISTstate(Relation index)
 	giststate->scanCxt = scanCxt;
 	giststate->tempCxt = scanCxt;	/* caller must change this if needed */
 	giststate->tupdesc = index->rd_att;
+	/*
+	 * truncTupdesc for truncated tuples may contain less attributes, because
+	 * it does not store INCLUDE attributes.
+	 * It is used to form tuples during tuple adjustement and page split.
+	 * B-tree creates shortened tuple descriptor for every truncated tuple,
+	 * because it is doing this less often: it does not have to form
+	 * truncated tuples during page split. Also, B-tree is not ajdusting
+	 * tuples on internal pages the way GiST does.
+	 */
+	giststate->truncTupdesc = CreateTupleDescCopyConstr(index->rd_att);
+	giststate->truncTupdesc->natts = IndexRelationGetNumberOfKeyAttributes(index);
 
-	for (i = 0; i < index->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(index); i++)
 	{
 		fmgr_info_copy(&(giststate->consistentFn[i]),
 					   index_getprocinfo(index, i + 1, GIST_CONSISTENT_PROC),
@@ -1531,6 +1542,20 @@ initGISTstate(Relation index)
 			giststate->supportCollation[i] = DEFAULT_COLLATION_OID;
 	}
 
+	for (; i < index->rd_att->natts; i++)
+	{
+		giststate->consistentFn[i].fn_oid = InvalidOid;
+		giststate->unionFn[i].fn_oid = InvalidOid;
+		giststate->compressFn[i].fn_oid = InvalidOid;
+		giststate->decompressFn[i].fn_oid = InvalidOid;
+		giststate->penaltyFn[i].fn_oid = InvalidOid;
+		giststate->picksplitFn[i].fn_oid = InvalidOid;
+		giststate->equalFn[i].fn_oid = InvalidOid;
+		giststate->distanceFn[i].fn_oid = InvalidOid;
+		giststate->fetchFn[i].fn_oid = InvalidOid;
+		giststate->supportCollation[i] = InvalidOid;
+	}
+
 	MemoryContextSwitchTo(oldCxt);
 
 	return giststate;
diff --git a/src/backend/access/gist/gistget.c b/src/backend/access/gist/gistget.c
index a96ef5c3ac..a4ee8763f6 100644
--- a/src/backend/access/gist/gistget.c
+++ b/src/backend/access/gist/gistget.c
@@ -769,11 +769,13 @@ gistgetbitmap(IndexScanDesc scan, TIDBitmap *tbm)
  *
  * Opclasses that implement a fetch function support index-only scans.
  * Opclasses without compression functions also support index-only scans.
+ * Included attributes always can be fetched for index-only scans.
  */
 bool
 gistcanreturn(Relation index, int attno)
 {
-	if (OidIsValid(index_getprocid(index, attno, GIST_FETCH_PROC)) ||
+	if (attno > IndexRelationGetNumberOfKeyAttributes(index) ||
+		OidIsValid(index_getprocid(index, attno, GIST_FETCH_PROC)) ||
 		!OidIsValid(index_getprocid(index, attno, GIST_COMPRESS_PROC)))
 		return true;
 	else
diff --git a/src/backend/access/gist/gistscan.c b/src/backend/access/gist/gistscan.c
index 78a8ede794..4fffc8b33a 100644
--- a/src/backend/access/gist/gistscan.c
+++ b/src/backend/access/gist/gistscan.c
@@ -158,6 +158,7 @@ gistrescan(IndexScanDesc scan, ScanKey key, int nkeys,
 	if (scan->xs_want_itup && !scan->xs_hitupdesc)
 	{
 		int			natts;
+		int			nkeyatts;
 		int			attno;
 
 		/*
@@ -167,13 +168,22 @@ gistrescan(IndexScanDesc scan, ScanKey key, int nkeys,
 		 * types.
 		 */
 		natts = RelationGetNumberOfAttributes(scan->indexRelation);
+		nkeyatts = IndexRelationGetNumberOfKeyAttributes(scan->indexRelation);
 		so->giststate->fetchTupdesc = CreateTemplateTupleDesc(natts);
-		for (attno = 1; attno <= natts; attno++)
+		for (attno = 1; attno <= nkeyatts; attno++)
 		{
 			TupleDescInitEntry(so->giststate->fetchTupdesc, attno, NULL,
 							   scan->indexRelation->rd_opcintype[attno - 1],
 							   -1, 0);
 		}
+
+		for (; attno <= natts; attno++)
+		{
+			/* taking opcintype from giststate->tupdesc */
+			TupleDescInitEntry(so->giststate->fetchTupdesc, attno, NULL,
+					TupleDescAttr(so->giststate->tupdesc, attno - 1)->atttypid,
+							   -1, 0);
+		}
 		scan->xs_hitupdesc = so->giststate->fetchTupdesc;
 
 		/* Also create a memory context that will hold the returned tuples */
diff --git a/src/backend/access/gist/gistsplit.c b/src/backend/access/gist/gistsplit.c
index f210e0c39f..cf3af2821a 100644
--- a/src/backend/access/gist/gistsplit.c
+++ b/src/backend/access/gist/gistsplit.c
@@ -207,7 +207,7 @@ placeOne(Relation r, GISTSTATE *giststate, GistSplitVector *v,
 	gistDeCompressAtt(giststate, r, itup, NULL, (OffsetNumber) 0,
 					  identry, isnull);
 
-	for (; attno < giststate->tupdesc->natts; attno++)
+	for (; attno < giststate->truncTupdesc->natts; attno++)
 	{
 		float		lpenalty,
 					rpenalty;
@@ -485,7 +485,7 @@ gistUserPicksplit(Relation r, GistEntryVector *entryvec, int attno, GistSplitVec
 	 */
 	v->spl_dontcare = NULL;
 
-	if (attno + 1 < giststate->tupdesc->natts)
+	if (attno + 1 < giststate->truncTupdesc->natts)
 	{
 		int			NumDontCare;
 
@@ -657,7 +657,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 		 */
 		v->spl_risnull[attno] = v->spl_lisnull[attno] = true;
 
-		if (attno + 1 < giststate->tupdesc->natts)
+		if (attno + 1 < giststate->truncTupdesc->natts)
 			gistSplitByKey(r, page, itup, len, giststate, v, attno + 1);
 		else
 			gistSplitHalf(&v->splitVector, len);
@@ -683,7 +683,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 				v->splitVector.spl_left[v->splitVector.spl_nleft++] = i;
 
 		/* Compute union keys, unless outer recursion level will handle it */
-		if (attno == 0 && giststate->tupdesc->natts == 1)
+		if (attno == 0 && giststate->truncTupdesc->natts == 1)
 		{
 			v->spl_dontcare = NULL;
 			gistunionsubkey(giststate, itup, v);
@@ -700,7 +700,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 			 * Splitting on attno column is not optimal, so consider
 			 * redistributing don't-care tuples according to the next column
 			 */
-			Assert(attno + 1 < giststate->tupdesc->natts);
+			Assert(attno + 1 < giststate->truncTupdesc->natts);
 
 			if (v->spl_dontcare == NULL)
 			{
@@ -771,7 +771,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 	 * that PickSplit (or the special cases above) produced correct union
 	 * datums.
 	 */
-	if (attno == 0 && giststate->tupdesc->natts > 1)
+	if (attno == 0 && giststate->truncTupdesc->natts > 1)
 	{
 		v->spl_dontcare = NULL;
 		gistunionsubkey(giststate, itup, v);
diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c
index 8d3dfad27b..f7024eda12 100644
--- a/src/backend/access/gist/gistutil.c
+++ b/src/backend/access/gist/gistutil.c
@@ -160,7 +160,7 @@ gistMakeUnionItVec(GISTSTATE *giststate, IndexTuple *itvec, int len,
 
 	evec = (GistEntryVector *) palloc((len + 2) * sizeof(GISTENTRY) + GEVHDRSZ);
 
-	for (i = 0; i < giststate->tupdesc->natts; i++)
+	for (i = 0; i < giststate->truncTupdesc->natts; i++)
 	{
 		int			j;
 
@@ -296,7 +296,7 @@ gistDeCompressAtt(GISTSTATE *giststate, Relation r, IndexTuple tuple, Page p,
 {
 	int			i;
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		Datum		datum;
 
@@ -329,7 +329,7 @@ gistgetadjusted(Relation r, IndexTuple oldtup, IndexTuple addtup, GISTSTATE *gis
 	gistDeCompressAtt(giststate, r, addtup, NULL,
 					  (OffsetNumber) 0, addentries, addisnull);
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		gistMakeUnionKey(giststate, i,
 						 oldentries + i, oldisnull[i],
@@ -442,7 +442,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 		zero_penalty = true;
 
 		/* Loop over index attributes. */
-		for (j = 0; j < r->rd_att->natts; j++)
+		for (j = 0; j < IndexRelationGetNumberOfKeyAttributes(r); j++)
 		{
 			Datum		datum;
 			float		usize;
@@ -470,7 +470,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 				result = i;
 				best_penalty[j] = usize;
 
-				if (j < r->rd_att->natts - 1)
+				if (j < IndexRelationGetNumberOfKeyAttributes(r) - 1)
 					best_penalty[j + 1] = -1;
 
 				/* we have new best, so reset keep-it decision */
@@ -500,7 +500,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 		 * If we looped past the last column, and did not update "result",
 		 * then this tuple is exactly as good as the prior best tuple.
 		 */
-		if (j == r->rd_att->natts && result != i)
+		if (j == IndexRelationGetNumberOfKeyAttributes(r) && result != i)
 		{
 			if (keep_current_best == -1)
 			{
@@ -575,11 +575,13 @@ gistFormTuple(GISTSTATE *giststate, Relation r,
 	Datum		compatt[INDEX_MAX_KEYS];
 	int			i;
 	IndexTuple	res;
+	/* currently we truncate tuples iif they are on internal pages */
+	bool isTruncated = !isleaf;
 
 	/*
 	 * Call the compress method on each attribute.
 	 */
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		if (isnull[i])
 			compatt[i] = (Datum) 0;
@@ -602,7 +604,21 @@ gistFormTuple(GISTSTATE *giststate, Relation r,
 		}
 	}
 
-	res = index_form_tuple(giststate->tupdesc, compatt, isnull);
+	if (!isTruncated)
+	{
+		/*
+		 * Emplace each included attribute.
+		 */
+		for (; i < r->rd_att->natts; i++)
+		{
+			if (isnull[i])
+				compatt[i] = (Datum) 0;
+			else
+				compatt[i] = attdata[i];
+		}
+	}
+
+	res = index_form_tuple(isTruncated ? giststate->truncTupdesc : giststate->tupdesc, compatt, isnull);
 
