A mips/mipsel gfortran issue

[Kevin B. McCarty-3]

Hi list,

FYI, there may be a showstopper for the gfortran transition on mips{,el}
for a little while, depending on how long it takes upstream to get a fix
for it:

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35662

The basic issue is that if your code calls both SIN() and COS() on the
same value in close succession, gfortran 4.3 [at -O1 or higher] on mips*
generates code that makes a call to sincosf() in libm -- and the call is
somehow broken.

I would imagine that a lot of scientific code may fall prey to this!
Incidentally, I hope all the packages that have been ported to gfortran
run test suites to check for this sort of problem -- better to have
something FTBFS than to have it get into the archive but be completely
(or worse, subtly) broken.

best regards,

--
Kevin B. McCarty <kmccarty@...>
WWW: http://www.starplot.org/
WWW: http://people.debian.org/~kmccarty/
GPG: public key ID 4F83C751