Re: Thoughts on mm-common and gmmproc

[Krzysztof Kosiński <tweenk pl gmail com>]

2010/1/7 Murray Cumming <murrayc murrayc com>:
> gmmproc's use of perl and m4 (and how those parts are interdependent)
> makes it very hard to add new features or to fix problems.

Agree 100%. I tried to add a feature to _WRAP_VFUNC that would make
wrapping GIO stream vfuncs easier (related to this bug:
https://bugzilla.gnome.org/show_bug.cgi?id=572471). In essence, I need
something like errthrow for vfuncs, and for async IO I need to replace
some pointer parameters with slots. It turned out I need to write a
lot of obscure Perl and M4 code. M4 is, to say the least, not for the
faint of heart, and the Perl part is only a little better. If the
entire script was in a single language (Perl or Python) it wouldn't be
very hard. With the current system I just gave up, because the M4 part
would become an incomprehensible jungle of nested ifelses, and because
the quoting rules are a complete mystery to me.

Python would be understood by more people, but Perl has a more concise
regex syntax. Either would be preferable to the hybrid Perl + M4
thing.

Regards, Krzysztof