 	/*
 	 * The offset number on tuples on internal pages is unused. For historical
@@ -644,7 +660,7 @@ gistFetchTuple(GISTSTATE *giststate, Relation r, IndexTuple tuple)
 	bool		isnull[INDEX_MAX_KEYS];
 	int			i;
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		Datum		datum;
 
@@ -679,6 +695,14 @@ gistFetchTuple(GISTSTATE *giststate, Relation r, IndexTuple tuple)
 			fetchatt[i] = (Datum) 0;
 		}
 	}
+
+	/*
+	 * Get each included attribute.
+	 */
+	for (; i < r->rd_att->natts; i++)
+	{
+		fetchatt[i] = index_getattr(tuple, i + 1, giststate->tupdesc, &isnull[i]);
+	}
 	MemoryContextSwitchTo(oldcxt);
 
 	return heap_form_tuple(giststate->fetchTupdesc, fetchatt, isnull);
diff --git a/src/include/access/gist_private.h b/src/include/access/gist_private.h
index 3698942f9d..ccf06e676c 100644
--- a/src/include/access/gist_private.h
+++ b/src/include/access/gist_private.h
@@ -79,6 +79,8 @@ typedef struct GISTSTATE
 	MemoryContext tempCxt;		/* short-term context for calling functions */
 
 	TupleDesc	tupdesc;		/* index's tuple descriptor */
+	TupleDesc	truncTupdesc;	/* truncated tuple descriptor
+								 * for internal pages */
 	TupleDesc	fetchTupdesc;	/* tuple descriptor for tuples returned in an
 								 * index-only scan */
 
diff --git a/src/test/regress/expected/amutils.out b/src/test/regress/expected/amutils.out
index 4570a39b05..d92a6d12c6 100644
--- a/src/test/regress/expected/amutils.out
+++ b/src/test/regress/expected/amutils.out
@@ -75,7 +75,7 @@ select prop,
  can_unique         | f  |       | 
  can_multi_col      | t  |       | 
  can_exclude        | t  |       | 
- can_include        | f  |       | 
+ can_include        | t  |       | 
  bogus              |    |       | 
 (19 rows)
 
@@ -159,7 +159,7 @@ select amname, prop, pg_indexam_has_property(a.oid, prop) as p
  gist   | can_unique    | f
  gist   | can_multi_col | t
  gist   | can_exclude   | t
- gist   | can_include   | f
+ gist   | can_include   | t
  gist   | bogus         | 
  hash   | can_order     | f
  hash   | can_unique    | f
diff --git a/src/test/regress/expected/index_including.out b/src/test/regress/expected/index_including.out
index 16b4be34de..896fc9a24a 100644
--- a/src/test/regress/expected/index_including.out
+++ b/src/test/regress/expected/index_including.out
@@ -313,23 +313,20 @@ SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl' ORDER BY indexname;
 
 DROP TABLE tbl;
 /*
- * 7. Check various AMs. All but btree must fail.
+ * 7. Check various AMs. All but btree and gist must fail.
  */
 CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
 CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "brin" does not support included columns
-CREATE INDEX on tbl USING gist(c3) INCLUDE (c4);
-ERROR:  access method "gist" does not support included columns
+CREATE INDEX on tbl USING gist(c3) INCLUDE (c1, c4);
 CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
 ERROR:  access method "spgist" does not support included columns
 CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "gin" does not support included columns
 CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "hash" does not support included columns
-CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
+CREATE INDEX on tbl USING rtree(c3) INCLUDE (c1, c4);
 NOTICE:  substituting access method "gist" for obsolete method "rtree"
-ERROR:  access method "gist" does not support included columns
-CREATE INDEX on tbl USING btree(c1, c2) INCLUDE (c3, c4);
 DROP TABLE tbl;
 /*
  * 8. Update, delete values in indexed table.
diff --git a/src/test/regress/expected/index_including_gist.out b/src/test/regress/expected/index_including_gist.out
new file mode 100644
index 0000000000..ed9906da66
--- /dev/null
+++ b/src/test/regress/expected/index_including_gist.out
@@ -0,0 +1,166 @@
+/*
+ * 1.1. test CREATE INDEX with buffered build
+ */
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+                                  pg_get_indexdef                                  
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+ c1 | c2 | c3 |     c4      
+----+----+----+-------------
+  1 |  2 |  3 | (2,3),(1,2)
+  2 |  4 |  6 | (4,5),(2,3)
+  3 |  6 |  9 | (6,7),(3,4)
+  4 |  8 | 12 | (8,9),(4,5)
+(4 rows)
+
+SET enable_bitmapscan TO off;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                   QUERY PLAN                   
+------------------------------------------------
+ Index Only Scan using tbl_gist_idx on tbl_gist
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+/*
+ * 1.2. test CREATE INDEX with inserts
+ */
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+                                  pg_get_indexdef                                  
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+ c1 | c2 | c3 |     c4      
+----+----+----+-------------
+  1 |  2 |  3 | (2,3),(1,2)
+  2 |  4 |  6 | (4,5),(2,3)
+  3 |  6 |  9 | (6,7),(3,4)
+  4 |  8 | 12 | (8,9),(4,5)
+(4 rows)
+
+SET enable_bitmapscan TO off;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                   QUERY PLAN                   
+------------------------------------------------
+ Index Only Scan using tbl_gist_idx on tbl_gist
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+/*
+ * 2. CREATE INDEX CONCURRENTLY
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX CONCURRENTLY tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                     indexdef                                      
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+DROP TABLE tbl_gist;
+/*
+ * 3. REINDEX
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                   indexdef                                    
+-------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c3)
+(1 row)
+
+REINDEX INDEX tbl_gist_idx;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                   indexdef                                    
+-------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c3)
+(1 row)
+
+ALTER TABLE tbl_gist DROP COLUMN c1;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+ indexdef 
+----------
+(0 rows)
+
+DROP TABLE tbl_gist;
+/*
+ * 4. Update, delete values in indexed table.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+UPDATE tbl_gist SET c1 = 100 WHERE c1 = 2;
+UPDATE tbl_gist SET c1 = 1 WHERE c1 = 3;
+DELETE FROM tbl_gist WHERE c1 = 5 OR c3 = 12;
+DROP TABLE tbl_gist;
+/*
+ * 5. Alter column type.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+ALTER TABLE tbl_gist ALTER c1 TYPE bigint;
+ALTER TABLE tbl_gist ALTER c3 TYPE bigint;
+\d tbl_gist
+              Table "public.tbl_gist"
+ Column |  Type   | Collation | Nullable | Default 
+--------+---------+-----------+----------+---------
+ c1     | bigint  |           |          | 
+ c2     | integer |           |          | 
+ c3     | bigint  |           |          | 
+ c4     | box     |           |          | 
+Indexes:
+    "tbl_gist_idx" gist (c4) INCLUDE (c1, c3)
+
+DROP TABLE tbl_gist;
+/*
+ * 6. EXCLUDE constraint.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box, EXCLUDE USING gist (c4 WITH &&) INCLUDE (c1, c2, c3));
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+ERROR:  conflicting key value violates exclusion constraint "tbl_gist_c4_c1_c2_c3_excl"
+DETAIL:  Key (c4)=((4,5),(2,3)) conflicts with existing key (c4)=((2,3),(1,2)).
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(3*x,2*x),point(3*x+1,2*x+1)) FROM generate_series(1,10) AS x;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                         QUERY PLAN                          
+-------------------------------------------------------------
+ Index Only Scan using tbl_gist_c4_c1_c2_c3_excl on tbl_gist
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+\d tbl_gist
+              Table "public.tbl_gist"
+ Column |  Type   | Collation | Nullable | Default 
+--------+---------+-----------+----------+---------
+ c1     | integer |           |          | 
+ c2     | integer |           |          | 
+ c3     | integer |           |          | 
+ c4     | box     |           |          | 
+Indexes:
+    "tbl_gist_c4_c1_c2_c3_excl" EXCLUDE USING gist (c4 WITH &&) INCLUDE (c1, c2, c3)
+
+DROP TABLE tbl_gist;
diff --git a/src/test/regress/parallel_schedule b/src/test/regress/parallel_schedule
index cc0bbf5db9..5684291ab0 100644
--- a/src/test/regress/parallel_schedule
+++ b/src/test/regress/parallel_schedule
@@ -55,7 +55,7 @@ test: copy copyselect copydml
 # ----------
 test: create_misc create_operator create_procedure
 # These depend on the above two
-test: create_index create_view index_including
+test: create_index create_view index_including index_including_gist
 
 # ----------
 # Another group of parallel tests
diff --git a/src/test/regress/serial_schedule b/src/test/regress/serial_schedule
index 0c10c7100c..309ccec81a 100644
--- a/src/test/regress/serial_schedule
+++ b/src/test/regress/serial_schedule
@@ -63,6 +63,7 @@ test: create_operator
 test: create_procedure
 test: create_index
 test: index_including
+test: index_including_gist
 test: create_view
 test: create_aggregate
 test: create_function_3
diff --git a/src/test/regress/sql/index_including.sql b/src/test/regress/sql/index_including.sql
index ef5fd882f5..78d0cb7578 100644
--- a/src/test/regress/sql/index_including.sql
+++ b/src/test/regress/sql/index_including.sql
@@ -173,16 +173,15 @@ SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl' ORDER BY indexname;
 DROP TABLE tbl;
 
 /*
- * 7. Check various AMs. All but btree must fail.
+ * 7. Check various AMs. All but btree and gist must fail.
  */
 CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
 CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
-CREATE INDEX on tbl USING gist(c3) INCLUDE (c4);
+CREATE INDEX on tbl USING gist(c3) INCLUDE (c1, c4);
 CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
 CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
 CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
-CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
-CREATE INDEX on tbl USING btree(c1, c2) INCLUDE (c3, c4);
+CREATE INDEX on tbl USING rtree(c3) INCLUDE (c1, c4);
 DROP TABLE tbl;
 
