Re: [Patch] Windows relation extension failure at 2GB and 4GB

On Wed, Nov 12, 2025 at 04:58:43PM +0900, Michael Paquier wrote:
> Thanks.  As the stamps have been pushed for the next minor release, I
> have applied and backpatched the meson check for now.  I'll look at
> your patch next, for HEAD.

Moving on to the I/O routine changes.  There was a little bit of
noise in the diffs, like one more comment removed that should still be
around.  Indentation has needed some adjustment as well, there was one
funny diff with a cast to pgoff_t.  And done this part as a first
step, because that's already a nice cut.

Then, about the test module.

src/test/modules/Makefile was missing, and once updated I have noticed
the extra REGRESS in the module's Makefile that made the tests fail
because we just have a TAP test.  This also meant that TAP_TESTS = 1
was also missing from the Makefile.  I've fixed these myself as per
the attached.

Anyway, I agree with the point about the restriction with WIN32: there
can be advantages in being able to run that in other places.  I think
that we should add a new value for PG_TEST_EXTRA and execute the test
based on that, or on small machines with little disk space (think
small SD cards), this is going to blow up.

Also, is there a point in making that a TAP test?  A SQL test should
work OK based on the set of SQL functions introduced for the file
creation step and the validation steps.  We could also use alternate
outputs if required.
--
Michael