Talk:Code stewardship reviews/Feedback solicitation/TimedMediaHandler
Official feedback period is now over; thank you for your input!
De-facto ownership and current limited resourcing
Currently Multimedia team has no resources to work on TimedMediaHandler and has hands full with other projects. As of January 2018, de-facto ownership is with me (Brion) under the scope of Audiences as sort of an adjunct micro-team finishing up long-lost tech debt in that area.
More dedicated resources to maintain would be very welcome, but improvements are ongoing, and since my shift to Audiences it'll be 80% of my time to work on until the current issues with frontend and subtitles are hammered out. (Remaining 20% of my time is TechCom/general architecture work.)
Ops needs are relatively small, and have been much reduced since the unmaintained ffmpeg2theora was removed from the dependency chain.
Future potential improvements like adaptive streaming, trim/editing tools, and resolving the upload formats question may require additional resources.
Future work on pushing many/large video contributions will eventually require capacity management for file storage, bandwidth, and server capacity for transcodes, but these are relatively straightforward problems once we get to them.
Future work making uploads easier for individuals may require better admin tools for review.
"improvements are ongoing, and since my shift to Audiences it'll be 80% of my time to work on until the current issues with frontend and subtitles are hammered out."
What would you call a "feature complete" TMH that can be "easily" maintained? IOW, are the subtitles and frontend work the only bits left for that?
Backend bits (handling of input, creating derivatives) is stable, if not ideal in all respects, so the frontend and the subtitle handler are the current work drivers.
Roughly what I intend to get it into comfy-space:
- fix the remaining bugs with the videojs control bar (source switcher icon is missing, and control resizer sometimes switches back and forth rapidly)
- bump resolution up/down on fullscreen switch
- finish working out subtitles to be videojs-compatible (pick up old patch with srt->vtt converter and clean API that will make subtitles be InstantCommons-compatible)
- if necessary go and manually fix up live subtitle pages that don't convert properly to WebVTT
- once we're happy with the front end, remove the old Kaltura player / MwEmbedPlayer bits to simplify the code
There's all kinds of future improvements to consider, of course, depending on time availability, but these are the most important.
>"improvements are ongoing, and since my shift to Audiences it'll be 80% of my time to work on until the current issues with frontend and subtitles are hammered out."
How long do you expect that to take (if it's just you working on it)? Once you've tackled those issues, do you anticipate being the maintainer of the audio/video stack or do we need to find dedicated resources to take on that role?
It looks like the full task breakdown is at https://phabricator.wikimedia.org/T100106.
A solution is needed
As described in phabricator, the front end is far from what users expect from a modern player. This needs to be addressed, but if this is done within TimedMediaHandler or within a new extension is strictly a technical decision. The important thing is for resources to be allocated, as video becomes more and more important, even for Wikipedia.
One point I would like to raise is the ability to upload non-free formats. While there have been a community consultation and the proposal was rejected, the software should have the ability to allow such formats, should the community reconsider.