attribute(mode) ignored by Sun cc

[Emmanuel Thomé]

On Sat, Jun 16, 2012 at 8:36 AM, Jay K <jay.krell at cornell.edu> wrote:
> I reported this years ago.

only if plural starts at 1.5

http://gmplib.org/list-archives/gmp-bugs/2010-November/002086.html

So it seems that you've been told the correct fix is to modify the
autoconf test, which nobody volunteered to do.

Honestly, I wonder what you expect. Fixing this issue is little work
(correcting the autoconf test for the mode attribute, and optionally
provide more sensible defaults for the #else part, if the elected ones
aren't good enough), and best done by yourself since you have an interest
in getting it fixed. There are many interesting things to do out there
rather than fixing somebody else's compilation issues on ultrasparc, so I
doubt anyone will do it for you. A sensible patch for that issue,
however, could probably be well received. For the benefit of all.

> It is still a problem in 5.0.2.

which is not current. Unimportant since I doubt 5.0.5 differs for that matter.

Best,

E.

>
>
>
> /* Define stuff for longlong.h.  */
> #if HAVE_ATTRIBUTE_MODE
> typedef unsigned int UQItype    __attribute__ ((mode (QI)));
> typedef         int SItype    __attribute__ ((mode (SI)));
> typedef unsigned int USItype    __attribute__ ((mode (SI)));
> typedef         int DItype    __attribute__ ((mode (DI)));
> typedef unsigned int UDItype    __attribute__ ((mode (DI)));
> #else
> typedef unsigned char UQItype;
> typedef         long SItype;
> typedef unsigned long USItype;
> #if HAVE_LONG_LONG
> typedef    long long int DItype;
> typedef unsigned long long int UDItype;
> #else /* Assume `long' gives us a wide enough type.  Needed for hppa2.0w.  */
> typedef long int DItype;
> typedef unsigned long int UDItype;
> #endif
> #endif
>
>
> -bash-4.2$ cc -V
> cc: Sun C 5.9 SunOS_sparc Patch 124867-16 2010/08/11
> usage: cc [ options] files.  Use 'cc -flags' for details
>
>
> gives
>
> "../../gcc-4.7/gmp/gmp-impl.h", line 3193: warning:  attribute "mode" is unknown, ignored
>
> mpfr has the same problem.
> In the #else, should it really be long, or should it be int?
>
>
>
>  - Jay
>
> _______________________________________________
> gmp-bugs mailing list
> gmp-bugs at gmplib.org
> https://gmplib.org/mailman/listinfo/gmp-bugs