Prepping for annual pgindent run
Hackorum builds and tests every patch posted to the lists, not only commitfest submissions. This is Hackorum's own CI rather than the PostgreSQL project's, and it is still under testing - please report anything that looks wrong.
You can run a PostgreSQL built from this patch straight from Docker, with no checkout and no build:
docker run --rm -p 5432:5432 ghcr.io/hackorum-dev/postgres-patch:t139559psql -h localhost -U postgresBuilt from patchset v1 (message #1), July 27, 2026 at 12:53 AM.
Every patchset is also pushed to a branch of our PostgreSQL fork, so you can check out the same tree CI built. Without a PostgreSQL checkout:
git clone --branch t139559_1 https://github.com/hackorum-dev/postgres.gitIn a checkout you already have, add the fork once:
git remote add hackorum https://github.com/hackorum-dev/postgres.gitthen, for this patchset and every later one:
git fetch hackorum t139559_1 && git checkout t139559_1Patchset v1 (message #1) is on t139559_1
I propose to do our annual pgindent run sometime pretty soon,
perhaps next week after the minor-release dust has settled.
One thing to be done is to update src/tools/pgindent/typedefs.list
from the canonical list constructed by the buildfarm. We've been
pretty good about maintaining that file manually, but not perfect,
so this has a few small effects --- see attached for what that
would look like today.
I also propose applying the pgindent patches discussed in [1]/messages/by-id/CAJ7c6TPQ0kkHQG-AqeAJ3PV_YtmDzcc7s+_V4=t+xgSnZm1cFw@mail.gmail.com and [2]/messages/by-id/c3327be8-09e2-46a1-88b4-228a339d6916@proxel.se,
which would have rather larger effects. [1]/messages/by-id/CAJ7c6TPQ0kkHQG-AqeAJ3PV_YtmDzcc7s+_V4=t+xgSnZm1cFw@mail.gmail.com improves formatting of
multiline comment blocks. 99% of the changes it would cause look like
@@ -933,7 +933,8 @@ _PG_init(void)
MarkGUCPrefixReserved("isn");
}
-/* isn_out
+/*
+ * isn_out
*/
PG_FUNCTION_INFO_V1(isn_out);
DatumWhile that's not that big a deal, it improves style uniformity since
most of our multiline comments do not have any text on the first line.
[2]: /messages/by-id/c3327be8-09e2-46a1-88b4-228a339d6916@proxel.se
which mostly affects variadic functions:
@@ -147,7 +147,7 @@ px_set_debug_handler(void (*handler) (const char *))
}
void
-px_debug(const char *fmt,...)
+px_debug(const char *fmt, ...)
{
va_list ap;
It also helps a few struct constructors:
@@ -62,7 +62,7 @@ struct RBTree
static RBTNode sentinel =
{
- .color = RBTBLACK,.left = RBTNIL,.right = RBTNIL,.parent = NULL
+ .color = RBTBLACK, .left = RBTNIL, .right = RBTNIL, .parent = NULL
};I don't see any places where it makes anything worse, and the
improvement in variadic functions is quite pleasing IMO.
So I think we should do that too.
We also have pgperltidy, renumber_oids.pl, and copyright.pl to
run sometime before beta1. Might as well do those at the same
time.
regards, tom lane
[1]: /messages/by-id/CAJ7c6TPQ0kkHQG-AqeAJ3PV_YtmDzcc7s+_V4=t+xgSnZm1cFw@mail.gmail.com
[2]: /messages/by-id/c3327be8-09e2-46a1-88b4-228a339d6916@proxel.se
On Tue, May 05, 2026 at 02:39:50PM -0400, Tom Lane wrote:
One thing to be done is to update src/tools/pgindent/typedefs.list
from the canonical list constructed by the buildfarm. We've been
pretty good about maintaining that file manually, but not perfect,
so this has a few small effects --- see attached for what that
would look like today.
Seems fine.
I also propose applying the pgindent patches discussed in [1] and [2],
which would have rather larger effects. [1] improves formatting of
multiline comment blocks. 99% of the changes it would cause look like
In general, +1.
@@ -933,7 +933,8 @@ _PG_init(void)
MarkGUCPrefixReserved("isn");
}-/* isn_out +/* + * isn_out */ PG_FUNCTION_INFO_V1(isn_out); DatumWhile that's not that big a deal, it improves style uniformity since
most of our multiline comments do not have any text on the first line.
Yeah, this seems like a good change, provided we've minimized all the
unintended side-effects. I know there was some discussion about whether
this change should be made in the Perl script, but I really can't blame
anyone for not touching the pg_bsd_indent source code.
[2] adds a space between comma and an immediately following period,
which mostly affects variadic functions:@@ -147,7 +147,7 @@ px_set_debug_handler(void (*handler) (const char *))
}void
-px_debug(const char *fmt,...)
+px_debug(const char *fmt, ...)
{
va_list ap;It also helps a few struct constructors:
@@ -62,7 +62,7 @@ struct RBTree
static RBTNode sentinel = { - .color = RBTBLACK,.left = RBTNIL,.right = RBTNIL,.parent = NULL + .color = RBTBLACK, .left = RBTNIL, .right = RBTNIL, .parent = NULL };I don't see any places where it makes anything worse, and the
improvement in variadic functions is quite pleasing IMO.
So I think we should do that too.
I read through that patch and it LGTM.
--
nathan