 /*
diff --git a/src/test/regress/sql/index_including_gist.sql b/src/test/regress/sql/index_including_gist.sql
new file mode 100644
index 0000000000..7d5c99b2e7
--- /dev/null
+++ b/src/test/regress/sql/index_including_gist.sql
@@ -0,0 +1,90 @@
+/*
+ * 1.1. test CREATE INDEX with buffered build
+ */
+
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO off;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+
+/*
+ * 1.2. test CREATE INDEX with inserts
+ */
+
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO off;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+
+/*
+ * 2. CREATE INDEX CONCURRENTLY
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX CONCURRENTLY tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+DROP TABLE tbl_gist;
+
+
+/*
+ * 3. REINDEX
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+REINDEX INDEX tbl_gist_idx;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+ALTER TABLE tbl_gist DROP COLUMN c1;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+DROP TABLE tbl_gist;
+
+/*
+ * 4. Update, delete values in indexed table.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+UPDATE tbl_gist SET c1 = 100 WHERE c1 = 2;
+UPDATE tbl_gist SET c1 = 1 WHERE c1 = 3;
+DELETE FROM tbl_gist WHERE c1 = 5 OR c3 = 12;
+DROP TABLE tbl_gist;
+
+/*
+ * 5. Alter column type.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+ALTER TABLE tbl_gist ALTER c1 TYPE bigint;
+ALTER TABLE tbl_gist ALTER c3 TYPE bigint;
+\d tbl_gist
+DROP TABLE tbl_gist;
+
+/*
+ * 6. EXCLUDE constraint.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box, EXCLUDE USING gist (c4 WITH &&) INCLUDE (c1, c2, c3));
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(3*x,2*x),point(3*x+1,2*x+1)) FROM generate_series(1,10) AS x;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+\d tbl_gist
+DROP TABLE tbl_gist;
-- 
2.20.1

#25

andreas@proxel.se

almost 7 years ago

In reply to: Andrey Borodin (#24)

1 attachment(s)

Re: Covering GiST indexes

On 29/01/2019 19.58, Andrey Borodin wrote:

PFA patch without redundant checks.

I hope it is ok that I fixed a typo and some wording in the comment,
plus re-added the btree query which I accidentally removed in my last mail.

I think the current state of the patch is fine, assuming the solution
with truncTupdesc is deemed to be good enough by the committer, so I
will mark this as ready for committer.

Thanks for the patch and the speedy fixes!

Andreas

Attachments:

0001-Covering-GiST-v10.patchtext/x-patch; name=0001-Covering-GiST-v10.patchDownload

From f1a85940d01e208a6502e689616a8806146c9f3d Mon Sep 17 00:00:00 2001
From: Andreas Karlsson <andreas@proxel.se>
Date: Wed, 30 Jan 2019 01:59:14 +0100
Subject: [PATCH] Covering GiST v10

This patch allow adding INCLUDE colums to GiST index.
These colums do not participate in GiST tree build,
are not used for chosing subtree for inserts or splits.
But they can be fetched during IndexOnlyScan.
---
 doc/src/sgml/ref/create_index.sgml                 |   4 +-
 doc/src/sgml/textsearch.sgml                       |   6 +
 src/backend/access/gist/gist.c                     |  33 +++-
 src/backend/access/gist/gistget.c                  |   4 +-
 src/backend/access/gist/gistscan.c                 |  12 +-
 src/backend/access/gist/gistsplit.c                |  12 +-
 src/backend/access/gist/gistutil.c                 |  42 ++++--
 src/include/access/gist_private.h                  |   2 +
 src/test/regress/expected/amutils.out              |   4 +-
 src/test/regress/expected/index_including.out      |   8 +-
 src/test/regress/expected/index_including_gist.out | 166 +++++++++++++++++++++
 src/test/regress/parallel_schedule                 |   2 +-
 src/test/regress/serial_schedule                   |   1 +
 src/test/regress/sql/index_including.sql           |   6 +-
 src/test/regress/sql/index_including_gist.sql      |  90 +++++++++++
 15 files changed, 358 insertions(+), 34 deletions(-)
 create mode 100644 src/test/regress/expected/index_including_gist.out
 create mode 100644 src/test/regress/sql/index_including_gist.sql

diff --git a/doc/src/sgml/ref/create_index.sgml b/doc/src/sgml/ref/create_index.sgml
index ad619cdcfe..f8657c6ae2 100644
--- a/doc/src/sgml/ref/create_index.sgml
+++ b/doc/src/sgml/ref/create_index.sgml
@@ -181,8 +181,8 @@ CREATE [ UNIQUE ] INDEX [ CONCURRENTLY ] [ [ IF NOT EXISTS ] <replaceable class=
        </para>
 
        <para>
-        Currently, only the B-tree index access method supports this feature.
-        In B-tree indexes, the values of columns listed in the
+        Currently, the B-tree and the GiST index access methods supports this
+        feature. In B-tree indexes, the values of columns listed in the
         <literal>INCLUDE</literal> clause are included in leaf tuples which
         correspond to heap tuples, but are not included in upper-level
         index entries used for tree navigation.
diff --git a/doc/src/sgml/textsearch.sgml b/doc/src/sgml/textsearch.sgml
index ecebade767..e4c75ebf6f 100644
--- a/doc/src/sgml/textsearch.sgml
+++ b/doc/src/sgml/textsearch.sgml
@@ -3675,6 +3675,12 @@ SELECT plainto_tsquery('supernovae stars');
   </para>
 
   <para>
+   A GiST index can be covering, i.e. use the <literal>INCLUDE</literal>
+   clause. Included columns can have data types without any GiST operator
+   class. Included attributes will be stored uncompressed.
+  </para>
+
+  <para>
    Lossiness causes performance degradation due to unnecessary fetches of table
    records that turn out to be false matches.  Since random access to table
    records is slow, this limits the usefulness of GiST indexes.  The
diff --git a/src/backend/access/gist/gist.c b/src/backend/access/gist/gist.c
index b75b3a8dac..2bbf7a7f49 100644
--- a/src/backend/access/gist/gist.c
+++ b/src/backend/access/gist/gist.c
@@ -75,7 +75,7 @@ gisthandler(PG_FUNCTION_ARGS)
 	amroutine->amclusterable = true;
 	amroutine->ampredlocks = true;
 	amroutine->amcanparallel = false;
-	amroutine->amcaninclude = false;
+	amroutine->amcaninclude = true;
 	amroutine->amkeytype = InvalidOid;
 
 	amroutine->ambuild = gistbuild;
@@ -1382,8 +1382,8 @@ gistSplit(Relation r,
 						IndexTupleSize(itup[0]), GiSTPageSize,
 						RelationGetRelationName(r))));
 
-	memset(v.spl_lisnull, true, sizeof(bool) * giststate->tupdesc->natts);
-	memset(v.spl_risnull, true, sizeof(bool) * giststate->tupdesc->natts);
+	memset(v.spl_lisnull, true, sizeof(bool) * giststate->truncTupdesc->natts);
+	memset(v.spl_risnull, true, sizeof(bool) * giststate->truncTupdesc->natts);
 	gistSplitByKey(r, page, itup, len, giststate, &v, 0);
 
 	/* form left and right vector */
@@ -1462,8 +1462,19 @@ initGISTstate(Relation index)
 	giststate->scanCxt = scanCxt;
 	giststate->tempCxt = scanCxt;	/* caller must change this if needed */
 	giststate->tupdesc = index->rd_att;
+	/*
+	 * The truncated tupdesc does not contain the INCLUDE attributes.
+	 *
+	 * It is used to form tuples during tuple adjustement and page split.
+	 * B-tree creates shortened tuple descriptor for every truncated tuple,
+	 * because it is doing this less often: it does not have to form
+	 * truncated tuples during page split. Also, B-tree is not adjusting
+	 * tuples on internal pages the way GiST does.
+	 */
+	giststate->truncTupdesc = CreateTupleDescCopyConstr(index->rd_att);
+	giststate->truncTupdesc->natts = IndexRelationGetNumberOfKeyAttributes(index);
 
-	for (i = 0; i < index->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(index); i++)
 	{
 		fmgr_info_copy(&(giststate->consistentFn[i]),
 					   index_getprocinfo(index, i + 1, GIST_CONSISTENT_PROC),
@@ -1531,6 +1542,20 @@ initGISTstate(Relation index)
 			giststate->supportCollation[i] = DEFAULT_COLLATION_OID;
 	}
 
+	for (; i < index->rd_att->natts; i++)
+	{
+		giststate->consistentFn[i].fn_oid = InvalidOid;
+		giststate->unionFn[i].fn_oid = InvalidOid;
+		giststate->compressFn[i].fn_oid = InvalidOid;
+		giststate->decompressFn[i].fn_oid = InvalidOid;
+		giststate->penaltyFn[i].fn_oid = InvalidOid;
+		giststate->picksplitFn[i].fn_oid = InvalidOid;
+		giststate->equalFn[i].fn_oid = InvalidOid;
+		giststate->distanceFn[i].fn_oid = InvalidOid;
+		giststate->fetchFn[i].fn_oid = InvalidOid;
+		giststate->supportCollation[i] = InvalidOid;
+	}
+
 	MemoryContextSwitchTo(oldCxt);
 
 	return giststate;
diff --git a/src/backend/access/gist/gistget.c b/src/backend/access/gist/gistget.c
index a96ef5c3ac..a4ee8763f6 100644
--- a/src/backend/access/gist/gistget.c
+++ b/src/backend/access/gist/gistget.c
@@ -769,11 +769,13 @@ gistgetbitmap(IndexScanDesc scan, TIDBitmap *tbm)
  *
  * Opclasses that implement a fetch function support index-only scans.
  * Opclasses without compression functions also support index-only scans.
+ * Included attributes always can be fetched for index-only scans.
  */
 bool
 gistcanreturn(Relation index, int attno)
 {
-	if (OidIsValid(index_getprocid(index, attno, GIST_FETCH_PROC)) ||
+	if (attno > IndexRelationGetNumberOfKeyAttributes(index) ||
+		OidIsValid(index_getprocid(index, attno, GIST_FETCH_PROC)) ||
 		!OidIsValid(index_getprocid(index, attno, GIST_COMPRESS_PROC)))
 		return true;
 	else
diff --git a/src/backend/access/gist/gistscan.c b/src/backend/access/gist/gistscan.c
index 78a8ede794..4fffc8b33a 100644
--- a/src/backend/access/gist/gistscan.c
+++ b/src/backend/access/gist/gistscan.c
@@ -158,6 +158,7 @@ gistrescan(IndexScanDesc scan, ScanKey key, int nkeys,
 	if (scan->xs_want_itup && !scan->xs_hitupdesc)
 	{
 		int			natts;
+		int			nkeyatts;
 		int			attno;
 
