[Bug 756000] gstrtmpsink: add unlock

[GStreamer (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=756000

--- Comment #13 from Nicolas Dufresne (stormer) <nicolas.dufresne at collabora.co.uk> ---
(In reply to Tim-Philipp Müller from comment #12)
> But then it's not a continuous stream any more, which is what my question
> boils down to really. unlock_stop() isn't really supposed to interfere with
> that (but I could live with this in this context).

Unlocking a source/sink cause some lost of data in general. Not sure it matters
really.

> 
> > There is no way to unlock network operations without having to
> > reset the connection. That is true for many protocols.
> 
> I don't entirely understand this comment, could you elaborate what you mean?
> Are you talking about the TCP protocol level or some other protocol? Isn't
> this mostly an implementation detail? The rtmp implementation could surely
> be done in a way that's always fully async and fully interruptible (worst
> case you leave behind a thread that's stuck in some non-async connect or dns
> lookup, no?)?

That could be another way to support librtmp indeed. Spawn a thread, and do
everything in that single thread.  That thread would ensure client/server state
stay sane, whatever happens. We know the librtmp is not thread-safe, we should
probably stop hacking around.

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