Hi,
On Tue, Aug 19, 2025 at 12:31:14PM +0200, Álvaro Herrera wrote:
> On 2025-Aug-18, Tom Lane wrote:
>
> > Michael Paquier <michael@paquier.xyz> writes:
> > > On Mon, Aug 18, 2025 at 09:47:14AM -0400, Tom Lane wrote:
> > >> Couldn't this have removed some #include-s from multixact.c?
> >
> > > Right. funcapi.h and fmgrprotos.h are direct dependencies, but
> > > looking closer it is also possible to remove four more of them.
> >
> > Sounds good!
>
> Hmm, don't you find strange that dbcommands.h is still included there?
> I wondered about that and found out that we have get_database_name()
> declared there. It sort-of makes sense, because it was originally done
> by scanning pg_database, but since commit cb98e6fb8fd4 introduced a
> syscache for it, we can have this routine in lsyscache.c/h instead,
> where it feels more at home.
That makes sense to me.
> It also seems more sensible to declare
> get_database_oid() in pg_database.h.
Yes.
> After these two changes, a lot of
> files can stop including dbcommands.h. This seems a nice cleanup to me,
> and passes headerscheck.
meson does report:
../src/test/modules/worker_spi/worker_spi.c: In function ‘worker_spi_launch’:
../src/test/modules/worker_spi/worker_spi.c:455:25: error: implicit declaration of function ‘get_database_oid’
[-Wimplicit-function-declaration]
455 | dboid = get_database_oid(worker_spi_database, false);
It looks like pg_database.h include is missing in worker_spi.c.
That said, autoconf does not report the issue, and that's because test/modules
is missing in src/Makefile. Is there any reasons for that? If not, the attached
fix it.
Regards,
--
Bertrand Drouvot
PostgreSQL Contributors Team
RDS Open Source Databases
Amazon Web Services: https://aws.amazon.com
