AX_PYTHON_DEVEL macro does not deal with debug build of Python correctly

[Jean-Pierre Flori]

Hi all,

As we remarked at Sage's Trac ticket #13867, pynac fails to build on top of a debug build of Python because of the AX_PYTHON_DEVEL autoconf macro which incorrectly tests for the distutils module by checking that the stderr output is empty, what is not the case with a debug build of Python.

I'll report this upstream as well, but in the end, even if the upstream macro gets fixed, we'll have to autoreconf pynac.
If not, custom tests or other macros should get used.

(That's not the most important problem to take car of, but if it can be done...)

Best,
JP

[Burcin Erocal]

Hi Jean-Pierre,

On Mon, 31 Dec 2012 01:56:57 -0800 (PST)
Jean-Pierre Flori <jpf...@gmail.com> wrote:

> As we remarked at Sage's Trac ticket #13867, pynac fails to build on
> top of a debug build of Python because of the AX_PYTHON_DEVEL
> autoconf macro which incorrectly tests for the distutils module by
> checking that the stderr output is empty, what is not the case with a
> debug build of Python.
>
> I'll report this upstream as well, but in the end, even if the
> upstream macro gets fixed, we'll have to autoreconf pynac.
> If not, custom tests or other macros should get used.

Thanks for the pointer. I agree that it would be good to fix our build
system to work with a debug build of Python (that is not patched like
Volker suggests on #13867). This would also help when we comb through
the code that initializes various numeric constants in Pynac.

> (That's not the most important problem to take car of, but if it can
> be done...)

As the comments at the beginning of m4/ax_python_devel.m4 suggest, we
just borrowed that file from the gnu autoconf archive. This was
suggested by Julien in this thread:

https://groups.google.com/d/topic/pynac-devel/fNo4Aed-7Bc/discussion

When you say report upstream, do you mean Python or the autoconf
archive?


In any case, I'd be happy to merge a patch that fixes the macro.


Happy new year!

Cheers,
Burcin

[Jean-Pierre Flori]

I've sent an email to some autoconf-archive mailing list.
It was waiting for approval, no news yet.
As on the macro page, there was a bunch of different authors, I thought it was better idea to post on the list than to spam one random or all of the authors listed.