bug-gsl
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: libgsl.so is broken: it has unresolved symbols cblas_*


From: Mark Galassi
Subject: Re: libgsl.so is broken: it has unresolved symbols cblas_*
Date: Fri, 16 Oct 2020 23:53:41 -0600
User-agent: Notmuch/0.29.3 (https://notmuchmail.org) Emacs/26.3 (x86_64-pc-linux-gnu)

> Instead, I provided the proof that it is broken, and the recipe how to 
> fix it.

Dear Yuri,

Thanks for trying to get your package to work with GSL.

The way this development works is that the gsl team develops the source
code to gsl, and various other teams prepare binary packages for various
platforms.  Yet other teams develop the various tools involved in
between.

So we have to find out where the problem comes up.

We must pick apart if your problem is a bug in gsl, or if your problem
lies in other parts of the system (autotools for setting up linking,
freebsd, the particular compiler toolchain, ...).  I offered to help
diagnose things.

And the standard way to do this is not by asking gsl volunteers to do
something big, which would not help isolate the problem anyway.  The
volunteer who helps you would still have to reduce it to a MWE to figure
how what component breaks.  So the convention is that when you report
the problem you provide the MWE.

When you say "the problem is clear", that is not clear to me: you have a
big web page with a long link line with very specific software.  It
becomes clear when you reduce it to a small link line with a known MWE
and no other moving parts.

And when you say you have provided a recipe, that would be great: you
can send us a patch against a current clone of the git repo at:

https://savannah.gnu.org/git/?group=gsl

If you would like help preparing an MWE (for example based on the sample
cblas program in the GSL manual) please write me off list and I can work
with you on that.  So far the examples I tried to create from what you
said simply worked, so we need the detail and *your* MWE that exhibits
the problem.  It should be quite fast if we do it together.



reply via email to

[Prev in Thread] Current Thread [Next in Thread]