On 2011-08-16 11:17, Mike Gabriel wrote:
Hi Morty,
On Di 16 Aug 2011 10:20:45 CEST Moritz Struebe wrote:
On 2011-08-16 10:12, Mike Gabriel wrote:
We should re-think this model for post-Baikal though. I would not touch it now...
Ack: We shold add a post-Baikal to the todo. There is also cups and pulse, which need review.
Could you do that?
Yes I think so.
Esp. I am interested what you think about pulse...
Pulse-clients don't like it, if the server is suddenly not available any more. My idea was therefore to run a server locally an forward the stream to the client. If the connection is lost the server can then automatically forward the stream to the null-sink. Once the client reconnects, it can redirect the stream to the client again. Another advantage is, that the client can be automatically removed, once the x2go-connection is terminated. This is especially the case with a TC, where the next user does not want to hear the sounds of the application of the previous user (and the other way round). Unfortunately the tcp-sink has some synchronisation issues (Buffer overrun at client -> player slowed down -> buffer underrun at server -> player sped up, etc). This is much more unlikely to happen with a direct connection.
Cheers Morty
-- Dipl.-Ing. Moritz 'Morty' Struebe (Wissenschaftlicher Mitarbeiter) Lehrstuhl für Informatik 4 (Verteilte Systeme und Betriebssysteme) Friedrich-Alexander-Universität Erlangen-Nürnberg Martensstr. 1 91058 Erlangen
Tel : +49 9131 85-25419 Fax : +49 9131 85-28732 eMail : struebe@informatik.uni-erlangen.de WWW : http://www4.informatik.uni-erlangen.de/~morty