Split gstreamer1.0-plugins-bad to facilitate inclusion in Ubuntu main
| Affects | Status | Importance | Assigned to | Milestone | |
|---|---|---|---|---|---|
| gst-plugins-bad1.0 (Ubuntu) |
In Progress
|
High
|
Jeremy Bícha | ||
Bug Description
Proposal
--------
Split the gstreamer1.
Background
----------
Many years ago, gstreamer split their plugins into good, bad, and ugly.
ugly have legal concerns, etc.
bad are lower quality than good
This is somewhat poetically described at
https:/
Years ago, Ubuntu moved some plugins needed for webcam support in the Cheese app from bad to good with assurances from upstream that they were planning on moving those plugins themselves in a future release. That never happened and it doesn't appear like upstream intends to move *any* plugins from bad to good.
For many years, GNOME has considered the bad plugins one of their standard dependencies. This is true not just for Cheese but also for GTK4 and webkitgtk.
However, it's not really all the -bad plugins. Fedora for instance explicitly disables some of the -bad set because of legal concerns. (If Fedora has legal concerns, then it sounds like those plugins should be moved to -bad but it really doesn't look like like plugins are being re-categorized after their initial inclusion.)
1.22
----
Ubuntu maintains the package move as a large patch. For the 1.22 release, upstream made a bigger change to one of those moved plugins to have it depend on another -bad plugin. Instead of moving yet another plugin to -good, our patch just kept that plugin at its 1.20 codebase. This is problematic.
Debian
------
By moving the webcam and related dependencies back to -bad, we can reduce our packaging diff with Debian. At least for the good set, the only remaining difference with Debian is our changes to disable some features on i386. (And Debian may align their i386 support to more closely match Ubuntu's).
I think we would need additional MIRs (at least libnice, libva) to push this new -bad split into Debian. But it may make sense there too.
Alternative
-----------
One approach would be to also split the source package. However, this was rejected because it would add to the complexity since there would be a need to keep 2 complex source packages in sync with each other.
Timeline
--------
Ubuntu Desktop needs the limited bad set in main in time for Ubuntu 24.04 LTS. Our goal is to get it done for 23.10.
Remaining issues
----------------
Could GNOME provide a set of recommended -bad plugins we should try to ship by default?
