Fri Nov 5 15:01:55 1999 Brian Gough * fp-sparclinux.c: added support for sparclinux Thu Oct 7 13:03:00 1999 Brian Gough * make_rep.c: more careful conversion from unsigned to signed integer for sign attribute to prevent warnings Sat Aug 21 01:05:01 1999 Tim Mooney * fp-aix.c: added, based on fp-solaris. Fri Aug 20 12:17:53 1999 Brian Gough * fp-tru64.c (gsl_ieee_set_mode): note that INEXACT is not easily supported on Tru64, and give an error if anyone tries to use it Wed Aug 18 21:36:01 1999 Tim Mooney * fp-irix.c: added, based on fp-solaris. IRIX 6 has a rounding and exception interface that is identical to Solaris, right down to the enums. Tue Aug 17 18:36:01 1999 Tim Mooney * fp-tru64.c: added, based on solaris and HP-UX fp-* files. Rounding mode and trap control requires that the compiler be passed special options, see the comments in fp-tru64.c. Fri Jul 23 19:00:51 1999 Brian Gough * print.c: added fprintf versions of the printf functions Sat May 8 20:39:28 1999 Brian Gough * fp-linux.c (gsl_ieee_set_mode): changed from using the function __fput_setcw() to the macro _FPU_SETCW() since Khimenko Victor reports that __setfpucw() is not in the shared lib version of glibc-2.1.1 Fri Apr 2 20:52:59 1999 Brian Gough * endian.c (setup_dynamic_endianness): removed useless test, u.c[i]<0 for unsigned Fri Aug 21 15:36:22 1998 Brian Gough * fp-unknown.c (gsl_ieee_set_mode): made return type int, as it should be Mon Jun 15 22:02:00 1998 Brian Gough * renamed read-mode-string.c to read.c and print-ieee.c to print.c Tue Jun 2 19:29:34 1998 Brian Gough * gsl_ieee_utils.h: renamed GSL_IEEE_CATCH_INEXACT to GSL_IEEE_TRAP_INEXACT, which is a better name Mon Jun 1 15:27:08 1998 Brian Gough * fp-sunos4.c: support for sunos4 IEEE interface * fp-solaris.c: support for solaris IEEE interface * renamed fp-mode-string.c to fp-env.c, in order to avoid short filename problems * added support for setting the IEEE mode from the environment variable GSL_IEEE_MODE (only works for the Linux kernel so far) Mon May 18 16:20:17 1998 Brian Gough * The determination of endianness is now done on each call instead of at configure time (autoconf complains about what would happen to the test if it were cross compiling). Ok, so it's a bit slower but this isn't a performance critical routine. Sat May 16 23:10:09 1998 Brian Gough * This directory contains some routines for examining IEEE format numbers at the bit level