Re: Please help testing/debugging math/openblas (and variants) on Darwin

["Dr. Thomas Orgis" <thomas.orgis%uni-hamburg.de@localhost>]

Am Sun, 1 Mar 2026 19:05:02 +0000
schrieb Jonathan Perkin <jperkin%pkgsrc.org@localhost>:

> You don't need to bump PKGREVISION for build fixes.  You only need to 
> bump when the binary package changes, and if the build is failing then 
> there's no previous binary package.

OK, this is assuming that I don't mess up my patch and potentially
change a build that worked before, of course. But I was unsure about a
revision change being good for bulk builds to know if they should
try again. That is done based on revision control changes, then?

I'll keep the focus on the binary package in mind. My point still
stands, though, about having to note the PKGREVISION in all packages in
the group, right? Pkglint really doesn't like it in Makefile.common.
I've seen a scheme scetched out in emacs that has a MYPKGREVISION and
adds that to a base revision … I wonder if such would be acceptable. Or
just a setting that only applies if not overridden in the specific
Makefile? I really would like to avoid having to change 6 Makefiles
when I only did one central tweak, and having to remember to remove the
revisions again on version update in 6 places instead of 1.


Alrighty then,

Thomas

-- 
Dr. Thomas Orgis
HPC @ Universität Hamburg