[Bug 795891] matroskademux: emits invalid Pango markup

[GStreamer (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=795891

--- Comment #18 from Philippe Normand <philn at igalia.com> ---
(In reply to Tim-Philipp Müller from comment #17)
> I think neither the naming nor the way this works is ideal yet :)
> 

Yeah this is a first shot, I basically extracted the code from subparse and
didn't change it.

> How does it work? Does it modify (and replace, if needed) an existing string
> in place?
> 

Yes.

> I dislike the _subrip() specific thing and also the is_webvtt boolean.
> 
> Could we do this as a generic function? (e.g. passing in a format enum with
> a flags field or something if needed)
> 

Hmm, yeah something like that would be possible.

> How about we pass a string in, function takes ownership and we return a new
> string? (Which may be the old one or not?)
> 

Yes this sounds better than modifying in-place and is most likely better for
bindings.

> What about error handling? Are there cases where we should/need to return an
> error or a warning?
> 

I don't think so. The idea is to output valid markup, not sure which case
should trigger errors.

> OR MAYBE we should just do this all differently, and force subparse to be
> plugged in this case via an extra caps field, and let subparse clean it up?

Actually I tried this first :) The problem is that subparse wouldn't be able to
auto-detect the subtitle format because it expects a full payload (cue timing
data and actual text data) whereas matroskademux only emits the text data.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.