More fun related to #1518464. This should fix everything really
hard! We backport a patch that should actually fix#1518464 if
frame-buffer is present but drm is not...but we also backport a
patch that fixes boot with 'rhgb' when neither renderer is
present, and move frame-buffer back to the subpackage, as we
think that's the 'most correct' package layout assuming no bugs.
Turns out that if we put *both* in the graphics-libs subpackage
minimal installs with 'rhgb' (which is what anaconda does out
of the box) break. As @pbrobinson wasn't around to explain the
rationale for moving the drm renderer to the subpackage in the
first place, @halfline and I agreed to put both renderers back
in the main package for now to avoid both that problem and
bug #1518464 . We can revisit this after @halfline does some
upstream code fixes.
The file was last updated in 2008, which was kind of a while ago.
Having files in /var conflicts with the rpm-ostree model of
`/var` starting empty.
Further, I don't think it makes sense to have a *global* boot
duration - it should likely be per-Fedora Edition. So probably
we should delete the one in `/usr` as well and have the Edition
packages carry it or something.
The file was last updated in 2008, which was kind of a while ago.
Having files in /var conflicts with the rpm-ostree model of
`/var` starting empty.
Further, I don't think it makes sense to have a *global* boot
duration - it should likely be per-Fedora Edition. So probably
we should delete the one in `/usr` as well and have the Edition
packages carry it or something.
- Fix color palette issue
- Fix splash at shutdown (if shutdown takes longer than 5 secs)
- Make sure text based splashes update terminal size when fbcon loads
It went from
0.2013.03.26.0 (F19)
to
0.2014.03.26.0 (F20)
to
0.1.2013.08.14.0 (F20)
Unfortunately, 0.2013 and 0.2014 compare higher than 0.1, so the release
number was actually going backwards.
https://lists.fedoraproject.org/pipermail/devel/2013-October/189917.html
Also, while at this, I've moved the snapshot_date to a %define so that
rpmdev-bumpspec wouldn't try to bump 2013 to 2014, like happened during
the F20 mass rebuild.
