make_it_cool: kdelibs/kdemm

[Marco Lohse]

Thomas Wilczek wrote:
> Am Dienstag, 22. März 2005 13:02 schrieb Marco Lohse:
> 
> 
>>>Would every NMM app have its own PlaybackNode or would you start one
>>>central NMM session per user (or per computer) that only handles the
>>>audio output, but then for all NMM apps (kind of like a soundserver).
>>
>>NMM is not a sound server. Every application will have its own
>>PlaybackNode. The PlaybackNode will then 'speak' with the sound server.
> 
> 
> hi!
> 
> but, i thought it was a main goal to get rid off a sound server (arts) ...?
> 
> right now, the playbacknode ist implemented as a "hardware node" afaik, 
> directly connecting to /dev/dsp (oss) but regarding your last sentence i get 
> the impression of a "facade node" for a given soundserver (mas, arts, ...).
> 
> it seems like we are missing some kind of simple "mixer node" muxing all 
> incoming request for a "hardware sink node" - the alsa dmix concept comes to 
> my mind.
> 

yes, I would prefer using alsa (or something else), too.

> hybrid approach: replace the current functionality of arts by a simple "mux 
> node" -> "playbacknode" (e.g. alsa - but configurable) combo which is started 
> only once for every existing "hardware sink node" and make only the "mux 
> node" shared. any later session requesting a "playback sink node" would just 
> connect to the "mux node".
> 

yes, we could do that. Actually, implementing such a mux node using NMM 
would be very easy. However, I would prefer to leave the actual audio 
mixing to existing solutions, e.g. alsa. (We will still need an alsa 
audio sink for NMM, but I guess we can provide that sooner or later)

> you could also insert some user defined "effect nodes" into that backend - 
> e.g. some equalizer to compansate your crappy notebook speakers. think of 
> your hi-fi amp unit... ;) 
> 

sure, I think Matthias currently prefers to use the NMM LADSPA node for 
that, right?

Marco.