bug#67042: ecl-cl-pcg fails to build due to a flaky test (was: Re: bug#67042: ecl-seedable-rng recently broken)

[Maxim Cournoyer <maxim.cournoyer@gmail.com>]

retitle 67042 ecl-cl-pcg fails to build due to a flaky test
quit

Hi Guillaume,

Guillaume Le Vaillant <glv@posteo.net> writes:

> Maxim Cournoyer <maxim.cournoyer@gmail.com> skribis:
>
>> Hi,
>>
>> guix-commits@gnu.org writes:
>>
>>> glv pushed a commit to branch master
>>> in repository guix.
>>>
>>> commit 42bec70a91d2205371c96287bcf565dcc5f5dd74
>>> Author: Andre A. Gomes <andremegafone@gmail.com>
>>> AuthorDate: Thu Nov 2 11:19:41 2023 +0200
>>>
>>>     gnu: cl-slynk: Update to 1.0.43-9.9c43bf6.
>>>     
>>>     * gnu/packages/lisp-xyz.scm (sbcl-slynk): Update to 1.0.43-9.9c43bf6.
>>>     
>>>     Signed-off-by: Guillaume Le Vaillant <glv@posteo.net>
>>>     Change-Id: I84ff141b7eefff470f72493d02f2cc24f02db7cf
>>
>> This commit or some preceding cl-* ones broke ecl-seedable-rng,
>> according to notifications from the CI:
>>
>> ecl-seedable-rng.x86_64-linux on master is broken
>>
>> Potentially related failures:
>>
>> - sbcl-cl-freetype2 (https://ci.guix.gnu.org/build/2584694/details)
>> - ecl-origin (https://ci.guix.gnu.org/build/2580823/details)
>>
>> See: https://ci.guix.gnu.org/build/2585168/details
>
> Hi.
>
> Locally I can build sbcl-cl-freetype2 and ecl-origin without issue.
>
> Concerning ecl-seedable-rng, there's something funny going on. The build
> details on the CI list only 1 dependency (ecl-ironclad), but there
> should be 3; ecl-cl-pcg and ecl-golden-utils are missing. And it is in
> fact the ecl-cl-pcg dependency that fails to build (because of a test).
> However, neither ecl-cl-pcg nor its dependencies have been updated
> recently, so why did the CI need to rebuild it, and why did it start
> failing?

Tests can sometimes be flaky (fail non-deterministically).  The best
course of action in these situations is to 1. report the problem
upstream and 2. disable the test in our packaging.  Re-titling
accordingly.

Cuirass mixing up derivations is tracked in bug #65505.

-- 
Thanks,
Maxim