[bug#69579] [PATCH] gnu: python-joblib: Update to 1.3.2.

[Nicolas Graves]

On 2024-03-13 14:16, Nicolas Graves wrote:

> On 2024-03-13 12:39, Christopher Baines wrote:
>
>> Nicolas Graves via Guix-patches via <guix-patches@gnu.org> writes:
>>
>>> On 2024-03-08 12:29, Vagrant Cascadian wrote:
>>>
>>>> On 2024-03-06, Nicolas Graves wrote:
>>>>> * gnu/packages/python-xyz.scm (python-joblib): Update to 1.3.2.
>>>>> [build-system]: Update to pyproject-build-system.
>>>>> [arguments]
>>>>>   <#:phases>: Remove check phase.
>>>>>   <#:test-flags>: Keep former test flag.
>>>>> [propagated-inputs]: Add python-psutil.
>>>>>
>>>>> Change-Id: Ide110a19431b1d26e1d7dd8d1adc3b3f0f417749
>>>>> ---
>>>>>  gnu/packages/python-xyz.scm | 22 ++++++++--------------
>>>>>  1 file changed, 8 insertions(+), 14 deletions(-)
>>>>>
>>>>> diff --git a/gnu/packages/python-xyz.scm b/gnu/packages/python-xyz.scm
>>>>> index 46232d3c98..122dca6e4e 100644
>>>>> --- a/gnu/packages/python-xyz.scm
>>>>> +++ b/gnu/packages/python-xyz.scm
>>>>> @@ -6169,26 +6169,20 @@ (define-public python-pystitcher
>>>>>  (define-public python-joblib
>>>>>    (package
>>>>>      (name "python-joblib")
>>>>> -    (version "1.1.1")
>>>>> +    (version "1.3.2")
>>>>>      (source (origin
>>>>>                (method url-fetch)
>>>>>                (uri (pypi-uri "joblib" version))
>>>>>                (sha256
>>>>>                 (base32
>>>>> -                
>>>>> "0019p280s2k941mihl67l7y6amwx86639xp3zvpsg1lmyish67rh"))))
>>>>> -    (build-system python-build-system)
>>>>> +                
>>>>> "1cbjjzsh9hzaqr2cqja95673p7j88b8bd02hjpkq8xz147k6by4j"))))
>>>>> +    (build-system pyproject-build-system)
>>>>>      (arguments
>>>>> -     `(#:phases
>>>>> -       (modify-phases %standard-phases
>>>>> -         (replace 'check
>>>>> -           (lambda* (#:key tests? #:allow-other-keys)
>>>>> -             (when tests?
>>>>> -               (setenv "JOBLIB_MULTIPROCESSING" "0")
>>
>> We should look at why this was added, and whether it should be kept.
>
> This is only present in the check phase, not in the final package. I
> guess when we switched to the pyproject-build-system or simply updated,
> this simply wasn't necessary anymore.
>
> Since tests were consistantly (--check) passing, I do not see the need
> to investigate that.

Christopher, is that good enough of an explanation to get this one
merged or would you rather have mpe really investigating it?

>
>>
>>> All in all, I don't see a precise reason why this patch is responsible
>>> for any of the failures.
>>
>> QA seems to agree at least, there looks to be some issues, but this
>> change seems to not make the situation worse at least.

-- 
Best regards,
Nicolas Graves