GNOME Bugzilla – Bug 728012
[regression] Python3 port introduces out of boundary clip waveforms or thumbnails at certain zoom levels
Last modified: 2014-10-24 18:57:36 UTC
I grabbed the .ogv files of http://jeff.ecchi.ca/public/sample-pitivi-projects/2011-desktop-summit-project/2011%20desktop%20summit%20-%20clips.tar and inserted them into a blank timeline, on a fresh computer (it had no cache folder). Zooming in and out, I eventually encoutered weird thumbnails and waveform placement. I first saw this in the terminal: ./pitivi/timeline/elements.py:318: Warning: value "-1.000000" of type 'gfloat' is invalid or out of range for property 'width' of type 'gfloat' self.background.props.width = width - 2 Continuing to zoom in/out in the timeline, I eventually hit this: -------------- GStreamer-CRITICAL **: gst_poll_write_control: assertion 'set != NULL' failed GStreamer-CRITICAL **: gst_poll_read_control: assertion 'set != NULL' failed GStreamer-CRITICAL **: gst_poll_free: assertion 'set != NULL' failed GStreamer-CRITICAL **: gst_poll_get_read_gpollfd: assertion 'set != NULL' failed GStreamer-CRITICAL **: gst_bus_create_watch: assertion 'bus->priv->poll != NULL' failed ./pitivi/timeline/previewers.py:245: Warning: g_source_set_callback: assertion 'source != NULL' failed Segmentation fault ------------- I tried running it through gdb, but wasn't able to reproduce it. That 2nd part might be related to bug #726904 ?
Created attachment 274060 [details] first problem - out-of-boundary thumbnails
Created attachment 274061 [details] second problem - out of boundary waveforms
Why do you think this issue is Python 3 related? I only encountered this issue, when zooming in and out: The waveform is out of boundaries. When the clip has an empty audio track, you can see a thin line. http://i.imgur.com/CH7MalL.png When I look into the log, I only get this: /home/bmonkey/workspace/Pitivi/pitivi/pitivi/timeline/previewers.py:805: Warning: Source ID 4511 was not found when attempting to remove it GLib.source_remove(self._callback_id) This is the same warning like in Bug #726904.
I was able to fix the bug for me, also the warning from Bug #726904 by running the compute geometry method synchronously. Also had to fix more pep8. I also found some rotten unused code in the Video Previewer. https://github.com/lubosz/pitivi/commits/previewer
> Why do you think this issue is Python 3 related? Simply because I compared F20's updates-testing Pitivi 0.93 to current git master, which, AFAIK, has not changed significantly from 0.93 (especially not the timeline) besides the python3 port... so my first reflex was to suspect a regression caused by that ;)
I would like to see a git bisect on that thesis. But even if this issue was caused / pointed out by upgrading the python interpreter, it has to be fixed some place else. Does synchronizing the waveform update help? (like on my branch) I am not sure if I am experiencing the thumbnail issue. It looks ok for me. Does this only occcur with your video file? After several hours I had to cancel the 500 mb download :)
Hey Lubosz, Finally retested this today with master, I'm still able to trigger the "waveforms overflow" issue sometimes at certain zoom levels. Just have a bunch of short clips with clearly different sound levels to be able to notice when a clip boundary is not respected. Interestingly, scrolling horizontally corrects/reclamps the waveform, not just changing the zoom level. Filing a new bug #739142 about that.