Re: [Bug-gsl] GSL 1.9 test_cblas vs Intel MKL

[Brian Gough]

At Thu, 26 Apr 2007 00:44:06 +0200,
Albert Strasheim wrote:
> GSL's sscal function does this as its first operation:
> 
> if (incX <= 0) { return; }
> 
> I might be mistaken, but I don't think other BLAS implementations have this 
> restriction. Instead incX=-1 just means step backward in memory from &X[0], 
> instead of forward.
> 
> Any thoughts?

Thanks for your email.  

The original BLAS Level-1 paper says "Only positive values of INCX are
allowed for operations ..... that have a single vector argument".  The
original Fortran reference BLAS returns zero in these cases, so we
followed that convention.

It looks like there are some errors in the test suite that don't
follow this pattern, do you know what is the problem with those?

-- 
Brian Gough
(GSL Maintainer)

Network Theory Ltd,
Commercial support for GSL --- http://www.network-theory.com/gsl/