[Pythonmac-SIG] timegm() C API causing extension-building probs - PLZ HELP!

[Just van Rossum]

At 7:00 PM -0500 12/8/98, Jason McVay wrote:
>hey all--
>
>i don't quite know if this is a simple question or not, but alas i've
>looked for an answer and couldn't find one, so i've decided to come to you
>guys.
>
>when i try to build lemberg's mxTextTools, mxTools, or mxDateTime
>extensions, i always have compilation errors dealing with a variable that's
>based on a file timegm.c

Hm, I've compiled mxTextTools succesfully before, and I just compiled the
other two you mention: no problem at all. I have no file called timegm.c on
my machine, so I really have no clue what that is all about.

Hm, I just checked, the symbol HAVE_TIMEGM is probably not defined, so
timegm() is not used. No clue what consequences that has for the packages.
test.py for DateTime fails with a SystemError, "tz_offset" not supported.
Hm.

So, in case you could live without timegm and tz_offset (whatever they are
;-) it is at least possible to compile the stuff, and it seems only
DateTime causes some trouble. The test for NewBuiltins seemed to pass,
except Python needs an awful lot of memory to do it.

Let me know if you want project files for CW4 (maybe these are compatible
with CW3, I'm not sure.)

Just