Re: PG_TEST_EXTRA and meson

From: Nazir Bilal Yavuz <byavuz81(at)gmail(dot)com>
To: Ashutosh Bapat <ashutosh(dot)bapat(dot)oss(at)gmail(dot)com>, Jacob Champion <jacob(dot)champion(at)enterprisedb(dot)com>
Cc: Tristan Partin <tristan(at)partin(dot)io>, PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org>, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>, Andrew Dunstan <andrew(at)dunslane(dot)net>
Subject: Re: PG_TEST_EXTRA and meson
Date: 2024-07-23 10:32:17
Message-ID: CAN55FZ1Kt2E2oPScdTxSfTSxVXYp6=d-WGsOnbLwf_8siVYycA@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Hi,

On Tue, 23 Jul 2024 at 12:26, Ashutosh Bapat
<ashutosh(dot)bapat(dot)oss(at)gmail(dot)com> wrote:
>
> Upthread Tom asked whether we should do a symmetric change to "make".
> This set of patches does not do that. Here are my thoughts:
> 1. Those who use make, are not using configure time PG_TEST_EXTRA
> anyway, so they don't need it.
> 2. Those meson users who use setup time PG_TEST_EXTRA and also want to
> use make may find the need for the feature in make.
> 3. https://www.postgresql.org/docs/devel/install-requirements.html
> says that the meson support is currently experimental and only works
> when building from a Git checkout. So there's a possibility (even if
> theoretical) that make and meson will co-exist. Also that we may
> abandon meson?
>
> Considering those, it seems to me that symmetry is required. I don't
> know how hard it is to introduce PG_TEST_EXTRA as a configure time
> option in "make". If it's simple, we should do that. Otherwise it will
> be better to just remove PG_EXTRA_TEST option support from meson
> support to keep make and meson symmetrical.

I agree that symmetry should be the ultimate goal.

Upthread Jacob said he could work on a patch about introducing the
PG_TEST_EXTRA configure option to make builds. Would you still be
interested in working on this? If not, I would gladly work on it.

> As far as the implementation is concerned the patch seems to be doing
> what's expected. If PG_TEST_EXTRA is specified at setup time, it is
> not needed to be specified as an environment variable at run time. But
> it can also be overridden at runtime. If PG_TEST_EXTRA is not
> specified at the time of setup, but specified at run time, it works. I
> have tested xid_wraparound and wal_consistency_check.

Thanks for testing it!

> I wonder whether we really require pg_test_extra argument to testwrap.
> Why can't we use the logic in testwrap, to set run time PG_TEST_EXTRA,
> in meson.build directly? I.e. set test_env['PG_TEST_EXTRA'] to
> os.environ[;PG_TEST_EXTRA'] if the latter is set, otherwise set the
> first to get_option('PG_TEST_EXTRA').

When test_env('PG_TEST_EXTRA') is set, it could not be overridden
afterwards. Perhaps there is a way to override test_env() but I do not
know how.

--
Regards,
Nazir Bilal Yavuz
Microsoft

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Ashutosh Bapat 2024-07-23 10:40:31 Re: PG_TEST_EXTRA and meson
Previous Message Thomas Munro 2024-07-23 10:31:07 Re: CI, macports, darwin version problems