[FFmpeg-devel] [PATCH] Consistent use of user_agent in hls.c

[Steven Liu]

2017-02-17 23:16 GMT+08:00 Jan Berkel <jan at berkel.fr>:

>
> > 2017-02-17 12:00 GMT+01:00 Jan Berkel <jan at berkel.fr>:
> > > This came up while debugging a problem with mpv:
> > > https://github.com/mpv-player/mpv/issues/4155
> > > ---
> > >  libavformat/hls.c | 6 +++---
> > >  1 file changed, 3 insertions(+), 3 deletions(-)
> > >
> > > diff --git a/libavformat/hls.c b/libavformat/hls.c
> > > index 3ae3c7cc5c..e1163a7927 100644
> > > --- a/libavformat/hls.c
> > > +++ b/libavformat/hls.c
> > > @@ -670,7 +670,7 @@ static int parse_playlist(HLSContext *c, const
> char *url,
> > >          av_dict_set(&opts, "seekable", "0", 0);
> > >
> > >          // broker prior HTTP options that should be consistent across
> requests
> > > -        av_dict_set(&opts, "user-agent", c->user_agent, 0);
> > > +        av_dict_set(&opts, "user_agent", c->user_agent, 0);
> >
> > Why is the code in http.c line 1048 not triggered that
> > should use the "user-agent" in case it was defined?.
>
> In my case the user_agent_* properties on HTTPContext never got set.
>
> The problem is that there is some client  code (mpv in this case)
> which uses "user_agent":
>
> https://github.com/mpv-player/mpv/blob/master/stream/stream_lavf.c#L205
>  av_dict_set(dict, "user_agent", opts->useragent, 0);
>
> This value should then be read from hls.c:1626
>
> update_options(&c->user_agent, "user-agent", u);
>
> I'm new to both mpv and ffmeg development so not exactly sure
> how this should be fixed. The result of the discussion on the mpv
> github issue was that ffmpeg needs to be changed, since user-agent
> is deprecated.
>
>          Jan
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>


Yes, the user-agent will deprecated.

And I think this can test with ticket #1291
https://trac.ffmpeg.org/ticket/1291