[HEADS UP] Fwd: svn commit: r328711 - in head: . Mk astro/viking cad/pdnmesh cad/repsnapper deskutils/pinot devel/xorg-macros emulators/catapult emulators/qemu-devel emulators/tme games/klavaro games/sea-defender ...

[Kevin Oberman]

On Tue, Oct 1, 2013 at 9:00 AM, Niclas Zeising <zeising at freebsd.org> wrote:

> On 10/01/13 17:01, Kevin Oberman wrote:
> > On Tue, Oct 1, 2013 at 1:35 AM, Niclas Zeising <zeising at freebsd.org>
> wrote:
> >
> >> On 10/01/13 00:16, Andrea Venturoli wrote:
> >>> On 09/29/13 19:17, Niclas Zeising wrote:
> >>>> HEADS UP!
> >>>>
> >>>> I just committed an update to the MESA graphics stack (libGL, dri) and
> >>>> various other xorg ports.  This mostly relates to people running with
> >>>> WITH_NEW_XORG=, but be sure to check UPDATING before updating.
> >>>> If you have any qeustions, don't hesitate to contact me or send a mail
> >>>> to the freebsd-x11 at FreeBSD.org mailing list.
> >>>> Regards!
> >>>>
> >>>
> >>> Great work!
> >>>
> >>> However, I think a little more info would be welcome by a lot of
> people.
> >>>
> >>> E.g.
> >>> In my case I have a Radeon card on 9.1... so:
> >>> _ I don't have KMS, right?
> >>
> >> No, you don't.  You have to update to the latest 10-CURRENT for that.
> >>
> >>> _ up to now WITH_NEW_XORG has prevented 3D hardware acceleration from
> >>> working; is it still so with this update?
> >>
> >> I think so.  Your best bet is to update to CURRENT.  10.0 isn't that far
> >> off, and we are in a code freeze, so 10 is reasonably stable.
> >>
> >>> _ any caveat?
> >>
> >> None known, just read UPDATING.
> >> Regards!
> >>
> >
> > Not to be contrary, but I think this is an exceptionally bad time to
> switch
> > to current if you want stability. Maybe in another week, but not quite
> yet.
> > When code freeze is upon us, there are LOTS of commits of WIP, and things
> > tend to break rather badly for a few days. These issues are normally
> fixed
> > fairly quickly, but it takes a little time and, from the issues raised in
> > mail to current@, things are still a bit rough.
> >
> > I really don't want to discourage the use of 10.0 as more use is far
> better
> > for a solid release, but running current can be a bit more risky, and
> > running it right after freeze is especially so. (So is running it right
> > after current is re-opened after release.) I'd recommend waiting until
> > around 10/10. (I like this date as it is the same whether your locality
> > puts month or day first). By then current should be pretty stable.
> >
>
> Just for the record, I've never really had any major trouble with
> CURRENT, and run it regularly on my desktop and laptop.
> Regards!
> --
> Niclas Zeising
>

I ran it through the "entertaining" times of going from v4 to v6 and seldom
had serious issues, but that was largely because I kept a close watch on
the current@ mailer and was familiar with the process. I don't think either
of us qualify as "typical" users. Nobody running head routinely really is
and I think it can impact our perceptions.

I can say with pretty high confidence that beginning of freeze and opening
to major changes after a release are the least stable times for both STABLE
and CURRENT. Instability with the freeze does not often last long. V5 was a
notable exception. The thaw is often a bit more disruptive.
-- 
R. Kevin Oberman, Network Engineer
E-mail: rkoberman at gmail.com