gap: use GAP_IN from kernel API

[fingolfin]

... to implement __contains__.

I am basing this on the code for _compare_equal which wraps GAP_EQ. Its use of sig_on etc. differs considerably from e.g. _mul_ which wraps GAP_PROD. I don't know if this has any significance, and which.

Contains PR #35761, once that is merged I'll rebase here. DONE

[fchapoton]

I am not sure if the finally is ever visited. Better assign the result to some variable and return this afterwards ?

[fingolfin]

I am not sure if the finally is ever visited.

Isn't that a core part of the Python semantics? Or are you suggesting cython might violate the usual Python semantics there?

[fingolfin]

Also note that other places in the code use finally just like that, including gap_eval

[fingolfin]

File "src/sage/libs/gap/element.pyx", line 619, in sage.libs.gap.element.GapElement.__contains__
Failed example:
    1 in libgap.eval('Integers')
Exception raised:
    Traceback (most recent call last):
      File "/sage/src/sage/doctest/forker.py", line 696, in _run
        self.compile_and_execute(example, compiler, test.globs)
      File "/sage/src/sage/doctest/forker.py", line 1106, in compile_and_execute
        exec(compiled, globs)
      File "<doctest sage.libs.gap.element.GapElement.__contains__[1]>", line 1, in <module>
        Integer(1) in libgap.eval('Integers')
      File "sage/libs/gap/element.pyx", line 635, in sage.libs.gap.element.GapElement.__contains__
        sig_on()
    cysignals.signals.SignalError: Segmentation fault
**********************************************************************
File "src/sage/libs/gap/element.pyx", line 622, in sage.libs.gap.element.GapElement.__contains__
Failed example:
    3 in libgap([1,5,3,2])
Expected:
    True
Got:
    False

So there is a segfault, seemingly on the sig_on added in this PR... I just don't get it.

[fingolfin]

Oh no, wait, of course facepalm the sig_on shows the error that it caught from the underlying code... and there I am passing in blindly the value other.value, but other might not be a GapElement at all -- indeed in the example it is a Python integer. So I guess I must inspect the other argument and possibly convert it? Hmm, but then again the other functions which call things like GAP_SUM or GAP_LT don't do that either...?!?

[fchapoton]

I am not sure if the finally is ever visited.

Isn't that a core part of the Python semantics? Or are you suggesting cython might violate the usual Python semantics there?

your are right:

https://docs.python.org/fr/3/reference/compound_stmts.html#finally-clause

[github-actions]

Documentation preview for this PR (built with commit 0096442; changes) is ready! 🎉
This preview will update shortly after each push to this PR.

[fchapoton]

@fingolfin : I have fixed things and it now works. Do you think it looks good ?

[fingolfin]

I couldn't say, I don't know enough about the Sage side. The bits I understand seems fine. If it works, it works...

[fchapoton]

ok, thanks. So let's run the CI once again and then I will set to positive