[Python-Dev] ANSIfication again ;)

Thomas Wouters thomas@xs4all.net
Sat, 22 Jul 2000 10:47:13 +0200

On Sat, Jul 22, 2000 at 03:06:59AM -0400, Tim Peters wrote:

> > I also found a number of files which were already converted to 
> > ANSI syntax, but with one function leftover. Should I bother' uploading
> > a patch for those, or can I just commit them ?

> Commit.

Well, I already did that ;)

I've also ANSIfied everything in Python/ and Parser/, and the couple of
declarations in PC/ that were still using empty argument lists.

The changes fall in two categories: those I can test (and have tested and
seem to work fine) and those that I can't test because I haven't such
expensive hardware. I'll upload the latter category, and I'm perfectly
willing to upload the former one, too, but I wonder if anyone is actually
going to review them ? If not, well, there's not much use in uploading them,
is there ?

Also, there is an issue with extern function declarations: should we change
all those to full prototypes ? Some aren't too obvious (like the ^*@#$&**
readline functions, which don't have prototypes to start with, and the info
pages are not very helpful) and in one spot a function of the wrong type is
passed to PyAddCallback, resulting in a new warning during -Wall ;) A cast
can fix that, though.

And then there is the question whether all those externs are really
necessary. They seem to be put in because one OS or another was missing them
in a header file, but giving them prototypes from a Linux system might give
more warnings on other platforms: a 'const char **' is not the same pointer
as a 'char **', even though a 'const char *' is the same as a 'char *'.
Perhaps these prototyped function declarations should be put in pyport.h,
instead, surrounded by platform #ifdefs ?

	-SourceForge-CVS-stats-ly y'rs, ;-)
Thomas Wouters <thomas@xs4all.net>

Hi! I'm a .signature virus! copy me into your .signature file to help me spread!