Re: GNOME 2.12 is slated to be our worst release -- ever!

[Vincent Untz <vuntz gnome org>]

Le lundi 05 septembre 2005 �2:51 -0600, Elijah Newren a �it :
> > Bugs that should block the release:
> > 
> > 1. text/plain documents on my desktop disappear
> > There is some suggestion (untested) that this is caused by broken
> > *notify kernels.
> > http://bugzilla.gnome.org/show_bug.cgi?id=314457
> 
> It seems to be less reproducible now (comment 11 says how to reliably
> reproduce but no one is able to reproduce anymore with those steps). 
> Also, it'd be really nice if someone could actually test with a kernel
> with the inotify fixes (i.e. >= 2.6.13), which no one seems to have
> done yet.
> 
> Really sucks, but I'm not sure I want to hold up the release for this
> (but then again, I keep stuff off my desktop anyway)...  What do
> others think?

I'd say we should punt this one: if it's not easily reproducable
anymore, it's not as major as it was before. Also, it seems we have
absolutely no idea how to fix it (for now) and it might take some
time...

> > 2. Splash screen doesn't disappear
> > I saw something about smproxy possibly fixing some of the splash
> > screen / session issues. Do I remember that correctly and has this
> > been done? Bugreport itself mentions long running programs causing
> > this.
> > http://bugzilla.gnome.org/show_bug.cgi?id=116814
> 
> It's in the release notes too and isn't a regression from previous
> releases (this bug has "always" been there), but I thought this was
> fixed with the smproxy removal too.  Mark?  Even if it isn't, I don't
> want to hold up the release given that it's already in the release
> notes.

I thought it was fixed too. And it's been here for a long time. Not
really a showstopper for me.

> > 3. keys refuse to "unbind"
> > Not totally sure if this is a showstopper. Not that many dupes. No
> > progress at all, if this is punted it must be fixed next release.
> > http://bugzilla.gnome.org/show_bug.cgi?id=133815
> 
> If it was punted in 2.10, I don't see the reason to hold up the 2.12
> release for it.  Given the few dupes, I don't think this adversely
> affects that many users (though it definitely does suck really badly
> and I hope we can advertise the showstoppers more next release and
> hopefully get Bastien on top of this)

Agree.

> > 4. Help browser content is inaccessible [REGRESSION]
> > Last comment suggest that this might be fixed. Should be tested! Also,
> > will 2.12 actually have that version?
> > http://bugzilla.gnome.org/show_bug.cgi?id=157941
> 
> >From the comments, it appears fixed now (though just not checked in
> gnopernicus).  This bug was brought up previously on this list (and on
> d-d-l months and months ago), and it was agreed that a11y shouldn't
> hold up forward progress on yelp for forever.

We already decided to ship with yelp 2.12, regardless of this issue.
Shaun added the F7 keybinding and Bill only needs to verify if it works
with gnopernicus. Not a showstopper.

> > 5. eog makes incorrect use of gthreads
> > When browsing many (not sure how many) images, eog will crash. Partial
> > patch has been committed in bug 312856. Has patch from Debian (so I
> > assume tested).
> > http://bugzilla.gnome.org/show_bug.cgi?id=315147
> 
> According to the Debian bug report, this also occurred in 2.8.x and
> 2.10.x, so it won't make this release any worse than previous ones. 
> Yeah, crashers suck, but I don't want to shove the release date back
> for non-regressions unless they're somehow more prominent now than
> before.

Agree.

> > 6. Leak in nautilus-icon-factory
> > A fix for a leak introduced a reliable crasher. Suggest to backout the
> > fix for the leak.
> > http://bugzilla.gnome.org/show_bug.cgi?id=307288
> 
> Yeah, adding new crashers sounds bad and I believe from Alex's comment
> this is probably pretty easy to get.  Luckily we have tested code
> (i.e. the old code, which we can get by reverting the patch that
> caused the crash).  I think we probably do need this fixed first
> though maybe nautilus-list can give us a better idea of the impact.

Patch has been reverted, according to the last comment.

> > 7. hal/non-hal incoherence: gnome_vfs_drive_get_activation_uri ()
> > This is the cause of bug 310270 (crasher, see below). Seems that patch
> > from 310270 works around this bug.
> > http://bugzilla.gnome.org/show_bug.cgi?id=312171
> 
> If it's worked around in gtk+ (as suggested in 310270), then I feel
> okay.  Otherwise, I'm scared to death by this bug.

Alex: any comment on this bug?

> > 8. GW Proxy - Evo hangs when started in offline mode
> > Start Evolution in offline mode, goto preferences, hang. No progress.
> > http://bugzilla.gnome.org/show_bug.cgi?id=314297
> 
> How often is evo started in offline mode--i.e. what's the impact of this bug?

Also, is GW Proxy used by a lot of people?

[... lots of other bugs, most are undecided ...]

> > 2. crash in file selector while open a directory
> > Many dupes. This would have been a blocker. However, a patch has been
> > committed. The patch seems to fix the issue. Should be verified.
> > http://bugzilla.gnome.org/show_bug.cgi?id=310270
> 
> Yes, I'm worried about this one... 

Last comment: "I can still reproduce this. The crash happens in every
app, with or without HAL running, when opening any directory, except for
home (home==desktop).
Interestingly, only selecting a directory and adding it to the bookmarks, then
opening it from there works OK."

Alex: any comment on this one? Is it really fixed?

Vincent

-- 
Les gens heureux ne sont pas press