We're building packages for the upcoming Fedora 24 with what will soon
be released as gcc 6.0, and also with a bleeding edge glibc. One of
the two of those, or perhaps both in conjunction, figured out that if
both pointer arguments to sincos() are not used, then the result is
equivalent to either sin() or cos(). As a result, the trick in
o/sfasli.c compiles down to a call to cos() rather than sincos().
This broke the maxima build.
I have applied the attached patch to the Fedora build to fix the issue.
Regards,
--
Jerry James
http://www.jamezone.org/
diff --git a/gcl/o/sfasli.c b/gcl/o/sfasli.c
index e0d6910..6ecf60a 100644
--- a/gcl/o/sfasli.c
+++ b/gcl/o/sfasli.c
@@ -136,13 +136,14 @@ extern int __remqu();
int
use_symbols(double d,...) {
- sincos(d,&d,&d);
+ double d2;
+ sincos(d,&d,&d2);
#ifdef GCL_GPROF
_mcount();
#endif
- return (int)d;
+ return (int)(d + d2);
}
#endif
_______________________________________________
Gcl-devel mailing list
[email protected]
https://lists.gnu.org/mailman/listinfo/gcl-devel
