GNOME Bugzilla – Bug 626142
ISO-8859-1 characters in tasks.ics crash evolution tasks
Last modified: 2011-01-10 13:03:14 UTC
Created attachment 167215 [details] File reproducing the bug I recently upgraded from 2.28 to 2.30. After this upgrade evolution tasks crash with: The Evolution tasks have quit unexpectedly. Your tasks will not be available until Evolution is restarted. and: e_cal_view_start: Message did not receive a reply (timeout by message bus) (evolution:1044): libecal-WARNING **: e-cal-view.c:425: Unable to start view (evolution:1044): libecal-WARNING **: e-cal.c:404: e-d-s proxy died (evolution:1044): libecal-CRITICAL **: file e-cal.c: line 1733: assertion `priv->proxy' failed in the terminal. I found that it's caused by some ISO-8859-1 characters (probably left over from a previous version) in one of my task lists. I join two files which can be use to replace tasks.ics in .evolution/tasks/local/system. The UTF-8 works correctly. The ISO-8859-1 will reproduce the bug.
Created attachment 167216 [details] Correct file
Thanks for a bug report. I tried to reproduce this, both attaching to the file only, and replacing the system's tasks.ics file, but no luck, the e-calendar-factory doesn't crash for me, the task is shown with its description. I'm testing with a git master, which is just before 2.31.92. Could you run e-calendar-factory under gdb and get a trace of the crash, please? Just in case. Thanks in advance.
I don't get a crash. Under gdb, I only get: Server is up and running... Program exited with code 01. In the terminal where I started evolution, I get: (evolution:26647): camel-WARNING **: Camel operation status stack non empty: e_cal_view_start: Message did not receive a reply (timeout by message bus) (evolution:26647): libecal-WARNING **: e-cal-view.c:425: Unable to start view (evolution:26647): calendar-gui-WARNING **: e-cal-model.c:2125: Unable to get query, The name :1.817 was not provided by any .service files (evolution:26647): libecal-WARNING **: e-cal.c:404: e-d-s proxy died I am using 2.30.2
(In reply to comment #3) > I don't get a crash. Under gdb, I only get: > Server is up and running... > > Program exited with code 01. > ... > I am using 2.30.2 I recall seeing this when there is some trouble, but I do not recall how to track where was the factory exited with code 1 (it's usually on the 'exit' function call). Maybe run evolution as you use to, go to memos (which will invoke the e-calendar-factory process), and then attach gdb to that already running factory. And when you'll have it under gdb then go to tasks and try to reproduce it. I would go to memos because it doesn't open tasks, but when you go to calendar then it opens both tasks and memos, which is something we do not want.
I get exactly the same result as soon as I go to tasks: Program exited with code 01.
Hi there. I found out one reason that results in the error message that the task component is closing and tasks are not available until restart of evolution. If the due date of a task is before the beginning date of the task any attempt to modify this task through evolution results in the above mentioned error. I have not done extensive tests but modifying /home/tuomas/.local/share/evolution/tasks/system/tasks.ics fixed this anyway in one case. The following entry results in tasks shutdown: BEGIN:VTODO UID:20110104T072830Z-839-100-1-2@rotta DTSTAMP:20110104T072830Z SUMMARY:Kurtin pönttis DUE;VALUE=DATE:20110406 DTSTART;VALUE=DATE:20110502 CLASS:PUBLIC CATEGORIES:Pikkuhommat PERCENT-COMPLETE:0 PRIORITY:0 SEQUENCE:1 CREATED:20110104T072903Z LAST-MODIFIED:20110107T133014Z END:VTODO After changing the DUE;VALUE=DATE field this task can be modified in the normal way in the evolution user-interface: BEGIN:VTODO UID:20110104T072830Z-839-100-1-2@rotta DTSTAMP:20110104T072830Z SUMMARY:Kurtin pönttis DUE;VALUE=DATE:20110506 DTSTART;VALUE=DATE:20110502 CLASS:PUBLIC CATEGORIES:Pikkuhommat PERCENT-COMPLETE:0 PRIORITY:0 SEQUENCE:1 CREATED:20110104T072903Z LAST-MODIFIED:20110107T133014Z END:VTODO This applies at-least to version 2.32.1-1 of evolution. Probably to 2.32.0-1 as well, because the first thing I did after running into this problem was an upgrade from 2.32.0-1 to 2.32.1-1. I cannot verify the behavior with older versions that were originally the target of this bug report.
Thanks for the update, Tuomas, though your issue is different from the initial one. The first is about non-UTF8 letters used in the component (which wasn't a problem with bonobo, but is a problem with dbus), where that yours is about a problem with a "bogus component", which has set incorrect times. The e-calendar-factory claims about such component on its console when opening a calendar, but when modifying the component it aborts the factory, which seems to be unnecessary, and I consider it rather a left-over from testing. The first issue was fixed within bug #635825 (I forgot of this one when looking at it and I'm marking this bug as a duplicate of it). The second issue is still there, I can reproduce it with the actual git master of evolution-data-server too. I fixed the second issue in these two commits: Created commit 781f53b in eds master (2.91.6+) Created commit cd47a3a in eds gnome-2-32 (2.32.2+) *** This bug has been marked as a duplicate of bug 635825 ***