On Fri, Nov 22, 2024 at 05:23:32PM +0000, Stuart Henderson wrote:
> > > Also I'm not 100% sure but I think the "updating" of @conflict markers
> > > goes too far - these really only want to be listed when there are actual
> > > conflicts (in some cases, e.g. share/texmf-dist/tex/latex/etoolbox,
> > > share/texmf-dist/tex/latex/everyshi, share/texmf-dist/tex/latex/xpatch,
> > > bin/eptex, they are needed). Problem with over-listing @conflicts is
> > > that pkg_add has to merge the updates together - and if there's one
> > > set of ports in the tree where this is a problem (for disk use on
> > > updates) that's texlive.
> >
> > I agree that too strict @conflicts aren't good. They seem to be
> > refreshed by texmf/files/update_plist_hints.py. Sadly it appears that
> > adding only minimal @conflicts will help much with this update. Using
will *not* help much
> > sort(1), comm(1), sed(1) etc on the old and new PLISTS, I get:
>
> Ah, that's a pity - still, I think it would be good to use the minimal
> @conflicts to give a hint for future updates.
+1
> > pbuild /usr/ports/print/texlive$ cvs diff 2>/dev/null | grep -e
> > '^Index.*pkg/PLIST-' -e '^+@conflict '
> > Index: base/pkg/PLIST-main
> > Index: texmf/pkg/PLIST-buildset
> > +@conflict texlive_texmf-main-<2024
>
> I think that should be texlive_texmf-minimal-<2024 (the subpackage
> extension doesn't match the PKGNAME, it confused me a few times when
> I was reading the diff!).
Good catch, duh... The output was something like:
texmf/pkg/PLIST-buildset and ../tl2023/texmf/pkg/PLIST-main conflict
texmf/pkg/PLIST-main and ../tl2023/texmf/pkg/PLIST-buildset conflict
texmf/pkg/PLIST-main and ../tl2023/texmf/pkg/PLIST-full conflict
--
jca
