GNOME Bugzilla – Bug 597704
Traceback when undoing 'split' operation
Last modified: 2010-02-22 00:18:05 UTC
Created attachment 144970 [details] backtrace See rhbz#527683: https://bugzilla.redhat.com/show_bug.cgi?id=527683 This only happened once and I haven't been able to reproduce it since.
Gah, bug has changed. Scenario: * blank timeline * add file to timeline [---------------A----------------] * seek to middle of file * cut => all good, we have two clips cut at current position [----A1-----][--------A2---------] * undo => ERROR : only second clip is kept, first one is removed and stays at the trimmed position [--------A2---------] * redo => ok I guess, two clips are back in place [----A1-----][--------A2---------] => BUT we get the following warning: pitivi/timeline/track.py:211: Warning: g_sequence_search: assertion `seq != NULL' failed self._controller.unset(self._property.name, keyframe.time) * undo again => ERROR : same bug as initial UNDO * redo again => ERROR : same as initial REDO + segfault
Setting the milestone so we don't forget about it... seems like a pretty important regression.
Setting as blocker as suggested by twi.
This is a gstreamer core bug, see bug #610444. Nonetheless something should be done on the pitivi side too. self._controller.unset(self._property.name, keyframe.time) This is deprecated API. Instead of setting/unsetting control points on the controller directly a new abstraction was added in 0.10.14. the control point stuff is inside the GstInterpolationControlSource now. No idea if that API is exposed in python already
commit 75e05d2679314ada4551623dd049fbab30efe87c Author: Alessandro Decina <alessandro.d@gmail.com> Date: Mon Feb 22 01:16:55 2010 +0100 Fix a regression undoing splitting of clips. Fixes #597704.