Horton
Menu

Re: CI and test improvements

Justin Pryzby <pryzby@telsasoft.com>

View thread
From: Justin Pryzby <pryzby@telsasoft.com>
To: Andres Freund <andres@anarazel.de>
Cc: Andrew Dunstan <andrew@dunslane.net>, Thomas Munro <thomas.munro@gmail.com>, pgsql-hackers@lists.postgresql.org, Noah Misch <noah@leadboat.com>, Michael Paquier <michael@paquier.xyz>, Anastasia Lubennikova <lubennikovaav@gmail.com>, Tom Lane <tgl@sss.pgh.pa.us>, Robert Haas <robertmhaas@gmail.com>, Melanie Plageman <melanieplageman@gmail.com>, Peter Eisentraut <peter.eisentraut@enterprisedb.com>, Daniel Gustafsson <daniel@yesql.se>, samay sharma <smilingsamay@gmail.com>
Date: 2023-01-18T01:10:18Z
Lists: pgsql-hackers 
On Tue, Jan 17, 2023 at 11:56:42AM -0800, Andres Freund wrote:
> > $(CF_PGP_TESTS) is missing from contrib/pgcrypto/meson.build
> 
> Assume that's the false positive?

Yes

> > I also tried but failed to write something to warn if "meson test" was
> > run with a list of tests but without tmp_install.  Help wanted.
> 
> That doesn't even catch the worst case - when there's tmp_install, but it's
> too old.

I don't understand what you mean by "too old" ?

> > I propose to put something like this into "SanityCheck".
> 
> Perhaps we instead could add it as a separate "meson-only" test? Then it'd
> fail on developer's machines, instead of later in CI. We could pass the test
> information from the 'tests' array, or it could look at the metadata in
> meson-info/intro-tests.json

I guess you mean that it should be *able* to fail on developer machines
*in addition* to cirrusci.

But, a meson-only test might not be so helpful, as it assumes that the
developer is using meson, in which case the problem would tend not to
have occured in the first place.

BTW I also noticed that:

meson.build:meson_binpath_r = run_command(python, 'src/tools/find_meson', check: true)
meson.build-
meson.build-if meson_binpath_r.returncode() != 0 or meson_binpath_r.stdout() == ''
meson.build-  error('huh, could not run find_meson.\nerrcode: @0@\nstdout: @1@\nstderr: @2@'.format(

The return code will never be nonzero since check==true, right ?

-- 
Justin

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. ci: Add test coverage of different pg_upgrade modes

  2. seg: Add test "security" in meson.build

  3. cirrus/freebsd: define ENFORCE_REGRESSION_TEST_NAME_RESTRICTIONS

  4. cirrus/ccache: Use G rather than GB suffix

  5. ci: Upgrade macOS version from 12 to 13.

  6. meson: Add two missing regress tests

  7. Push lpp variable closer to usage in heapgetpage()

  8. ci: Change macOS builds from Intel to ARM.

  9. ci: Introduce SanityCheck task that other tasks depend on

  10. ci: Use -fsanitize=undefined,alignment,address in linux tasks

  11. ci: Clean up pre-meson cruft in windows task

  12. meson: Mark PROVE as not required

  13. ci: enable various runtime checks on FreeBSD and macOS