[bug#30761] [PATCH staging 1/9] gnu: meson: Don't wrap the meson executable.

[ludo@gnu.org (Ludovic Courtès)]

Marius Bakke <mbakke@fastmail.com> skribis:

> Ludovic Courtès <ludo@gnu.org> writes:
>
>> Hello!
>>
>> Marius Bakke <mbakke@fastmail.com> skribis:
>>
>>> * gnu/packages/build-tools.scm (meson)[arguments]: Delete WRAP-PHASE.
>>> ---
>>>  gnu/packages/build-tools.scm | 5 +++++
>>>  1 file changed, 5 insertions(+)
>>>
>>> diff --git a/gnu/packages/build-tools.scm b/gnu/packages/build-tools.scm
>>> index 2abb44fdb..1ebccca43 100644
>>> --- a/gnu/packages/build-tools.scm
>>> +++ b/gnu/packages/build-tools.scm
>>> @@ -81,6 +81,11 @@ makes a few sacrifices to acquire fast full and incremental build times.")
>>>                 (base32
>>>                  "06r8limj38mv884s5riiz6lpzw37cvhbf9jd0smzcbi7fwmv3yah"))))
>>>      (build-system python-build-system)
>>> +    (arguments
>>> +     `(#:phases (modify-phases %standard-phases
>>> +                  ;; Meson runs itself through the Python interpreter, so
>>> +                  ;; we cannot use the shell wrapper.
>>> +                  (delete 'wrap))))
>>
>> I’m not sure what this means.  The ‘wrap’ phase is precisely for things
>> with #!/…/python, right?
>>
>> What problem does it solve?
>
> The problem is that at some point after 0.42 (the current
> meson-for-build), Meson does the equivalent of `python $(which meson)`
> at runtime.  Which ends up tripping over the shell wrapper.
>
> I don't remember exactly where I saw the problem, it could have been
> with "mesontest" which also resides in "meson/bin".  The only downside
> to deleting the wrapper is that you cannot call meson directly from the
> store; it works fine as long as PYTHONPATH is set (e.g. in a profile).
>
> I'll try to clarify the comment.

Yes, that’s a good idea.  :-)

It does mean that people installing meson in their profile will get
something missing-python-module errors, right?

Should we do the wrapping differently?

Thanks,
Ludo’.