error in Python 2.0 dist?

[Roland Mas]

D-Man (2000-11-08 01:14:23 -0500) :

> I downloaded the SRPM from python.org.  When I tried building I got
> the following:
> 
> [snip]
> gcc -g -O2 -Wall -Wstrict-prototypes -I./../Include -I.. -DHAVE_CONFIG_H   -c -o node.o node.c
> In file included from ../Include/Python.h:54,
>                  from node.c:3:
> ./Include/pyport.h:390:2: #error "LONG_BIT definition appears wrong for platform (bad gcc config?)."
> make[1]: *** [node.o] Error 1
> [snip]
[...]
> I have a PentiumII 300 with RH7.0 (gcc 2.96).  Python 1.5.2 runs
> fine on my machine.

This has been discussed several times here and on the Sourceforge
bug-tracking pages.  Basically the reason is that Red Hat 7 ships with
a most unstable version of libc, along with a most unofficial and most
unstable version of GCC.  The name GCC 2.96 in itself is wrong, since
it is not even a released GCC snapshot but a development version taken
at time t from the CVS tree (or something like that).  Those two in
combination (GCC + libc) trigger a bug defining LONG_BIT to be 64 even
on i386 machines, which is detected by the Python code.  The error
being in GCC+libc, the Python code does not try to fix it.  It stops
compilation instead.

Roland.
-- 
Roland Mas

Plus on en fout, plus y'en a du riz.
  -- Proverbe chinois.