 		/*
@@ -167,13 +168,22 @@ gistrescan(IndexScanDesc scan, ScanKey key, int nkeys,
 		 * types.
 		 */
 		natts = RelationGetNumberOfAttributes(scan->indexRelation);
+		nkeyatts = IndexRelationGetNumberOfKeyAttributes(scan->indexRelation);
 		so->giststate->fetchTupdesc = CreateTemplateTupleDesc(natts);
-		for (attno = 1; attno <= natts; attno++)
+		for (attno = 1; attno <= nkeyatts; attno++)
 		{
 			TupleDescInitEntry(so->giststate->fetchTupdesc, attno, NULL,
 							   scan->indexRelation->rd_opcintype[attno - 1],
 							   -1, 0);
 		}
+
+		for (; attno <= natts; attno++)
+		{
+			/* taking opcintype from giststate->tupdesc */
+			TupleDescInitEntry(so->giststate->fetchTupdesc, attno, NULL,
+					TupleDescAttr(so->giststate->tupdesc, attno - 1)->atttypid,
+							   -1, 0);
+		}
 		scan->xs_hitupdesc = so->giststate->fetchTupdesc;
 
 		/* Also create a memory context that will hold the returned tuples */
diff --git a/src/backend/access/gist/gistsplit.c b/src/backend/access/gist/gistsplit.c
index f210e0c39f..cf3af2821a 100644
--- a/src/backend/access/gist/gistsplit.c
+++ b/src/backend/access/gist/gistsplit.c
@@ -207,7 +207,7 @@ placeOne(Relation r, GISTSTATE *giststate, GistSplitVector *v,
 	gistDeCompressAtt(giststate, r, itup, NULL, (OffsetNumber) 0,
 					  identry, isnull);
 
-	for (; attno < giststate->tupdesc->natts; attno++)
+	for (; attno < giststate->truncTupdesc->natts; attno++)
 	{
 		float		lpenalty,
 					rpenalty;
@@ -485,7 +485,7 @@ gistUserPicksplit(Relation r, GistEntryVector *entryvec, int attno, GistSplitVec
 	 */
 	v->spl_dontcare = NULL;
 
-	if (attno + 1 < giststate->tupdesc->natts)
+	if (attno + 1 < giststate->truncTupdesc->natts)
 	{
 		int			NumDontCare;
 
@@ -657,7 +657,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 		 */
 		v->spl_risnull[attno] = v->spl_lisnull[attno] = true;
 
-		if (attno + 1 < giststate->tupdesc->natts)
+		if (attno + 1 < giststate->truncTupdesc->natts)
 			gistSplitByKey(r, page, itup, len, giststate, v, attno + 1);
 		else
 			gistSplitHalf(&v->splitVector, len);
@@ -683,7 +683,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 				v->splitVector.spl_left[v->splitVector.spl_nleft++] = i;
 
 		/* Compute union keys, unless outer recursion level will handle it */
-		if (attno == 0 && giststate->tupdesc->natts == 1)
+		if (attno == 0 && giststate->truncTupdesc->natts == 1)
 		{
 			v->spl_dontcare = NULL;
 			gistunionsubkey(giststate, itup, v);
@@ -700,7 +700,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 			 * Splitting on attno column is not optimal, so consider
 			 * redistributing don't-care tuples according to the next column
 			 */
-			Assert(attno + 1 < giststate->tupdesc->natts);
+			Assert(attno + 1 < giststate->truncTupdesc->natts);
 
 			if (v->spl_dontcare == NULL)
 			{
@@ -771,7 +771,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 	 * that PickSplit (or the special cases above) produced correct union
 	 * datums.
 	 */
-	if (attno == 0 && giststate->tupdesc->natts > 1)
+	if (attno == 0 && giststate->truncTupdesc->natts > 1)
 	{
 		v->spl_dontcare = NULL;
 		gistunionsubkey(giststate, itup, v);
diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c
index 8d3dfad27b..f7024eda12 100644
--- a/src/backend/access/gist/gistutil.c
+++ b/src/backend/access/gist/gistutil.c
@@ -160,7 +160,7 @@ gistMakeUnionItVec(GISTSTATE *giststate, IndexTuple *itvec, int len,
 
 	evec = (GistEntryVector *) palloc((len + 2) * sizeof(GISTENTRY) + GEVHDRSZ);
 
-	for (i = 0; i < giststate->tupdesc->natts; i++)
+	for (i = 0; i < giststate->truncTupdesc->natts; i++)
 	{
 		int			j;
 
@@ -296,7 +296,7 @@ gistDeCompressAtt(GISTSTATE *giststate, Relation r, IndexTuple tuple, Page p,
 {
 	int			i;
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		Datum		datum;
 
@@ -329,7 +329,7 @@ gistgetadjusted(Relation r, IndexTuple oldtup, IndexTuple addtup, GISTSTATE *gis
 	gistDeCompressAtt(giststate, r, addtup, NULL,
 					  (OffsetNumber) 0, addentries, addisnull);
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		gistMakeUnionKey(giststate, i,
 						 oldentries + i, oldisnull[i],
@@ -442,7 +442,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 		zero_penalty = true;
 
 		/* Loop over index attributes. */
-		for (j = 0; j < r->rd_att->natts; j++)
+		for (j = 0; j < IndexRelationGetNumberOfKeyAttributes(r); j++)
 		{
 			Datum		datum;
 			float		usize;
@@ -470,7 +470,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 				result = i;
 				best_penalty[j] = usize;
 
-				if (j < r->rd_att->natts - 1)
+				if (j < IndexRelationGetNumberOfKeyAttributes(r) - 1)
 					best_penalty[j + 1] = -1;
 
 				/* we have new best, so reset keep-it decision */
@@ -500,7 +500,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 		 * If we looped past the last column, and did not update "result",
 		 * then this tuple is exactly as good as the prior best tuple.
 		 */
-		if (j == r->rd_att->natts && result != i)
+		if (j == IndexRelationGetNumberOfKeyAttributes(r) && result != i)
 		{
 			if (keep_current_best == -1)
 			{
@@ -575,11 +575,13 @@ gistFormTuple(GISTSTATE *giststate, Relation r,
 	Datum		compatt[INDEX_MAX_KEYS];
 	int			i;
 	IndexTuple	res;
+	/* currently we truncate tuples iif they are on internal pages */
+	bool isTruncated = !isleaf;
 
 	/*
 	 * Call the compress method on each attribute.
 	 */
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		if (isnull[i])
 			compatt[i] = (Datum) 0;
@@ -602,7 +604,21 @@ gistFormTuple(GISTSTATE *giststate, Relation r,
 		}
 	}
 
-	res = index_form_tuple(giststate->tupdesc, compatt, isnull);
+	if (!isTruncated)
+	{
+		/*
+		 * Emplace each included attribute.
+		 */
+		for (; i < r->rd_att->natts; i++)
+		{
+			if (isnull[i])
+				compatt[i] = (Datum) 0;
+			else
+				compatt[i] = attdata[i];
+		}
+	}
+
+	res = index_form_tuple(isTruncated ? giststate->truncTupdesc : giststate->tupdesc, compatt, isnull);
 
 	/*
 	 * The offset number on tuples on internal pages is unused. For historical
@@ -644,7 +660,7 @@ gistFetchTuple(GISTSTATE *giststate, Relation r, IndexTuple tuple)
 	bool		isnull[INDEX_MAX_KEYS];
 	int			i;
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		Datum		datum;
 
@@ -679,6 +695,14 @@ gistFetchTuple(GISTSTATE *giststate, Relation r, IndexTuple tuple)
 			fetchatt[i] = (Datum) 0;
 		}
 	}
+
+	/*
+	 * Get each included attribute.
+	 */
+	for (; i < r->rd_att->natts; i++)
+	{
+		fetchatt[i] = index_getattr(tuple, i + 1, giststate->tupdesc, &isnull[i]);
+	}
 	MemoryContextSwitchTo(oldcxt);
 
 	return heap_form_tuple(giststate->fetchTupdesc, fetchatt, isnull);
diff --git a/src/include/access/gist_private.h b/src/include/access/gist_private.h
index 3698942f9d..ccf06e676c 100644
--- a/src/include/access/gist_private.h
+++ b/src/include/access/gist_private.h
@@ -79,6 +79,8 @@ typedef struct GISTSTATE
 	MemoryContext tempCxt;		/* short-term context for calling functions */
 
 	TupleDesc	tupdesc;		/* index's tuple descriptor */
+	TupleDesc	truncTupdesc;	/* truncated tuple descriptor
+								 * for internal pages */
 	TupleDesc	fetchTupdesc;	/* tuple descriptor for tuples returned in an
 								 * index-only scan */
 
diff --git a/src/test/regress/expected/amutils.out b/src/test/regress/expected/amutils.out
index 4570a39b05..d92a6d12c6 100644
--- a/src/test/regress/expected/amutils.out
+++ b/src/test/regress/expected/amutils.out
@@ -75,7 +75,7 @@ select prop,
  can_unique         | f  |       | 
  can_multi_col      | t  |       | 
  can_exclude        | t  |       | 
- can_include        | f  |       | 
+ can_include        | t  |       | 
  bogus              |    |       | 
 (19 rows)
 
@@ -159,7 +159,7 @@ select amname, prop, pg_indexam_has_property(a.oid, prop) as p
  gist   | can_unique    | f
  gist   | can_multi_col | t
  gist   | can_exclude   | t
- gist   | can_include   | f
+ gist   | can_include   | t
  gist   | bogus         | 
  hash   | can_order     | f
  hash   | can_unique    | f
diff --git a/src/test/regress/expected/index_including.out b/src/test/regress/expected/index_including.out
index 16b4be34de..464629af88 100644
--- a/src/test/regress/expected/index_including.out
+++ b/src/test/regress/expected/index_including.out
@@ -313,22 +313,20 @@ SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl' ORDER BY indexname;
 
