Michael Schwendt mschwendt at gmail.com
Fri Feb 6 18:30:50 UTC 2015

On Sat, 07 Feb 2015 02:50:05 +0900, Mamoru TASAKA wrote:

> Just info:
> - Actually even zeitgeist upstream has no or little active development, so
>    even if Fedora downstream maintainers tries to do something
>    it can be little expected that these "bugs" can be fixed:

 | 0.9.14 (Diamond) release from the 0.9 series released 2013-06-18

 | zeitgeist-0.9.16-0.2.20140808.git.ce9affa

How does that snapshot fit into the activity scheme then? Simply
rebuilding 0.9.14 for Fedora made it crash for multiple users after
hitting stable updates repo too quickly. Building the snapshot may have
fixed the crash due to upstream changes, but a crash in sqlite3 in Oct
2014 could be a new problem, couldn't it? Bugzilla doesn't give any hint
about whether such issues are forwarded or compared with upstream
changelogs. Users trying to delete local sqlite db files is kinda a
brute-force way to fight such problems.

>    http://bazaar.launchpad.net/~zeitgeist/zeitgeist/trunk-freedesktop/changes/
>    https://answers.launchpad.net/zeitgeist/+question/260500

Odd. They consider it mature. Even more reason to communicate with them
about the stability problems at Fedora. IOW, their mature product somehow
fails here. And there's one fix that has been applied after creating the
snapshot, too.

> - As far as I quickly glanced at those bugs, some (many?) bugs don't seem
>    to have happened on direct zeitgeist side, and some (many?) bugs seem to
>    require GLib knowledge or so, and I doubt reassigning them (to glib2 or so) make
>    progress...

Well, just because it crashed within glib2 does not imply that glib2 is
the culprit. The rushed out broken build of zeitgeist resulted in users
seeing assertion aborts such as

  $2 = 0x1127cd0 "GLib:ERROR:gvarianttypeinfo.c:165:g_variant_type_info_check: assertion failed: (info->alignment == 0 || info->alignment == 1 || info->alignment == 3 || info->alignment == 7)"

  $2 = 0x1ab8ca0 "GLib:ERROR:gvarianttypeinfo.c:186:g_variant_type_info_check: assertion failed: (0 <= index && index < 24)"

and the snapshot made those go away. (Verified? Expected?)

More information about the devel mailing list