> Am Dienstag 22 Juli 2008 03:19:24 schrieb Russell Sears: >> Michael 'Mickey' Lauer wrote: > Well, pulseaudio is great. Honestly, I love the concept, that's why I pushed > it in to Openmoko in the first place. But alas, the current state is way > unusable on our slow systems. Perhaps this will improve with the new branch, > lets see. Until then, it's alsa dmix for me.
i agree. pulse needs to get better first. still, dmix will bite us in the ass on the long run, since it has no concept of anything but static buffers, can nor handle fullduplex audio, or stream routing (its alsa after all). means moving playing audio from the bt headset to the line-out will never be possible with alsa only. doesn't matter. still enough other stuff to get working first, and in the end your apps should only use highlevel interfaces, and not need to tamper with alsa-interfaces anyways. so it is transparent and your app will do the same calls, regardless which underlying backend. >> Also, is GStreamer definitely in the future? I'm looking into >> fixed-point graphic equalizers; gstreamer seems to be a good place to >> add one. > > Absolutely. GStreamer is great and will be a vital part of the Openmoko > framework. ack ;) >> The choices for adding new audio processing code seem to be adding a >> plugin to gstreamer, alsa or pulseaudio. > > If you want to maximize performance, go alsa. If you want to maximize > reusing, > go GStreamer. > since alsa is hw-abstraction only, not stream routing, not stream processing (besides of in and output) and doesn't have any sane framework for stuff like equalizers (or any software based signal processing) i would recommend gstreamer for that. alsa should stay what it is. a driver framework. adding more layering violations doesn't help (and would hinder pushing stuff upstream in the long run) -- Joachim Steiger Openmoko Central Services _______________________________________________ support mailing list [email protected] https://lists.openmoko.org/mailman/listinfo/support