 DROP TABLE tbl;
 /*
- * 7. Check various AMs. All but btree must fail.
+ * 7. Check various AMs. All but btree and gist must fail.
  */
 CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
 CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "brin" does not support included columns
-CREATE INDEX on tbl USING gist(c3) INCLUDE (c4);
-ERROR:  access method "gist" does not support included columns
+CREATE INDEX on tbl USING gist(c3) INCLUDE (c1, c4);
 CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
 ERROR:  access method "spgist" does not support included columns
 CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "gin" does not support included columns
 CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "hash" does not support included columns
-CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
+CREATE INDEX on tbl USING rtree(c3) INCLUDE (c1, c4);
 NOTICE:  substituting access method "gist" for obsolete method "rtree"
-ERROR:  access method "gist" does not support included columns
 CREATE INDEX on tbl USING btree(c1, c2) INCLUDE (c3, c4);
 DROP TABLE tbl;
 /*
diff --git a/src/test/regress/expected/index_including_gist.out b/src/test/regress/expected/index_including_gist.out
new file mode 100644
index 0000000000..ed9906da66
--- /dev/null
+++ b/src/test/regress/expected/index_including_gist.out
@@ -0,0 +1,166 @@
+/*
+ * 1.1. test CREATE INDEX with buffered build
+ */
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+                                  pg_get_indexdef                                  
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+ c1 | c2 | c3 |     c4      
+----+----+----+-------------
+  1 |  2 |  3 | (2,3),(1,2)
+  2 |  4 |  6 | (4,5),(2,3)
+  3 |  6 |  9 | (6,7),(3,4)
+  4 |  8 | 12 | (8,9),(4,5)
+(4 rows)
+
+SET enable_bitmapscan TO off;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                   QUERY PLAN                   
+------------------------------------------------
+ Index Only Scan using tbl_gist_idx on tbl_gist
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+/*
+ * 1.2. test CREATE INDEX with inserts
+ */
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+                                  pg_get_indexdef                                  
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+ c1 | c2 | c3 |     c4      
+----+----+----+-------------
+  1 |  2 |  3 | (2,3),(1,2)
+  2 |  4 |  6 | (4,5),(2,3)
+  3 |  6 |  9 | (6,7),(3,4)
+  4 |  8 | 12 | (8,9),(4,5)
+(4 rows)
+
+SET enable_bitmapscan TO off;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                   QUERY PLAN                   
+------------------------------------------------
+ Index Only Scan using tbl_gist_idx on tbl_gist
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+/*
+ * 2. CREATE INDEX CONCURRENTLY
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX CONCURRENTLY tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                     indexdef                                      
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+DROP TABLE tbl_gist;
+/*
+ * 3. REINDEX
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                   indexdef                                    
+-------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c3)
+(1 row)
+
+REINDEX INDEX tbl_gist_idx;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                   indexdef                                    
+-------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c3)
+(1 row)
+
+ALTER TABLE tbl_gist DROP COLUMN c1;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+ indexdef 
+----------
+(0 rows)
+
+DROP TABLE tbl_gist;
+/*
+ * 4. Update, delete values in indexed table.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+UPDATE tbl_gist SET c1 = 100 WHERE c1 = 2;
+UPDATE tbl_gist SET c1 = 1 WHERE c1 = 3;
+DELETE FROM tbl_gist WHERE c1 = 5 OR c3 = 12;
+DROP TABLE tbl_gist;
+/*
+ * 5. Alter column type.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+ALTER TABLE tbl_gist ALTER c1 TYPE bigint;
+ALTER TABLE tbl_gist ALTER c3 TYPE bigint;
+\d tbl_gist
+              Table "public.tbl_gist"
+ Column |  Type   | Collation | Nullable | Default 
+--------+---------+-----------+----------+---------
+ c1     | bigint  |           |          | 
+ c2     | integer |           |          | 
+ c3     | bigint  |           |          | 
+ c4     | box     |           |          | 
+Indexes:
+    "tbl_gist_idx" gist (c4) INCLUDE (c1, c3)
+
+DROP TABLE tbl_gist;
+/*
+ * 6. EXCLUDE constraint.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box, EXCLUDE USING gist (c4 WITH &&) INCLUDE (c1, c2, c3));
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+ERROR:  conflicting key value violates exclusion constraint "tbl_gist_c4_c1_c2_c3_excl"
+DETAIL:  Key (c4)=((4,5),(2,3)) conflicts with existing key (c4)=((2,3),(1,2)).
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(3*x,2*x),point(3*x+1,2*x+1)) FROM generate_series(1,10) AS x;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                         QUERY PLAN                          
+-------------------------------------------------------------
+ Index Only Scan using tbl_gist_c4_c1_c2_c3_excl on tbl_gist
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+\d tbl_gist
+              Table "public.tbl_gist"
+ Column |  Type   | Collation | Nullable | Default 
+--------+---------+-----------+----------+---------
+ c1     | integer |           |          | 
+ c2     | integer |           |          | 
+ c3     | integer |           |          | 
+ c4     | box     |           |          | 
+Indexes:
+    "tbl_gist_c4_c1_c2_c3_excl" EXCLUDE USING gist (c4 WITH &&) INCLUDE (c1, c2, c3)
+
+DROP TABLE tbl_gist;
diff --git a/src/test/regress/parallel_schedule b/src/test/regress/parallel_schedule
index cc0bbf5db9..5684291ab0 100644
--- a/src/test/regress/parallel_schedule
+++ b/src/test/regress/parallel_schedule
@@ -55,7 +55,7 @@ test: copy copyselect copydml
 # ----------
 test: create_misc create_operator create_procedure
 # These depend on the above two
-test: create_index create_view index_including
+test: create_index create_view index_including index_including_gist
 
 # ----------
 # Another group of parallel tests
diff --git a/src/test/regress/serial_schedule b/src/test/regress/serial_schedule
index 0c10c7100c..309ccec81a 100644
--- a/src/test/regress/serial_schedule
+++ b/src/test/regress/serial_schedule
@@ -63,6 +63,7 @@ test: create_operator
 test: create_procedure
 test: create_index
 test: index_including
+test: index_including_gist
 test: create_view
 test: create_aggregate
 test: create_function_3
diff --git a/src/test/regress/sql/index_including.sql b/src/test/regress/sql/index_including.sql
index ef5fd882f5..3f9b9b51d7 100644
--- a/src/test/regress/sql/index_including.sql
+++ b/src/test/regress/sql/index_including.sql
@@ -173,15 +173,15 @@ SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl' ORDER BY indexname;
 DROP TABLE tbl;
 
 /*
- * 7. Check various AMs. All but btree must fail.
+ * 7. Check various AMs. All but btree and gist must fail.
  */
 CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
 CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
-CREATE INDEX on tbl USING gist(c3) INCLUDE (c4);
+CREATE INDEX on tbl USING gist(c3) INCLUDE (c1, c4);
 CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
 CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
 CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
-CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
+CREATE INDEX on tbl USING rtree(c3) INCLUDE (c1, c4);
 CREATE INDEX on tbl USING btree(c1, c2) INCLUDE (c3, c4);
 DROP TABLE tbl;
 
diff --git a/src/test/regress/sql/index_including_gist.sql b/src/test/regress/sql/index_including_gist.sql
new file mode 100644
index 0000000000..7d5c99b2e7
--- /dev/null
+++ b/src/test/regress/sql/index_including_gist.sql
@@ -0,0 +1,90 @@
+/*
+ * 1.1. test CREATE INDEX with buffered build
+ */
+
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO off;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+
+/*
+ * 1.2. test CREATE INDEX with inserts
+ */
+
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO off;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+
+/*
+ * 2. CREATE INDEX CONCURRENTLY
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX CONCURRENTLY tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+DROP TABLE tbl_gist;
+
+
+/*
+ * 3. REINDEX
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+REINDEX INDEX tbl_gist_idx;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+ALTER TABLE tbl_gist DROP COLUMN c1;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+DROP TABLE tbl_gist;
+
+/*
+ * 4. Update, delete values in indexed table.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+UPDATE tbl_gist SET c1 = 100 WHERE c1 = 2;
+UPDATE tbl_gist SET c1 = 1 WHERE c1 = 3;
+DELETE FROM tbl_gist WHERE c1 = 5 OR c3 = 12;
+DROP TABLE tbl_gist;
+
+/*
+ * 5. Alter column type.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+ALTER TABLE tbl_gist ALTER c1 TYPE bigint;
+ALTER TABLE tbl_gist ALTER c3 TYPE bigint;
+\d tbl_gist
+DROP TABLE tbl_gist;
+
+/*
+ * 6. EXCLUDE constraint.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box, EXCLUDE USING gist (c4 WITH &&) INCLUDE (c1, c2, c3));
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(3*x,2*x),point(3*x+1,2*x+1)) FROM generate_series(1,10) AS x;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+\d tbl_gist
+DROP TABLE tbl_gist;
-- 
2.11.0

#26

a.korotkov@postgrespro.ru

almost 7 years ago

In reply to: Andreas Karlsson (#25)

1 attachment(s)

Re: Covering GiST indexes

On Wed, Jan 30, 2019 at 4:16 AM Andreas Karlsson <andreas@proxel.se> wrote:

On 29/01/2019 19.58, Andrey Borodin wrote:

PFA patch without redundant checks.

I hope it is ok that I fixed a typo and some wording in the comment,
plus re-added the btree query which I accidentally removed in my last mail.

I think the current state of the patch is fine, assuming the solution
with truncTupdesc is deemed to be good enough by the committer, so I
will mark this as ready for committer.

Thanks for the patch and the speedy fixes!

I made some adjustments for this patch:
* Rename tupledesc and tructTupledesc to leafTupdesc and
nonLeafTupdesc correspondingly. I think this makes things more clear.
* Some improvements to docs and comments.
* Revise commit message.

I'm going to push this on no objections.

------
Alexander Korotkov
Postgres Professional: http://www.postgrespro.com
The Russian Postgres Company

Attachments:

0001-Covering-GiST-v11.patchapplication/octet-stream; name=0001-Covering-GiST-v11.patchDownload

commit 626d144227848d8532d50c8984a9a73f861633af
Author: Alexander Korotkov <akorotkov@postgresql.org>
Date:   Fri Mar 8 11:28:03 2019 +0300

    Support for INCLUDE attributes in GiST indexes
    
    Similarly to B-tree, GiST index access method gets support of INCLUDE
    attributes.  These attributes aren't used for tree navigation and aren't
    present in non-leaf pages.  But they are present in leaf pages and can be
    fetched during index-only scan.
    
    The point of having INCLUDE attributes in GiST indexes is slightly different
    from the point of having them in B-tree.  The main point of INCLUDE attributes
    in B-tree is to define UNIQUE constraint over part of attributes enabled for
    index-only scan.  In GiST the main point of INCLUDE attributes is to use
    index-only scan for attributes, whose data types don't have GiST opclasses.
    
