Re: [OctDev] control-2.3.51 released in package forum - please upload

[Thomas Weber]

On Mon, Jun 04, 2012 at 07:19:14PM +0200, c. wrote:
> Lukas, 
> Thanks for your great work!
> 
> I ran the tests on OSX 10.6 with Apple's vecLib and get the failures below [1]
> which you already said should be considered as "no problem".
> 
> On 4 Jun 2012, at 18:56, marco atzeri wrote:
> > I expect a package test function to behave reasonable correctly
> > for most of the system. Borderline case or alternative valid solution
> > should be avoided.
> 
> I agree with Marco, if you think these results are to be considered correct, 
> you should change the tests so that they are not reported as failures.
> This is particularly important as, for example, Debian packagers will not 
> produce a .deb 
> for your package until they see failing tests, and we reaally do want to see 
> the 
> control package in Debian!

Eh no, not quite. We usually look at the errors and decide. But here we
are talking about a factor of 10000 between expected and actual result.
That's a bug.

I also don't buy the argument that ATLAS and BLAS will produce wildly
differing results that are both correct. Whenever the differences
between ATLAS and BLAS were huge in the past, there was a bug in the
library or in the calling code.

        Thomas