#12466 Please update `openh264`to 2.6.0 in F41
Closed: Duplicate by patrikp. Opened by abitrolly.

  • Describe the issue

openh264 released a new version that fixes https://github.com/cisco/openh264/issues/3808 and others - https://github.com/cisco/openh264/releases/tag/v2.5.0

  • When do you need this? (YYYY/MM/DD)

Three months ago :D

https://community.deeplearning.ai/t/videos-dont-play-on-firefox/695251

  • When is this no longer needed or useful? (YYYY/MM/DD)

No ETA.

  • If we cannot complete your request, what is the impact?

Problems with playing inline gifs and videos onReddit, sluggish and unplayable videos in messengers, and inability for Fedora users to watch Deep Learning AI courses.


There was no build of this for f41 until... this morning. :)

@kalev I assume you want us to ship this one? or are you still testing?

Metadata Update from @phsmoura:
- Issue tagged with: low-gain, low-trouble, ops

Yes, it would be nice to ship it. I did a quick round of local testing and didn't find any regressions.

I mostly did the f41 build in response to https://bugzilla.redhat.com/show_bug.cgi?id=2325032, but it doesn't appear to fix the issue reported there, sadly.

@kalev the original issue "Openh264 loop video gets stuck in .mp4" may have been fixed, actually. But we don't have test file to know that for sure.

My issue with Reddit gifs in Firefox and DeepLearning.AI video is not fixed, but they might be unrelated to the original.

OpenH264 2.6.0 has been released, so this issue is now obsolete.

(I don't think we should ever update OpenH264 in stable releases anyway, unless there is some specific bug we are trying to fix. The updates are relatively risky. I would only update rawhide and maybe branched.)

@catanzaro why do you think new OpenH264 releases are unstable? There is no API to break.

Although applications build against noopenh264, It's still a library with API/ABI and it has accidentally broken ABI without a soname bump twice in recent years (example 1, example 2). There are also commonly other serious regressions. We just don't know whether an update will work before it's tested.

And in fact, 2.6.0 is unfortunately broken so we should not update to it.

Metadata Update from @patrikp:
- Issue assigned to patrikp

This is done now and the tracker ticket is here: https://pagure.io/releng/issue/12617

Metadata Update from @patrikp:
- Issue close_status updated to: Duplicate
- Issue status updated to: Closed (was: Open)

Metadata