    Discussion: https://postgr.es/m/73A1A452-AD5F-40D4-BD61-978622FF75C1%40yandex-team.ru
    Author: Andrey Borodin
    Reviewed-by: Andreas Karlsson

diff --git a/doc/src/sgml/ref/create_index.sgml b/doc/src/sgml/ref/create_index.sgml
index ad619cdcfe4..d8f018f4dac 100644
--- a/doc/src/sgml/ref/create_index.sgml
+++ b/doc/src/sgml/ref/create_index.sgml
@@ -181,10 +181,10 @@ CREATE [ UNIQUE ] INDEX [ CONCURRENTLY ] [ [ IF NOT EXISTS ] <replaceable class=
        </para>
 
        <para>
-        Currently, only the B-tree index access method supports this feature.
-        In B-tree indexes, the values of columns listed in the
-        <literal>INCLUDE</literal> clause are included in leaf tuples which
-        correspond to heap tuples, but are not included in upper-level
+        Currently, the B-tree and the GiST index access methods supports this
+        feature.  In B-tree and the GiST indexes, the values of columns listed
+        in the <literal>INCLUDE</literal> clause are included in leaf tuples
+        which correspond to heap tuples, but are not included in upper-level
         index entries used for tree navigation.
        </para>
       </listitem>
diff --git a/doc/src/sgml/textsearch.sgml b/doc/src/sgml/textsearch.sgml
index ecebade767d..3281f7cd33a 100644
--- a/doc/src/sgml/textsearch.sgml
+++ b/doc/src/sgml/textsearch.sgml
@@ -3674,6 +3674,12 @@ SELECT plainto_tsquery('supernovae stars');
    retrieved to see if the match is correct.
   </para>
 
+  <para>
+   A GiST index can be covering, i.e. use the <literal>INCLUDE</literal>
+   clause.  Included columns can have data types without any GiST operator
+   class.  Included attributes will be stored uncompressed.
+  </para>
+
   <para>
    Lossiness causes performance degradation due to unnecessary fetches of table
    records that turn out to be false matches.  Since random access to table
diff --git a/src/backend/access/gist/gist.c b/src/backend/access/gist/gist.c
index 5ea774661a9..2ce5425ef98 100644
--- a/src/backend/access/gist/gist.c
+++ b/src/backend/access/gist/gist.c
@@ -75,7 +75,7 @@ gisthandler(PG_FUNCTION_ARGS)
 	amroutine->amclusterable = true;
 	amroutine->ampredlocks = true;
 	amroutine->amcanparallel = false;
-	amroutine->amcaninclude = false;
+	amroutine->amcaninclude = true;
 	amroutine->amkeytype = InvalidOid;
 
 	amroutine->ambuild = gistbuild;
@@ -1382,8 +1382,10 @@ gistSplit(Relation r,
 						IndexTupleSize(itup[0]), GiSTPageSize,
 						RelationGetRelationName(r))));
 
-	memset(v.spl_lisnull, true, sizeof(bool) * giststate->tupdesc->natts);
-	memset(v.spl_risnull, true, sizeof(bool) * giststate->tupdesc->natts);
+	memset(v.spl_lisnull, true,
+		   sizeof(bool) * giststate->nonLeafTupdesc->natts);
+	memset(v.spl_risnull, true,
+		   sizeof(bool) * giststate->nonLeafTupdesc->natts);
 	gistSplitByKey(r, page, itup, len, giststate, &v, 0);
 
 	/* form left and right vector */
@@ -1461,9 +1463,23 @@ initGISTstate(Relation index)
 
 	giststate->scanCxt = scanCxt;
 	giststate->tempCxt = scanCxt;	/* caller must change this if needed */
-	giststate->tupdesc = index->rd_att;
+	giststate->leafTupdesc = index->rd_att;
 
-	for (i = 0; i < index->rd_att->natts; i++)
+	/*
+	 * The truncated tupdesc for non-leaf index tuples, which doesn't contain
+	 * the INCLUDE attributes.
+	 *
+	 * It is used to form tuples during tuple adjustement and page split.
+	 * B-tree creates shortened tuple descriptor for every truncated tuple,
+	 * because it is doing this less often: it does not have to form truncated
+	 * tuples during page split.  Also, B-tree is not adjusting tuples on
+	 * internal pages the way GiST does.
+	 */
+	giststate->nonLeafTupdesc = CreateTupleDescCopyConstr(index->rd_att);
+	giststate->nonLeafTupdesc->natts =
+		IndexRelationGetNumberOfKeyAttributes(index);
+
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(index); i++)
 	{
 		fmgr_info_copy(&(giststate->consistentFn[i]),
 					   index_getprocinfo(index, i + 1, GIST_CONSISTENT_PROC),
@@ -1531,6 +1547,21 @@ initGISTstate(Relation index)
 			giststate->supportCollation[i] = DEFAULT_COLLATION_OID;
 	}
 
+	/* No opclass information for INCLUDE attributes */
+	for (; i < index->rd_att->natts; i++)
+	{
+		giststate->consistentFn[i].fn_oid = InvalidOid;
+		giststate->unionFn[i].fn_oid = InvalidOid;
+		giststate->compressFn[i].fn_oid = InvalidOid;
+		giststate->decompressFn[i].fn_oid = InvalidOid;
+		giststate->penaltyFn[i].fn_oid = InvalidOid;
+		giststate->picksplitFn[i].fn_oid = InvalidOid;
+		giststate->equalFn[i].fn_oid = InvalidOid;
+		giststate->distanceFn[i].fn_oid = InvalidOid;
+		giststate->fetchFn[i].fn_oid = InvalidOid;
+		giststate->supportCollation[i] = InvalidOid;
+	}
+
 	MemoryContextSwitchTo(oldCxt);
 
 	return giststate;
diff --git a/src/backend/access/gist/gistget.c b/src/backend/access/gist/gistget.c
index a96ef5c3acc..156b9d699f7 100644
--- a/src/backend/access/gist/gistget.c
+++ b/src/backend/access/gist/gistget.c
@@ -164,7 +164,7 @@ gistindex_keytest(IndexScanDesc scan,
 
 		datum = index_getattr(tuple,
 							  key->sk_attno,
-							  giststate->tupdesc,
+							  giststate->leafTupdesc,
 							  &isNull);
 
 		if (key->sk_flags & SK_ISNULL)
@@ -244,7 +244,7 @@ gistindex_keytest(IndexScanDesc scan,
 
 		datum = index_getattr(tuple,
 							  key->sk_attno,
-							  giststate->tupdesc,
+							  giststate->leafTupdesc,
 							  &isNull);
 
 		if ((key->sk_flags & SK_ISNULL) || isNull)
@@ -769,11 +769,13 @@ gistgetbitmap(IndexScanDesc scan, TIDBitmap *tbm)
  *
  * Opclasses that implement a fetch function support index-only scans.
  * Opclasses without compression functions also support index-only scans.
+ * Included attributes always can be fetched for index-only scans.
  */
 bool
 gistcanreturn(Relation index, int attno)
 {
-	if (OidIsValid(index_getprocid(index, attno, GIST_FETCH_PROC)) ||
+	if (attno > IndexRelationGetNumberOfKeyAttributes(index) ||
+		OidIsValid(index_getprocid(index, attno, GIST_FETCH_PROC)) ||
 		!OidIsValid(index_getprocid(index, attno, GIST_COMPRESS_PROC)))
 		return true;
 	else
diff --git a/src/backend/access/gist/gistscan.c b/src/backend/access/gist/gistscan.c
index 78a8ede7949..ef7e4c8aa34 100644
--- a/src/backend/access/gist/gistscan.c
+++ b/src/backend/access/gist/gistscan.c
@@ -158,6 +158,7 @@ gistrescan(IndexScanDesc scan, ScanKey key, int nkeys,
 	if (scan->xs_want_itup && !scan->xs_hitupdesc)
 	{
 		int			natts;
+		int			nkeyatts;
 		int			attno;
 
 		/*
@@ -167,13 +168,22 @@ gistrescan(IndexScanDesc scan, ScanKey key, int nkeys,
 		 * types.
 		 */
 		natts = RelationGetNumberOfAttributes(scan->indexRelation);
+		nkeyatts = IndexRelationGetNumberOfKeyAttributes(scan->indexRelation);
 		so->giststate->fetchTupdesc = CreateTemplateTupleDesc(natts);
-		for (attno = 1; attno <= natts; attno++)
+		for (attno = 1; attno <= nkeyatts; attno++)
 		{
 			TupleDescInitEntry(so->giststate->fetchTupdesc, attno, NULL,
 							   scan->indexRelation->rd_opcintype[attno - 1],
 							   -1, 0);
 		}
+
+		for (; attno <= natts; attno++)
+		{
+			/* taking opcintype from giststate->tupdesc */
+			TupleDescInitEntry(so->giststate->fetchTupdesc, attno, NULL,
+							   TupleDescAttr(so->giststate->leafTupdesc, attno - 1)->atttypid,
+							   -1, 0);
+		}
 		scan->xs_hitupdesc = so->giststate->fetchTupdesc;
 
 		/* Also create a memory context that will hold the returned tuples */
diff --git a/src/backend/access/gist/gistsplit.c b/src/backend/access/gist/gistsplit.c
index f210e0c39f4..6a9c54d86ce 100644
--- a/src/backend/access/gist/gistsplit.c
+++ b/src/backend/access/gist/gistsplit.c
@@ -207,7 +207,7 @@ placeOne(Relation r, GISTSTATE *giststate, GistSplitVector *v,
 	gistDeCompressAtt(giststate, r, itup, NULL, (OffsetNumber) 0,
 					  identry, isnull);
 
-	for (; attno < giststate->tupdesc->natts; attno++)
+	for (; attno < giststate->nonLeafTupdesc->natts; attno++)
 	{
 		float		lpenalty,
 					rpenalty;
@@ -485,7 +485,7 @@ gistUserPicksplit(Relation r, GistEntryVector *entryvec, int attno, GistSplitVec
 	 */
 	v->spl_dontcare = NULL;
 
-	if (attno + 1 < giststate->tupdesc->natts)
+	if (attno + 1 < giststate->nonLeafTupdesc->natts)
 	{
 		int			NumDontCare;
 
@@ -639,7 +639,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 		Datum		datum;
 		bool		IsNull;
 
-		datum = index_getattr(itup[i - 1], attno + 1, giststate->tupdesc,
+		datum = index_getattr(itup[i - 1], attno + 1, giststate->leafTupdesc,
 							  &IsNull);
 		gistdentryinit(giststate, attno, &(entryvec->vector[i]),
 					   datum, r, page, i,
@@ -657,7 +657,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 		 */
 		v->spl_risnull[attno] = v->spl_lisnull[attno] = true;
 
