Solaris 64 bit compilation

[Martin v. Löwis]

phoebe_1 at att.net (Holden Caulfield) writes:

> > > "./Include/pyport.h", line 480: #error: "LONG_BIT definition appears
> > > wrong for platform (bad gcc/glibc config?)."
> > 
> > So what value has SIZEOF_LONG in pyconfig.h?
[...]
> To answer your questions:
>   - pyconfig.h does say 8 as SIZEOF_LONG

That raises the question then why the error in pyport.h is
produced. LONG_BIT comes from limits.h, and, if _LP64 is defined, is
64. So I'm still uncertain what you did to provoke this error.

>   - 'curses' modules does not compile. 
>   - 'dbm' modules compiles fine but core dumps during test.
> So, I disabled these modules in 'setup.py' and built it and ran the
> make test.

curses fails for me as well, since it finds a 32-bit ncurses
installation; dbm also core dumps here.

> test test_time failed -- Traceback (most recent call last):
>   File "./Lib/test/test_time.py", line 46, in test_mktime
>     time.mktime, (999999, 999999, 999999, 999999,
>   File "/apps/python/build/Python-2.2.1/Lib/unittest.py", line 279, in
> failUnlessRaises
>     raise self.failureException, excName
> AssertionError: OverflowError

This is a known bug (#460357), which has been fixed in CVS.

> 2)
> test_gettext
> test test_gettext crashed -- exceptions.OSError: [Errno 20] Not a
> directory: 'xx/LC_MESSAGES/gettext.mo'

I believe this also has been fixed in CVS.

> Are there any plans to support 64-bit python executable/module on
> solaris platforms for future versions?

Building python in 64-bit mode on Solaris works already. Some modules
may still not work; patches are welcome.

> Is anyone even interested in the configure.in patches?

I don't really see a need, since it works fine without the patches.

Regards,
Martin