On Fri, 27 Apr 2007, Brian Gough wrote: > At Wed, 25 Apr 2007 14:22:15 +0100 (BST), > Henry Tillotson wrote: > > it repeats like this for a while, and later on different sequences, with > > more negative numbers, appear. > > If you need any more, please let me know > > Thanks for the info. The problem seems specific to Solaris, it hasn't > been reported before. > > It looks like there is an issue with signed-vs-unsigned char in the > preparation of the data to be sorted (they should all be positive, but > there is a -128 in the original data). The actual code looks correct > to me. > > If you want to investigate it further the function where that is done > is "FUNCTION(my, initialize)" at line 159 in sort/test_source.c -- I > can't really offer any more than that from here. > > -- > Brian Gough >
thanks for the hint about the signed/unsigned char issue: I've just tried adding the Sun cc option: -xchar=unsigned to CFLAGS in configure. The sort test now works, as do all the others. I guess that Sun cc must assume that char is signed by default, and that gcc assumes unsigned... or maybe it's Sun versus Intel hardware... anyway it works now thanks again Henry ===================================================================== Henry Tillotson e-mail: [EMAIL PROTECTED] Applications and Academic Support phone: 020 7679 7827 Information Systems fax: 020 7388 5406 University College London, Gower Street, London WC1 E 6BT ===================================================================== _______________________________________________ Bug-gsl mailing list [email protected] http://lists.gnu.org/mailman/listinfo/bug-gsl