-		if (attno + 1 < giststate->tupdesc->natts)
+		if (attno + 1 < giststate->nonLeafTupdesc->natts)
 			gistSplitByKey(r, page, itup, len, giststate, v, attno + 1);
 		else
 			gistSplitHalf(&v->splitVector, len);
@@ -683,7 +683,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 				v->splitVector.spl_left[v->splitVector.spl_nleft++] = i;
 
 		/* Compute union keys, unless outer recursion level will handle it */
-		if (attno == 0 && giststate->tupdesc->natts == 1)
+		if (attno == 0 && giststate->nonLeafTupdesc->natts == 1)
 		{
 			v->spl_dontcare = NULL;
 			gistunionsubkey(giststate, itup, v);
@@ -700,7 +700,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 			 * Splitting on attno column is not optimal, so consider
 			 * redistributing don't-care tuples according to the next column
 			 */
-			Assert(attno + 1 < giststate->tupdesc->natts);
+			Assert(attno + 1 < giststate->nonLeafTupdesc->natts);
 
 			if (v->spl_dontcare == NULL)
 			{
@@ -771,7 +771,7 @@ gistSplitByKey(Relation r, Page page, IndexTuple *itup, int len,
 	 * that PickSplit (or the special cases above) produced correct union
 	 * datums.
 	 */
-	if (attno == 0 && giststate->tupdesc->natts > 1)
+	if (attno == 0 && giststate->nonLeafTupdesc->natts > 1)
 	{
 		v->spl_dontcare = NULL;
 		gistunionsubkey(giststate, itup, v);
diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c
index 8d3dfad27bd..a1415f6842d 100644
--- a/src/backend/access/gist/gistutil.c
+++ b/src/backend/access/gist/gistutil.c
@@ -160,7 +160,7 @@ gistMakeUnionItVec(GISTSTATE *giststate, IndexTuple *itvec, int len,
 
 	evec = (GistEntryVector *) palloc((len + 2) * sizeof(GISTENTRY) + GEVHDRSZ);
 
-	for (i = 0; i < giststate->tupdesc->natts; i++)
+	for (i = 0; i < giststate->nonLeafTupdesc->natts; i++)
 	{
 		int			j;
 
@@ -171,7 +171,8 @@ gistMakeUnionItVec(GISTSTATE *giststate, IndexTuple *itvec, int len,
 			Datum		datum;
 			bool		IsNull;
 
-			datum = index_getattr(itvec[j], i + 1, giststate->tupdesc, &IsNull);
+			datum = index_getattr(itvec[j], i + 1, giststate->leafTupdesc,
+								  &IsNull);
 			if (IsNull)
 				continue;
 
@@ -296,11 +297,11 @@ gistDeCompressAtt(GISTSTATE *giststate, Relation r, IndexTuple tuple, Page p,
 {
 	int			i;
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		Datum		datum;
 
-		datum = index_getattr(tuple, i + 1, giststate->tupdesc, &isnull[i]);
+		datum = index_getattr(tuple, i + 1, giststate->leafTupdesc, &isnull[i]);
 		gistdentryinit(giststate, i, &attdata[i],
 					   datum, r, p, o,
 					   false, isnull[i]);
@@ -329,7 +330,7 @@ gistgetadjusted(Relation r, IndexTuple oldtup, IndexTuple addtup, GISTSTATE *gis
 	gistDeCompressAtt(giststate, r, addtup, NULL,
 					  (OffsetNumber) 0, addentries, addisnull);
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		gistMakeUnionKey(giststate, i,
 						 oldentries + i, oldisnull[i],
@@ -442,14 +443,15 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 		zero_penalty = true;
 
 		/* Loop over index attributes. */
-		for (j = 0; j < r->rd_att->natts; j++)
+		for (j = 0; j < IndexRelationGetNumberOfKeyAttributes(r); j++)
 		{
 			Datum		datum;
 			float		usize;
 			bool		IsNull;
 
 			/* Compute penalty for this column. */
-			datum = index_getattr(itup, j + 1, giststate->tupdesc, &IsNull);
+			datum = index_getattr(itup, j + 1, giststate->leafTupdesc,
+								  &IsNull);
 			gistdentryinit(giststate, j, &entry, datum, r, p, i,
 						   false, IsNull);
 			usize = gistpenalty(giststate, j, &entry, IsNull,
@@ -470,7 +472,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 				result = i;
 				best_penalty[j] = usize;
 
-				if (j < r->rd_att->natts - 1)
+				if (j < IndexRelationGetNumberOfKeyAttributes(r) - 1)
 					best_penalty[j + 1] = -1;
 
 				/* we have new best, so reset keep-it decision */
@@ -500,7 +502,7 @@ gistchoose(Relation r, Page p, IndexTuple it,	/* it has compressed entry */
 		 * If we looped past the last column, and did not update "result",
 		 * then this tuple is exactly as good as the prior best tuple.
 		 */
-		if (j == r->rd_att->natts && result != i)
+		if (j == IndexRelationGetNumberOfKeyAttributes(r) && result != i)
 		{
 			if (keep_current_best == -1)
 			{
@@ -576,10 +578,13 @@ gistFormTuple(GISTSTATE *giststate, Relation r,
 	int			i;
 	IndexTuple	res;
 
+	/* currently we truncate tuples iif they are on internal pages */
+	bool		isTruncated = !isleaf;
+
 	/*
 	 * Call the compress method on each attribute.
 	 */
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		if (isnull[i])
 			compatt[i] = (Datum) 0;
@@ -602,7 +607,23 @@ gistFormTuple(GISTSTATE *giststate, Relation r,
 		}
 	}
 
-	res = index_form_tuple(giststate->tupdesc, compatt, isnull);
+	if (!isTruncated)
+	{
+		/*
+		 * Emplace each included attribute.
+		 */
+		for (; i < r->rd_att->natts; i++)
+		{
+			if (isnull[i])
+				compatt[i] = (Datum) 0;
+			else
+				compatt[i] = attdata[i];
+		}
+	}
+
+	res = index_form_tuple(isTruncated ? giststate->nonLeafTupdesc :
+						   giststate->leafTupdesc,
+						   compatt, isnull);
 
 	/*
 	 * The offset number on tuples on internal pages is unused. For historical
@@ -644,11 +665,11 @@ gistFetchTuple(GISTSTATE *giststate, Relation r, IndexTuple tuple)
 	bool		isnull[INDEX_MAX_KEYS];
 	int			i;
 
-	for (i = 0; i < r->rd_att->natts; i++)
+	for (i = 0; i < IndexRelationGetNumberOfKeyAttributes(r); i++)
 	{
 		Datum		datum;
 
-		datum = index_getattr(tuple, i + 1, giststate->tupdesc, &isnull[i]);
+		datum = index_getattr(tuple, i + 1, giststate->leafTupdesc, &isnull[i]);
 
 		if (giststate->fetchFn[i].fn_oid != InvalidOid)
 		{
@@ -679,6 +700,15 @@ gistFetchTuple(GISTSTATE *giststate, Relation r, IndexTuple tuple)
 			fetchatt[i] = (Datum) 0;
 		}
 	}
+
+	/*
+	 * Get each included attribute.
+	 */
+	for (; i < r->rd_att->natts; i++)
+	{
+		fetchatt[i] = index_getattr(tuple, i + 1, giststate->leafTupdesc,
+									&isnull[i]);
+	}
 	MemoryContextSwitchTo(oldcxt);
 
 	return heap_form_tuple(giststate->fetchTupdesc, fetchatt, isnull);
diff --git a/src/include/access/gist_private.h b/src/include/access/gist_private.h
index 3698942f9da..463d2bfc7b9 100644
--- a/src/include/access/gist_private.h
+++ b/src/include/access/gist_private.h
@@ -78,7 +78,9 @@ typedef struct GISTSTATE
 	MemoryContext scanCxt;		/* context for scan-lifespan data */
 	MemoryContext tempCxt;		/* short-term context for calling functions */
 
-	TupleDesc	tupdesc;		/* index's tuple descriptor */
+	TupleDesc	leafTupdesc;	/* index's tuple descriptor */
+	TupleDesc	nonLeafTupdesc; /* truncated tuple descriptor for non-leaf
+								 * pages */
 	TupleDesc	fetchTupdesc;	/* tuple descriptor for tuples returned in an
 								 * index-only scan */
 
diff --git a/src/test/regress/expected/amutils.out b/src/test/regress/expected/amutils.out
index 4570a39b058..d92a6d12c62 100644
--- a/src/test/regress/expected/amutils.out
+++ b/src/test/regress/expected/amutils.out
@@ -75,7 +75,7 @@ select prop,
  can_unique         | f  |       | 
  can_multi_col      | t  |       | 
  can_exclude        | t  |       | 
- can_include        | f  |       | 
+ can_include        | t  |       | 
  bogus              |    |       | 
 (19 rows)
 
@@ -159,7 +159,7 @@ select amname, prop, pg_indexam_has_property(a.oid, prop) as p
  gist   | can_unique    | f
  gist   | can_multi_col | t
  gist   | can_exclude   | t
- gist   | can_include   | f
+ gist   | can_include   | t
  gist   | bogus         | 
  hash   | can_order     | f
  hash   | can_unique    | f
diff --git a/src/test/regress/expected/index_including.out b/src/test/regress/expected/index_including.out
index f86e5953fac..77ec29f2a33 100644
--- a/src/test/regress/expected/index_including.out
+++ b/src/test/regress/expected/index_including.out
@@ -330,22 +330,20 @@ SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl' ORDER BY indexname;
 
