Closed
Bug 1190829
Opened 9 years ago
Closed 9 years ago
Video duration on YouTube to long
Categories
(Core :: Audio/Video: Playback, defect)
Tracking
()
RESOLVED
DUPLICATE
of bug 1180214
People
(Reporter: CommodoreFabianus, Unassigned)
Details
User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:42.0) Gecko/20100101 Firefox/42.0
Build ID: 20150804030204
Steps to reproduce:
I'm using the most recent nightly of Firefox (from August 4th) although it already happened in versions before (unfortunately not sure which). Open almost any YouTube video and use the HTML5 player. It appears more reliably on long videos and when starts when it completes caching the video.
Actual results:
The video duration increases to a multiple of the original video. Here a few examples:
* https://www.youtube.com/watch?v=s8aQSxXoLFk 39:10 to 2:50:40 (about 4.25)
* https://www.youtube.com/watch?v=cKBI9tHAjkg 2:10 to 6:26 (about 2.5), or 4:05 (about 1.6), or 2:32
Even for the same video it doesn't always increase by the same amount. Sometimes it doesn't even increase at all. When I reload the page there seems to be no difference between F5 and Ctrl+F5.
It is possible to fast forward to any point in the video as long that point is before the original video ends. If you go further and in the “additional” area it stops working and you have to reload that tab. The video will also work normal and as soon as it reaches the original end stops normally (but the time jumps up to the shown duration).
It also doesn't start at that increased duration but at some point (sometimes it looks like it has finished caching the video) it slowly increases in steps.
Expected results:
The video duration shouldn't change.
|
Reporter | |
Comment 1•9 years ago
|
||
Okay I actually got that also on a Twitter video: https://twitter.com/MythBusters/status/629751996359118848
Could you give the details about your gstreamer config, please? (version)
|
|
Reporter | |
Comment 3•9 years ago
|
||
Apparently I've installed 1.2.4-4.fc20 and 0.10.36-7.fc20… Is there a way to get which version is actually Firefox using? I also stumbled upon http://www.quirksmode.org/html5/tests/video.html where only the H264 video showed an increased duration.
|
||
Comment 4•9 years ago
|
||
I'm on Debian testing, I had installed the gstreamer0.10 (0.10.36-2) and gstreamer1.0 (1.4.5-2) packages with the same issue. I removed gstreamer0.10 and now the duration looks good.
Linux, Fedora 21, x86_64.
I've been seeing this issue in nightly on Vimeo for about 3 weeks. I'll play a video, and the position indicator will move a little bit into the video, and stop there. The video keeps playing, and the time indicator keeps incrementing, but the position doesn't move anymore. If I click in the position indicator far ahead, it shows a time longer than the video length, the video starts over, and this second time, everything is correct. The duration that showed really long when it exhibited the stall behavior is now the correct duration. It doesn't seem to happen on all videos. If a video is very short, (less than a minute), it plays correctly without any second clicking.
Over the time that it has been happening, it has improved. The times used to be ridiculous, now they're only 3 or 4 times as long as the actual video. And clicking used to lock the video, now it restarts and is correct.
Using 39.03, everything works correctly, the way nightly used to work. Same system, so I don't think it is the libraries.
Here's a video link on vimeo that this just happened on.
https://vimeo.com/136237174
(In reply to stan from comment #5)
> Linux, Fedora 21, x86_64.
Do you feel like narrowing it down to a specific commit using mozregression?
http://mozilla.github.io/mozregression/
Note that distro builds probably use gstreamer-1.0 whereas mozilla builds still use gstreamer-0.10.
That's a good idea. (narrowing it down to a specific commit using mozregression). I custom compile, but this seems like something that would be the same for both the generic and custom firefox.
...mozilla builds still use gstreamer-0.10...
Except I'm compiling firefox here using mach build, ac_add_options --enable-gstreamer
Would it still use the hg library instead of the system library?
(In reply to stan from comment #8)
> Except I'm compiling firefox here using mach build, ac_add_options
> --enable-gstreamer
> Would it still use the hg library instead of the system library?
It uses the default which I think is 0.10. This is a duplicate of bug 1180214.
Status: UNCONFIRMED → RESOLVED
Closed: 9 years ago
Resolution: --- → DUPLICATE
You need to log in
before you can comment on or make changes to this bug.
Description
•