[Bug 268348] math/py-numpy: introduce FORTRAN / NOBLAS options

From: <bugzilla-noreply_at_freebsd.org>
Date: Wed, 10 Sep 2025 16:05:45 UTC
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=268348

Mateusz Piotrowski <0mp@FreeBSD.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|python@FreeBSD.org          |0mp@FreeBSD.org
              Flags|                            |maintainer-feedback?(python
                   |                            |@FreeBSD.org)
             Status|Closed                      |In Progress
                URL|                            |https://reviews.freebsd.org
                   |                            |/D52469
                 CC|                            |0mp@FreeBSD.org,
                   |                            |python@FreeBSD.org
         Resolution|Not Accepted                |---

--- Comment #19 from Mateusz Piotrowski <0mp@FreeBSD.org> ---
Hello python@,

I've reworked the original patch and I wonder if you could take another look.

I agree with Charlie that we need to be careful when adding extra options to
Python ports. A potential pilot error could lead to a lot of frustration and
long debugging. To address that I added an explicit warning to the option
description. There is a precedence for such warnings in the ports tree and it
is a good middle ground for different ports usergroups.

Regarding NumPy's support for NOBLAS: it is fully supported via internal
fallback routines. Their implementation is not very performant, however, so the
option description warns about that.

The new patch is here:
https://reviews.freebsd.org/D52469

Best,
Mateusz

-- 
You are receiving this mail because:
You are on the CC list for the bug.
You are the assignee for the bug.