Re: Issue with eselect-ldso -> flexiblas migration

On 2026-03-17 10:56:36, Poncho wrote:
> 
> It's a pesky one. Unfortunately, those additional flags didn't help 
> either. Attached the ldd output and my current settings to compile 
> packages with debug info.

Trying another approach, I started to build scipy myself to see what
is happening here:

  #1  0x00007faab5e3d8df in f2py_rout__flapack_dgelsd_lwork
  ...at scipy/linalg/_flapackmodule.c:18697

With the stable amd64 versions of scipy/numpy, that line is...

  /*callfortranroutine*/
  (*f2py_func)(&m,&n,&nrhs,&a,&m,&b,&maxmn,&s,&cond,&r,&work,&lwork,
               &iwork,&info) ;

and that fits with what you're seeing (backtrace mysteriously stops as
it switches from C to fortran).

At this point I would open a bug against scipy on bugs.gentoo.org.
It's probably not a scipy bug, but it is crashing in scipy, and the
people who maintain scipy (the python and science projects) might
have better advice on how to troubleshoot further.