I've seen this bug for a REALLY long time. Like everyone who sees it, I guess. Anyway, I unmasked updated to latest ~x86, still not in there... so, in frustration, I found there is a fix and it is, in fact, utterly trivial. I'll attach the needed, with hope a DEV will put these in the tree, and it should apply to all versions in the tree, I think. Reproducible: Always Steps to Reproduce: 1. Gnome 2. Open 7 windows 3. open window 8, gnome panel freezes Actual Results: I have 8 windows open and gnome-panel freezes Expected Results: I have 8 or more windows open and gnome-panel still works Is all in the very long report I linked to. I'll propose 'major' severity, because it's a such trivial fix compared to the hassle it creates to the users, I can't believe it hasn't been applied already. ;-)
Created attachment 184735 [details, diff] patch against recent gnome-panel I tweaked Paskma's patch to match path of more recent version.
Created attachment 184738 [details, diff] patch against recent libwnck too... Again, I just tweaked the path for more recent version. I'm no maintainer or DEV.
So... just got an -r1 in the tree for gnome-panel. This patch set isn't in it. What did I do wrong? ;-) Trivial changes... and it works. What could I do to help get this applied in the gentoo patchset, then?
gee, forgot there was some patchs here. They look fine so if you say they work for you, let's include them.
replacing the launchpad issue by upstream. Alternative link: http://bugzilla.gnome.org/show_bug.cgi?id=86382 upstream seems not too happy about the patch because it is wrong wrt to gtk+ design/meaning for size request afaiu also this is still valid with 2.26 it seems.
(In reply to comment #5) > also this is still valid with 2.26 it seems. > Not a regression => Not blocking 2.26 stabilization. + Upstream does not like patch => no cookie :)
This patch rocks! Why is not included in the official ebuild?
(In reply to comment #7) > This patch rocks! Why is not included in the official ebuild? > Well, there is a new patch, here: http://bugzilla-attachments.gnome.org/attachment.cgi?id=140334 It applies cleanly. There is another patch for the 'notification area' which doesn't apply cleanly, so don't put any there in a vertical panel... Testing the patch, I find the behavior stupid compared to the last patch, in that it makes two columns of icons for open windows > 7 (rendering them uselessly small as icons), but at least it doesn't freeze and become totally useless.
(In reply to comment #8) > (In reply to comment #7) > > This patch rocks! Why is not included in the official ebuild? I have some new information that begs this question again. It seems that gnome-panel will be deprecated in favor of gnome-shell in 'gnome 3', aka 2.32. So, unlikely that there will ever be a patch accepted upstream. > > > > Well, there is a new patch, here: > http://bugzilla-attachments.gnome.org/attachment.cgi?id=140334 > > It applies cleanly. There is another patch for the 'notification area' which > doesn't apply cleanly, so don't put any there in a vertical panel... > > Testing the patch, I find the behavior stupid compared to the last patch, in > that it makes two columns of icons for open windows > 7 (rendering them > uselessly small as icons), but at least it doesn't freeze and become totally > useless. > In the end, the Paskma patchset is my preference. Also, since it won't be fixed upstream... please fix it here..?? As for gnome-shell as a replacement, I'm not sure that it will work well for everyone, so maybe folks will be stuck with current (buggy) vertical panel option (should they choose it at all) unless something is done. I propose these patches get included into the official ebuilds... I think other distros are using them, but I can't be sure. They've applied cleanly through several versions of the packages for me, so seems like a stable enough part of the code not to be a big hassle. Anyway, for someone finding this in the future, http://launchpadlibrarian.net/15679096/paskma-patch.tar is a tidy link. I changed the path in the patch to match recent version, dropped in /files, made ebuilds... a PITA. Anyway, my feeling is that the behavior is so bad without any fix, it should be addressed, one way or the other. I lost many hours before finding it was a bug in gnome-panel that locked my system for no apparent reason... just saying. ;-)
This needs to be handled directly to upstream because: - Suggested patches are too intrusive and cause other problems are seen in upstream report. - There are still no consensus about what alternative (patch) is the best option. We will track this directly on upstream bug and, once a consensus about what patch to use if reached there, we can try to consider its applying downstream if possible.
