Re: [linux-audio-dev] Re: Linux VERSUS OSS ???

the mixer interface is a problem. OSS glosses over this by hiding 90% of the capabilities of most hardware mixers and stuffing it into an incredibly simplified model that then prevents users from doing things that they can do under Windows. ALSA messes it up by exporting 90% of the capabilities of most hardware mixers into user space and leaving the complexity for someone else to deal with :) takashi and jaroslav are working on the issues. i think their approach is correct (export the hardware capabilities, wrap them in alsa-lib to provide a simple interface for most apps and users) but it requires significant amounts of coding and won't emerge in a few days. assuming that your requirements are met by OSS's incredibly simplistic model of an audio device driver. need to control xrun detection? want to avoid starting the device until you've got enough data ready? want to use non-interleaved access? want to use a sample rate or sample format not supported by OSS? well, it won't take 5 lines, or 50 lines or 500 lines of code: you simply can't do any of this in OSS. JACK is explicitly NOT more flexible than OSS. its vastly LESS flexible. that's the whole point. 90% of audio apps don't need to concern themselves with the details of interacting with audio interface hardware. so JACK removes all of that: you get one sample format, a predefined sample rate and a somewhat inflexible buffer size. the tradeoff: your code gets much simpler, and you can forget about the hardware, which is good because it might not even be running or in use! plus you're running in a way that is well suited for the fundamentally real-time nature of audio. OTOH, JACK requires less lines of code than OSS to get up and running :) --p