Remove columns pg_index.haskeytype and pg_index.indisclustered

Bruce Momjian <pgman@candle.pha.pa.us> writes:

I have removed pg_index.haskeytype and pg_index.indisclustered with the
following patch. They were not used.

haskeytype IS used, by GIST. Kindly revert this entire patch.

regards, tom lane

Bruce Momjian <pgman@candle.pha.pa.us> writes:

I have removed pg_index.haskeytype and pg_index.indisclustered with the
following patch. They were not used.

haskeytype IS used, by GIST. Kindly revert this entire patch.

Fixing now. Thanks.

-- 
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026

Bruce Momjian <pgman@candle.pha.pa.us> writes:

I have removed pg_index.haskeytype and pg_index.indisclustered with the
following patch. They were not used.

haskeytype IS used, by GIST. Kindly revert this entire patch.

Fixing now. Thanks.

Patch applied. Field re-added. (I removed it incorrectly anyway.)

What threw me off is that there was a comment saying the field was not
used.

How does GIST set this field? Does GIST insert directly into pg_index?
I don't see this field set anywhere except to 'true'. Here are the
references I see from the 7.1.X code:

./backend/access/gist/gist.c: giststate->haskeytype = itupform->indhaskeytype;
./backend/access/gist/gist.c: if (giststate->haskeytype)
./backend/access/gist/gist.c: if (giststate->haskeytype)
./backend/access/gist/gist.c: if (giststate->haskeytype)
./backend/catalog/index.c: indexForm->indhaskeytype = true; /* not actually used anymore */
./include/access/gist.h: bool haskeytype;
./include/catalog/pg_index.h: * first variable length field. so I moved indislossy, indhaskeytype,
./include/catalog/pg_index.h: bool indhaskeytype; /* does key type != attribute type? */
./include/catalog/pg_index.h:#define Anum_pg_index_indhaskeytype 8

-- 
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026

Bruce Momjian <pgman@candle.pha.pa.us> writes:

How does GIST set this field?

I didn't say GIST *set* it, I said GIST *used* it --- and your patch
would've changed GIST's behavior.

Perhaps the field is not really necessary, but some detailed analysis
would be required to establish that. I'm not very clear on what GIST
is doing with it.

regards, tom lane

Bruce Momjian <pgman@candle.pha.pa.us> writes:

How does GIST set this field?

I didn't say GIST *set* it, I said GIST *used* it --- and your patch
would've changed GIST's behavior.

Perhaps the field is not really necessary, but some detailed analysis
would be required to establish that. I'm not very clear on what GIST
is doing with it.

Nor am I. GIST folks?

-- 
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026