> 1) Poor documentation That's why we should put some effort to change this.
> 2) Difficult installation and configuration (not for me, but for users of > my > software) Of course! This is because Alsa is not a part of the kernel. Once it becomes a part of kernel, it will be the same like OSS in this respect. > 3) Constantly changing API (required lots of changes to my code 0.5 -> > 0.9) Again, if you have been reading up on the past couple posts in this topic, you'd realize that the API is rather close to the 1.0 release (something that I learned by participating in this thread as well), and as such is not prone any more to dramatic changes, as was the case in the past. > 4) Currently supports only Linux. (required: FreeBSD, OpenBSD, NetBSD, > Solaris, HP-UX, AIX and IRIX (I know, OSS doesn't support IRIX either)) > > I _don't_ like to support single OS API's, it reminds me too much about > Win... One of the main reasons for writing support for ALSA was that QNX > has > ALSA API. So it's not single platform API, but almost. I have to completely disagree with your statement here. I could not care less for BSD, Solaris or any other flavor or *nix. I use Linux and that is all I care for. Neither do I see any similarity with Windows since I find this aspect to be a rather weak basis for comparison, as well as irrelevant argument in respect to Alsa/Linux audio. Besides this is a typical issue of quantity vs. quality, and right now with OSS we got all the quantity we could possibly get with very little or no quality. On top of that, if Alsa proves to be all that is supposed to be, it will set the standard, and there will be an initiative from the other OS camps to port the architecture. So I see no point in doing their homework. OTOH I do understand there is the issue of cross-platform compatibility, but then it all boils down again to the issue of quantity vs. quality: do I want to make an audio app that will compile even on a palm pilot that will thus be capable of only producing different-pitched bleeps, or do I want to work for a specific platform and make an Ardour-like monster app? I would much rather see Alsa developers invest their time into perfecting Alsa's architecture, than use the same time for porting an API that is yet to be completed/widely used to other OS's. Ico