[Python-Dev] Re: merging my*.h into pyport.h

[Vladimir Marangozov]

Peter Schneider-Kamp wrote:
> 
> What do you think would be appropriate? I have been toying around
> with the my*.h headers a bit and loaded up a "proof of concept" patch
> that makes all my*.h obsolete.

Where's the patch? ... Ah, found it - assigned to / postponed by nowonder.

This include business is tricky.
I for one am against seeing mymalloc.h moved to pyport.h. It doesn't belong
there and deserves a file on its own. If my*.[ch] files really disturb you,
you can make them obsolete by renaming them to py*.[ch] :-).

~/python/dev> ls Include/my*
Include/mymalloc.h  Include/myproto.h   Include/mytime.h
Include/mymath.h    Include/myselect.h

- feel free to move to pyport.h both myproto.h and mymath.h (as you did)
- mymalloc.h shouldn't be touched. You may eventyally want to rename
  it to pymalloc.h
- not sure what to do with mytime.h and myselect.h
- not sure whether it's worth renaming my*.c to py*.c

> 
> P.S.: What's the status of sre? I just ask because test_re
>       keeps nagging all the time. :-]

The problem is that /F has no idea where this bug comes from, and
the suspicions he had about memory glitches in the SRE code haven't
been confirmed by Barry's Purify analyses so far.

I did some testing too, but found only unrelated buglets. I second
Barry's opinion that the memory picture of the CVS snapshot is good
enough.

So this seems to be a bug related with the logic of the code, in an area
where Barry and I have already publicly qualified ourselves as clueless.

-- 
       Vladimir MARANGOZOV          | Vladimir.Marangozov@inrialpes.fr
http://sirac.inrialpes.fr/~marangoz | tel:(+33-4)76615277 fax:76615252