At Wed, 1 Jul 2009 20:30:17 +0200, José Luis García Pallero wrote: > But, in my opinion, the problem not is that the gsl cblas > implementation check for the same error than refblas. The problem is > that, if you link your code with no gsl cblas and you are using the > returning codes in gsl for checking errors, if a cblas function > fails you cannot catch the exception and the program stop (and the > problem of not *freed* memory)
I've checked the BLAS papers again[1] and you are right, we should be catching the other errors like M<0 listed there. So we need to do two things (i) add the checks for those errors and (ii) provide a way of setting a custom cblas_xerbla function to avoid the hard-coded abort() at least when using GSL CBLAS. Let me know if these are things you could help with. Thanks. -- Brian Gough (GSL Maintainer) Support freedom by joining the FSF http://www.fsf.org/associate/support_freedom/join_fsf?referrer=37 [1] http://www.netlib.org/blas/blas2-paper.ps http://www.netlib.org/blas/blas3-paper.ps _______________________________________________ Help-gsl mailing list [email protected] http://lists.gnu.org/mailman/listinfo/help-gsl
