[Python-Dev] forwarded message from Gerald Pfeifer

Thomas Wouters thomas@xs4all.net
Fri, 6 Oct 2000 02:29:11 +0200

On Thu, Oct 05, 2000 at 08:14:20PM -0400, Fred L. Drake, Jr. wrote:

>   I just got this message from the gcc-announce list.  We've seen bug
> reports related to GCC 2.96 (which may or may not be the problem).  We
> need to be aware that these releases are out there and can be a
> problem.
>   I've noticed that the version of GCC shipped with Mandrake 7.1 is
> "gcc version 2.95.3 19991030 (prerelease)", which doesn't give my much
> confidence either.  ;(

The 'unstable' branch of Debian currently ships with:
gcc version 2.95.2 20000220 (Debian GNU/Linux)

and it seems to be working fine. I did notice a problem, but it was related
to glibc 2.1.94, not gcc: LONG_BIT was wrongly defined to 64. Tim's
sanity-check, which checks LONG_BIT against (SIZEOF_LONG * 8), would have
caught that one, but I found it on my machine before he checked in that fix
;) For me, that problem was caused by the /usr/include/bits/xopen_lim.h
include file:

/* Number of bits in a word of type ong int'.  */
#if LONG_MAX == 2147483647
# define LONG_BIT       32
/* Safe assumption.  */
# define LONG_BIT       64

where LONG_MAX was not defined (yet). I fixed it 'manually' for more than
just Python by adding

#ifndef LONG_MAX
#define LONG_MAX 2147483647

above it. I had to laugh, though, when I saw that assuming longs had 64 bits
is considered 'a safe assumption'. I guess most people use 64 bit machines
nowadays ? :-) 

I'm not complaining about this, though. Woody (Debian's current unstable
tree) is bleeding edge, and I'm fully prepared to live with it. In fact, I
love it! But people testing out glibc 2.1.90+ should keep this in mind. I'm
also wondering where to send my bugreport, but I think I'll read some
documentation before I do that, first ;)

Thomas Wouters <thomas@xs4all.net>

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