On 11/16 09:23, George Koehler wrote:
> On Tue, 10 Nov 2020 18:22:27 -0800
> Jeremy Evans <jeremy@openbsd.org> wrote:
>
> > OKs to switch the default Ruby version to 2.7?
>
> ok gkoehler@
>
> There are also a few ports that don't depend on Ruby, but do reference
> the default Ruby version. You didn't bump the REVISION of these few
> ports, and I don't insist on a REVISION bump, because these ports are
> strange. They are,
>
> devel/git-lfs
> Probably should not have a REVISION bump, because it only
> uses MODRUBY_BINREV to run textproc/ruby-ronn at build time.
This shouldn't get a bump as the package is not affected.
> graphics/inkscape
> Substs ${RUBY} in a patch to inkscape's list of script
> interpreters. If a user wrote a script in Ruby, then inkscape
> would run ${RUBY} (which is /usr/local/bin/ruby27 after your
> diff). I don't know if any Ruby extensions exist; their main
> extension language is Python.
This should probably get a bump, since it changes the package. I've
bumped it locally just in case.
> print/texlive/base
> Mentions ${MODRUBY_LIBREV} in pkg/README, but never uses this
> Ruby version anywhere, because the Ruby scripts are in texmf,
> but...
> print/texlive/texmf
> Adjusts Ruby scripts to "#!/usr/local/bin/ruby18".
> Ignores the default Ruby version.
I guess no bump currently needed, but this sounds like something that
should be fixed. However, TeXLive is complex enough as it is, and if it
has been broken for many years and nobody has complained, it's probably
not a big deal.
> On powerpc64, some tests in lang/ruby/2.7 crash with a segfault
> on address 0x18. I haven't seen this crash outside the tests.
> I built a few ruby27-* ports on powerpc64 and found no obvious
> problems. --George
It's possible the better fiber fix that the Ruby fiber maintainer is
working on may help this case, but a more educated guess would require
a backtrace.
Thank you very much for your detailed review, I'll commit the changes
tonight.
Thanks,
Jeremy
No comments:
Post a Comment