Re: Windows CFBot is broken because ecpg dec_test.c error - Mailing list pgsql-hackers

Hi,

On 2025-01-29 18:24:45 +0300, Nazir Bilal Yavuz wrote:
> On Tue, 28 Jan 2025 at 17:02, Andres Freund <andres@anarazel.de> wrote:
> >
> > Hi,
> >
> > On January 28, 2025 7:13:16 AM EST, Jelte Fennema-Nio <postgres@jeltef.nl> wrote:
> > >Since about ~11 hours ago the ecpg test is consistently failing on
> > >Window with this error[1]:
> > >
> > >> Could not open file C:/cirrus/build/src/interfaces/ecpg/test/compat_informix/dec_test.c for reading
> > >
> > >I took a quick look at possible causes but couldn't find a clear
> > >winner. My current guess is that there's some dependency rule missing
> > >in the meson file and due to some infra changes files now get compiled
> > >in the wrong order.
> > >
> > >One recent suspicious commit seems to be:
> > >7819a25cd101b574f5422edb00fe3376fbb646da
> > >But there are a bunch of successful changes that include that commit,
> > >so it seems to be a red herring. (CC-ed Noah anyway)
> >
> > I think it's due to a new version of meson. Seems we under specified test dependencies. I'll write up a patch.

Sorry, got distracted with somewhat pressing matters.


> The cause is that meson fixed a bug [1] in v.1.7.0. Before meson
> v1.7.0; although --no-rebuild is used while running tests, meson was
> building all targets. This is fixed with v.1.7.0.
>
> [1] https://mesonbuild.com/Release-notes-for-1-7-0.html#test-targets-no-longer-built-by-default

That's not *quite* right - it wasn't that the targets were built when
--no-rebuild was specified, it's that the default build target (for just
'ninja'), built all test dependencies.


> The change below fixes the problem:
>
> diff --git a/.cirrus.tasks.yml b/.cirrus.tasks.yml
> index 18e944ca89d..c7a94ff6471 100644
> --- a/.cirrus.tasks.yml
> +++ b/.cirrus.tasks.yml
> @@ -17,7 +17,7 @@ env:
>    CHECK: check-world PROVE_FLAGS=$PROVE_FLAGS
>    CHECKFLAGS: -Otarget
>    PROVE_FLAGS: --timer
> -  MTEST_ARGS: --print-errorlogs --no-rebuild -C build
> +  MTEST_ARGS: --print-errorlogs -C build
>    PGCTLTIMEOUT: 120 # avoids spurious failures during parallel tests
>    TEMP_CONFIG: ${CIRRUS_WORKING_DIR}/src/tools/ci/pg_ci_base.conf
>    PG_TEST_EXTRA: kerberos ldap ssl libpq_encryption load_balance
>
> And I think this is the correct approach. It builds all of the
> not-yet-built targets before running the tests. Another solution might
> be manually building ecpg target before running tests but I think the
> former approach is more suitable for the CI.
>
> CI run after this change applied: https://cirrus-ci.com/build/6264369203380224

I don't think that's the entirety of the issue.

Our dependencies aren't quite airtight enough. With a sufficiently modern
meson, try doing e.g.

  rm -rf tmp_install/ && ninja clean && meson test --suite setup --suite ecpg

It'll fail, because the dependencies of the tests are insufficient.

See the set of patches at
https://www.postgresql.org/message-id/qh4c5tvkgjef7jikjig56rclbcdrrotngnwpycukd2n3k25zi2%4044hxxvtwmgum


I think the only reason your patch on its own suffices, is that the "all"
target, that we ran separately beforehand, actually has sufficient
dependencies to make things work.



The nice thing is that with this meson improvement we should be able to get
rid of the "setup" test suite and instead generate the test install via
dependencies.  Obviously we either have to wait a fair bit or do it depending
on the meson version...


Greetings,

Andres Freund