Re: gmp-4.0 header files not installed

[Mike Castle]

On Wed, Dec 12, 2001 at 08:37:10PM +0100, Torbjorn Granlund wrote:
> Mike Castle <address@hidden> writes:
> 
>   With 4.0, at least longlong.h and gmp-mparam.h are NOT installed.
> 
>   Only gmp.h/mp.h are installed.
> 
> Yes?
> 
> Are you reporting that this is somehow an error?
> gmp-mparam.h and longlong.h, as well as most other GMP source
> files are not installed and should not be installed.

Ok.  Turns out I didn't take enough time to properly analyze the situation
(not surprising since the network to the machine in question is slow making
telnet painful).

The problem is, I was seeing the following:

gcc -g -O2 -Wall -Wstrict-prototypes -I. -I../Python-2.1.1/Include
-DHAVE_CONFIG
_H   -I/usr/lib -c ../Python-2.1.1/Modules/mpzmodule.c -o
Modules/mpzmodule.o
../Python-2.1.1/Modules/mpzmodule.c:71: gmp-mparam.h: No such file or
directory
../Python-2.1.1/Modules/mpzmodule.c:105: longlong.h: No such file or
directory
make[1]: *** [Modules/mpzmodule.o] Error 1


And I jumped to the incorrect conclusion that gmp had changed policies
about installing certain header file.

More analysis shows that it is indeed a Python issue, particularly:

#if __GNU_MP__ + 0 == 2
#define GMP2
#define BITS_PER_MP_LIMB mp_bits_per_limb
#else
#define MPZ_GET_STR_BUG
#include "gmp-mparam.h"
#endif


Of course, that's broken now.

I'll inform the Python people (I didn't see anything on their lists).

Sorry for the confusion,
mrc
-- 
     Mike Castle      address@hidden      www.netcom.com/~dalgoda/
    We are all of us living in the shadow of Manhattan.  -- Watchmen
fatal ("You are in a maze of twisty compiler features, all different"); -- gcc