[vlc-devel] [PATCH 00/14] Move VLM implementation to a module

[Pierre Ynard]

Hello Denis,

> IMO I think the first step should be to evaluate what is actually used 
> of VLM (in VLC's code and user base) to determine which features are 
> used and for what and then triage which ones should be in the core and 
> which ones could benefit to be in a module.

I already mentioned the 4 in-tree users of the VLM; in 3.0 there was
also a LibVLC API for it, but it was removed. --vlm-conf and the lua CLI
expose a raw full-feature access to the VLM.

The HTTP interface offers both a raw access API, and a stream output
manager in the web front end that relies on it but uses only VLM
broadcasts. It had some bugs, but apparently some people tried to use
it and reported them (#9590). There is also a VLM mosaic dialog, but it
appears to have been completely non-functional in the current version of
the front end for the past 9 years (#25133).

The situation is worse for the Qt VLM dialog. It crashes (#25123),
some button icons still don't display properly (#25131), and basic
broadcast play/pause/stop controls are still buggy (#25132). The dialog
attempts to map all VLM features, however support for VLM schedule
is some one-way pretend (#25126). There are also some more secondary
display duplication bugs (#25134). There have been people trying to use
it and reporting bugs though (#2186, #2482, #6085, #17839...).

More generally, you'd be hard-pressed to find any interface properly
implementing a front end for VLM schedules, as the VLM core offers no
proper API for them (#25125).

The forums probably hold more information about how users try to use the
VLM.

-- 
Pierre Ynard
"Une âme dans un corps, c'est comme un dessin sur une feuille de papier."