GNOME Bugzilla – Bug 745518
http.playback.reverse_playback.vorbis_theora.1.ogg should be failed
Last modified: 2018-11-03 11:06:41 UTC
http.playback.reverse_playback.vorbis_theora.1.ogg seems to have false negative result. reverse_playback makes test media playback from end of file to 0 for testing reverse playback, but in this test vorbis_theora.1.ogg is played from 0 to end of file. So, while this test seems like it should fail, it actually 'passed' because it is ended successfully. The test scenario need to be modified to achieve intented result.
Created attachment 298418 [details] vorbis_theora_1_ogg.validate.logs my vorbis_theora_1_ogg.validate.logs file is attached.
How about adding, for example, 'error' action type like following. (reverse_playback.scenario) description, seek=true, reverse-playback=true, min-media-duration=4.0 seek, name=Revserse-seek, playback-time=0.0, rate=-1.0, start=0.0, stop=duration, flags=accurate+flush error, name=should-not-be-reached, playback-time=1.0 eos, playback-time=3.0
Created attachment 298796 [details] [review] Add 'error' action Attach a patch for discussion.
Created attachment 298871 [details] [review] Add 'error' action
As discussed on IRC your solution is not waht should be done fmpov. I proposed a solution at http://phabricator.freedesktop.org/D35 where I verify that the position reporting decrease if the rate of the last seek < 0 and increase otherwize. would be nice if you could check it out.
Created attachment 298990 [details] [review] New patch to test rate after a seek This change tests rate is set properly after a seek. In the result, this makes the following test failed which were passed but false negative. validate.http.playback.reverse_playback.vorbis_theora_1_ogg One weird side-effect is that some fast-forward tests fail additionally if gst-validate-launch runs with -j option. In this case, the rate value immediately checked after seek and async-done is not changed, so test fails. I have a plan to investigate this problem. p.s. Thibault Saunier, I created this patch before reading your comment. I'll test your suggestion soon.
Review of attachment 298990 [details] [review]: It is pretty much the same thing as what I proposed, can you check the patch I proposed at: http://phabricator.freedesktop.org/D35 ? ::: validate/gst/validate/gst-validate-report.h @@ +87,3 @@ #define EVENT_SEEK_NOT_HANDLED _QUARK("event::seek-not-handled") #define EVENT_SEEK_RESULT_POSITION_WRONG _QUARK("event::seek-result-position-wrong") +#define EVENT_SEEK_RESULT_RATE_WRONG _QUARK("event::seek-result-rate-wrong") You would have to also register that type, without doing it reporting will just no work I think
Review of attachment 298990 [details] [review]: It is pretty much the same thing as what I propose ::: validate/gst/validate/gst-validate-report.h @@ +87,3 @@ #define EVENT_SEEK_NOT_HANDLED _QUARK("event::seek-not-handled") #define EVENT_SEEK_RESULT_POSITION_WRONG _QUARK("event::seek-result-position-wrong") +#define EVENT_SEEK_RESULT_RATE_WRONG _QUARK("event::seek-result-rate-wrong") You would have to also register that type, without doing it reporting will just no work I think
Review of attachment 298990 [details] [review]: It is pretty much the same thing as what I proposed but my patch is a bit more complete I think. ::: validate/gst/validate/gst-validate-report.h @@ +87,3 @@ #define EVENT_SEEK_NOT_HANDLED _QUARK("event::seek-not-handled") #define EVENT_SEEK_RESULT_POSITION_WRONG _QUARK("event::seek-result-position-wrong") +#define EVENT_SEEK_RESULT_RATE_WRONG _QUARK("event::seek-result-rate-wrong") You would have to also register that type, without doing it reporting will just no work I think
OK, really sorry about that, I pressed several time the "commit" button because it looked like it was not working.
It's okay :) I'm testing your patch to check whether there is the side-effect I experienced.
This issue is waiting for merge of the patch in http://phabricator.freedesktop.org/D35
(In reply to Young Han Lee from comment #15) > This issue is waiting for merge of the patch in > http://phabricator.freedesktop.org/D35 I have it in mind but found regressions in the GES testsuite due to it so I need to find some time to investigate that before merging.
-- GitLab Migration Automatic Message -- This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity. You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/gstreamer/gst-devtools/issues/12.