Hello, I just did some rereading of some parts of the "What Parts of Linux Audio Simply Work Great?" thread, that talk about the problems with soundcards that do not support multiple streams, and thought it would be good if we could actually come up with an advice to the desktop developers (Gnome and KDE mainly), distribution developers, and audio application developers in general.
This document should contain a detailed description of the current situation, of how we got there (i.e. how the desktop "sound daemons" actually created bigger problem than a solution, and why alsa does not do mixing in software by default), of how different user requirements lead to different solutions that are not always compatible (i.e. the "professional audio" vs "normal" users), and of all the different solutions currently available (and interfering with eachother). I believe such an overview is essential. I think most people on this mailing list have a pretty good idea about his, but do others? For example, I get the impression that there is a lot of misunderstanding or ignorance about alsa and dmix. Then it should propose an ad-hoc solution, and some guidelines of how to work towards a future in which everybody (including jwz ;-) ) is happy with linux audio that "just works". (I found jwz rant unjustified and unpleasant, but we can use it in our advantage if we give the right response, which, with a bit of luck (?) will get the same attention from the slashdot hords as jwz's blog) The ad-hoc solution, I believe, is something that should work "right now", or at least, as good as possible, with as little as possible changes to existing applications. For one, this would mean: making sure that dmix is being used when necesary, that no applications use the hw: devices explicitely, but the "default", that OSS applications use libaoss (If I understand correctly, libaoss can be told the use the dmix plugin, while alsa oss emulation will always use the hw device, or am I wrong here?). This is mainly a thing of the distro's. The remaining problem here is what to do with jackd. When the "professional" user runs jackd, and jackd complains that it is not using the hw: device directly, the solution should be obvious for him, and the non-jack apps should continue to work like before (but should be restarted, I suppose). Could anyone comment on this? The "occasional" jackd user can just run jack through the dmix plugin, which, if I understand correctly, would cause higher latency, but we are not talking about the "professional" user here. Proposing a roadmap for the future is much harder, but I think we can talk about that later. For now, I would like your opinion on the issue. Do you agree that such a document would be feasible and useful, and that the proposed structure/contents make sense? I am not sure that the mailinglist is the best way to write this document, maybe we could use a Wiki. I guess the first step would be to look for the relevant messages in the "What Parts of Linux Audio Simply Work Great?" thread, and write a short overview of those. maarten