Re: Rygel created UPnP stream help please

[Jens Georg <mail jensge org>]

On Mo, 2013-03-11 at 19:44 +1300, Andrej Falout wrote:

>         That requirement is pretty clear: no DLNA.ORG_OP parameter ->
>         no seek
>         headers allowed, not even those requesting the whole file. And
>         endless
>         streams as produced by GstLaunch are not seekable.
>
>
> It is then even more necessary to note how pointless it is to blindly
> obey such specification, when it is possible that at least three major
> manufacturers obtained certification based on it, while there devices
> are not implementing it.

Ah, you mean like not allowing to pause/fast forward videos when
DLNA.ORG_OP is missing, like ehm Samsung does, for example?

> Create an audio stream that can play on DLNA devices, namely Samsung
> and Onkyo. And therefore be able to treat DLNA device just the same as
> Apple Airport Express, Logitech Squeezebox, Sonos or an BlueTooth A2DL
> device can be treated - as a destination that plays sound, instead of
> having to depend on control point, built-in or (worse) otherwise. This
> would solve the problem of renderer needing a CP to skip to next
> track, problem of gapless playback, as well as for me even larger
> problem of integration UPnP devices with all other audio network
> devices I have in one interface... including synchronization, playlist
> transfer, etc etc...

Ok.

 
>         > Please forgive my ignorance, is "user-agent" some king of
>         software I
>         > can add or configure?  If it refers to the software
>         component in
>         
>         
>         User-Agent: A string devices SHOULD use to identify
>         themselves. Samsung
>         sucks at that hard times up until until the 2011 models. See
>         http://en.wikipedia.org/wiki/User_agent
>         
> Ah - HTTP 1.1 "user-agent" field.... how about using UPnP
> "manufacturer" ID instead?

Because UPnP uses HTTP quite a lot and so reuses the User-Agent (and
Server) header for this.

>         Although I just realised with commit
>         https://git.gnome.org/browse/rygel/commit/?id=32311fd80d1101c91b7796c2ea0920999c4bb142 that should 
> be at least solved for the recent models. Trying to make a patch.
>
>
> That is good to know - does it apply to audio seek issue?

https://bugzilla.gnome.org/show_bug.cgi?id=674736 now has a patch that
should ignore seeks based on user-agent.
 
>         > This patch is still marked as Committed - did it not fix the
>         issue or
>         > was it not really committed because of your comment about
>         breaking the
>         > test?
>         >
>         https://bugzilla.gnome.org/attachment.cgi?id=201807&action=diff
>         
>         
>         It was reverted since it broke DLNA conformity tests for the
>         N9.
>         
> So now we have two certified devices that cant be used together... 

Yes. And in fact to some degree that was intentional by DLNA. Some
devices are not meant to work together without a third one.