 DROP TABLE tbl;
 /*
- * 7. Check various AMs. All but btree must fail.
+ * 7. Check various AMs. All but btree and gist must fail.
  */
 CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
 CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "brin" does not support included columns
-CREATE INDEX on tbl USING gist(c3) INCLUDE (c4);
-ERROR:  access method "gist" does not support included columns
+CREATE INDEX on tbl USING gist(c3) INCLUDE (c1, c4);
 CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
 ERROR:  access method "spgist" does not support included columns
 CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "gin" does not support included columns
 CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
 ERROR:  access method "hash" does not support included columns
-CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
+CREATE INDEX on tbl USING rtree(c3) INCLUDE (c1, c4);
 NOTICE:  substituting access method "gist" for obsolete method "rtree"
-ERROR:  access method "gist" does not support included columns
 CREATE INDEX on tbl USING btree(c1, c2) INCLUDE (c3, c4);
 DROP TABLE tbl;
 /*
diff --git a/src/test/regress/expected/index_including_gist.out b/src/test/regress/expected/index_including_gist.out
new file mode 100644
index 00000000000..ed9906da669
--- /dev/null
+++ b/src/test/regress/expected/index_including_gist.out
@@ -0,0 +1,166 @@
+/*
+ * 1.1. test CREATE INDEX with buffered build
+ */
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+                                  pg_get_indexdef                                  
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+ c1 | c2 | c3 |     c4      
+----+----+----+-------------
+  1 |  2 |  3 | (2,3),(1,2)
+  2 |  4 |  6 | (4,5),(2,3)
+  3 |  6 |  9 | (6,7),(3,4)
+  4 |  8 | 12 | (8,9),(4,5)
+(4 rows)
+
+SET enable_bitmapscan TO off;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                   QUERY PLAN                   
+------------------------------------------------
+ Index Only Scan using tbl_gist_idx on tbl_gist
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+/*
+ * 1.2. test CREATE INDEX with inserts
+ */
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+                                  pg_get_indexdef                                  
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+ c1 | c2 | c3 |     c4      
+----+----+----+-------------
+  1 |  2 |  3 | (2,3),(1,2)
+  2 |  4 |  6 | (4,5),(2,3)
+  3 |  6 |  9 | (6,7),(3,4)
+  4 |  8 | 12 | (8,9),(4,5)
+(4 rows)
+
+SET enable_bitmapscan TO off;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                   QUERY PLAN                   
+------------------------------------------------
+ Index Only Scan using tbl_gist_idx on tbl_gist
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+/*
+ * 2. CREATE INDEX CONCURRENTLY
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX CONCURRENTLY tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                     indexdef                                      
+-----------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c2, c3)
+(1 row)
+
+DROP TABLE tbl_gist;
+/*
+ * 3. REINDEX
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                   indexdef                                    
+-------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c3)
+(1 row)
+
+REINDEX INDEX tbl_gist_idx;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+                                   indexdef                                    
+-------------------------------------------------------------------------------
+ CREATE INDEX tbl_gist_idx ON public.tbl_gist USING gist (c4) INCLUDE (c1, c3)
+(1 row)
+
+ALTER TABLE tbl_gist DROP COLUMN c1;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+ indexdef 
+----------
+(0 rows)
+
+DROP TABLE tbl_gist;
+/*
+ * 4. Update, delete values in indexed table.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+UPDATE tbl_gist SET c1 = 100 WHERE c1 = 2;
+UPDATE tbl_gist SET c1 = 1 WHERE c1 = 3;
+DELETE FROM tbl_gist WHERE c1 = 5 OR c3 = 12;
+DROP TABLE tbl_gist;
+/*
+ * 5. Alter column type.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+ALTER TABLE tbl_gist ALTER c1 TYPE bigint;
+ALTER TABLE tbl_gist ALTER c3 TYPE bigint;
+\d tbl_gist
+              Table "public.tbl_gist"
+ Column |  Type   | Collation | Nullable | Default 
+--------+---------+-----------+----------+---------
+ c1     | bigint  |           |          | 
+ c2     | integer |           |          | 
+ c3     | bigint  |           |          | 
+ c4     | box     |           |          | 
+Indexes:
+    "tbl_gist_idx" gist (c4) INCLUDE (c1, c3)
+
+DROP TABLE tbl_gist;
+/*
+ * 6. EXCLUDE constraint.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box, EXCLUDE USING gist (c4 WITH &&) INCLUDE (c1, c2, c3));
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+ERROR:  conflicting key value violates exclusion constraint "tbl_gist_c4_c1_c2_c3_excl"
+DETAIL:  Key (c4)=((4,5),(2,3)) conflicts with existing key (c4)=((2,3),(1,2)).
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(3*x,2*x),point(3*x+1,2*x+1)) FROM generate_series(1,10) AS x;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+                         QUERY PLAN                          
+-------------------------------------------------------------
+ Index Only Scan using tbl_gist_c4_c1_c2_c3_excl on tbl_gist
+   Index Cond: (c4 <@ '(10,10),(1,1)'::box)
+(2 rows)
+
+\d tbl_gist
+              Table "public.tbl_gist"
+ Column |  Type   | Collation | Nullable | Default 
+--------+---------+-----------+----------+---------
+ c1     | integer |           |          | 
+ c2     | integer |           |          | 
+ c3     | integer |           |          | 
+ c4     | box     |           |          | 
+Indexes:
+    "tbl_gist_c4_c1_c2_c3_excl" EXCLUDE USING gist (c4 WITH &&) INCLUDE (c1, c2, c3)
+
+DROP TABLE tbl_gist;
diff --git a/src/test/regress/parallel_schedule b/src/test/regress/parallel_schedule
index 4051a4ad4e1..ace703179e8 100644
--- a/src/test/regress/parallel_schedule
+++ b/src/test/regress/parallel_schedule
@@ -55,7 +55,7 @@ test: copy copyselect copydml
 # ----------
 test: create_misc create_operator create_procedure
 # These depend on the above two
-test: create_index create_view index_including
+test: create_index create_view index_including index_including_gist
 
 # ----------
 # Another group of parallel tests
diff --git a/src/test/regress/serial_schedule b/src/test/regress/serial_schedule
index ac1ea622d65..309a907ece8 100644
--- a/src/test/regress/serial_schedule
+++ b/src/test/regress/serial_schedule
@@ -63,6 +63,7 @@ test: create_operator
 test: create_procedure
 test: create_index
 test: index_including
+test: index_including_gist
 test: create_view
 test: create_aggregate
 test: create_function_3
diff --git a/src/test/regress/sql/index_including.sql b/src/test/regress/sql/index_including.sql
index 2c6dd6211e6..c0ae71d0cbd 100644
--- a/src/test/regress/sql/index_including.sql
+++ b/src/test/regress/sql/index_including.sql
@@ -176,15 +176,15 @@ SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl' ORDER BY indexname;
 DROP TABLE tbl;
 
 /*
- * 7. Check various AMs. All but btree must fail.
+ * 7. Check various AMs. All but btree and gist must fail.
  */
 CREATE TABLE tbl (c1 int,c2 int, c3 box, c4 box);
 CREATE INDEX on tbl USING brin(c1, c2) INCLUDE (c3, c4);
-CREATE INDEX on tbl USING gist(c3) INCLUDE (c4);
+CREATE INDEX on tbl USING gist(c3) INCLUDE (c1, c4);
 CREATE INDEX on tbl USING spgist(c3) INCLUDE (c4);
 CREATE INDEX on tbl USING gin(c1, c2) INCLUDE (c3, c4);
 CREATE INDEX on tbl USING hash(c1, c2) INCLUDE (c3, c4);
-CREATE INDEX on tbl USING rtree(c1, c2) INCLUDE (c3, c4);
+CREATE INDEX on tbl USING rtree(c3) INCLUDE (c1, c4);
 CREATE INDEX on tbl USING btree(c1, c2) INCLUDE (c3, c4);
 DROP TABLE tbl;
 
diff --git a/src/test/regress/sql/index_including_gist.sql b/src/test/regress/sql/index_including_gist.sql
new file mode 100644
index 00000000000..7d5c99b2e79
--- /dev/null
+++ b/src/test/regress/sql/index_including_gist.sql
@@ -0,0 +1,90 @@
+/*
+ * 1.1. test CREATE INDEX with buffered build
+ */
+
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO off;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+
+/*
+ * 1.2. test CREATE INDEX with inserts
+ */
+
+-- Regular index with included columns
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+-- size is chosen to exceed page size and trigger actual truncation
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,8000) AS x;
+SELECT pg_get_indexdef(i.indexrelid)
+FROM pg_index i JOIN pg_class c ON i.indexrelid = c.oid
+WHERE i.indrelid = 'tbl_gist'::regclass ORDER BY c.relname;
+SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO off;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+SET enable_bitmapscan TO default;
+DROP TABLE tbl_gist;
+
+/*
+ * 2. CREATE INDEX CONCURRENTLY
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX CONCURRENTLY tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c2,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+DROP TABLE tbl_gist;
+
+
+/*
+ * 3. REINDEX
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+REINDEX INDEX tbl_gist_idx;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+ALTER TABLE tbl_gist DROP COLUMN c1;
+SELECT indexdef FROM pg_indexes WHERE tablename = 'tbl_gist' ORDER BY indexname;
+DROP TABLE tbl_gist;
+
+/*
+ * 4. Update, delete values in indexed table.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+UPDATE tbl_gist SET c1 = 100 WHERE c1 = 2;
+UPDATE tbl_gist SET c1 = 1 WHERE c1 = 3;
+DELETE FROM tbl_gist WHERE c1 = 5 OR c3 = 12;
+DROP TABLE tbl_gist;
+
+/*
+ * 5. Alter column type.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box);
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+CREATE INDEX tbl_gist_idx ON tbl_gist using gist (c4) INCLUDE (c1,c3);
+ALTER TABLE tbl_gist ALTER c1 TYPE bigint;
+ALTER TABLE tbl_gist ALTER c3 TYPE bigint;
+\d tbl_gist
+DROP TABLE tbl_gist;
+
+/*
+ * 6. EXCLUDE constraint.
+ */
+CREATE TABLE tbl_gist (c1 int, c2 int, c3 int, c4 box, EXCLUDE USING gist (c4 WITH &&) INCLUDE (c1, c2, c3));
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(x,x+1),point(2*x,2*x+1)) FROM generate_series(1,10) AS x;
+INSERT INTO tbl_gist SELECT x, 2*x, 3*x, box(point(3*x,2*x),point(3*x+1,2*x+1)) FROM generate_series(1,10) AS x;
+EXPLAIN  (costs off) SELECT * FROM tbl_gist where c4 <@ box(point(1,1),point(10,10));
+\d tbl_gist
+DROP TABLE tbl_gist;

#27

a.korotkov@postgrespro.ru

almost 7 years ago

In reply to: Alexander Korotkov (#26)

Re: Covering GiST indexes

On Fri, Mar 8, 2019 at 11:58 AM Alexander Korotkov
<a.korotkov@postgrespro.ru> wrote:

I made some adjustments for this patch:
* Rename tupledesc and tructTupledesc to leafTupdesc and
nonLeafTupdesc correspondingly. I think this makes things more clear.
* Some improvements to docs and comments.
* Revise commit message.

I'm going to push this on no objections.

Pushed with some more small changes in docs.

------
Alexander Korotkov
Postgres Professional: http://www.postgrespro.com
The Russian Postgres Company

#28