What about... MME to WDM?

[stephang]

Would that be hard to implement? The WDM backend is already there from ASIO4All, one would "merely" need to replace ASIO by waveOut/MME on the other side (which seems to be well documented and all - at least to me as a non-programmer ). Benefits I can think of:

• 24 bit, 192 kHz support for MME apps (with WAVE_FORMAT_EXTENSIBLE support) in Win2k (and Win98SE/ME with WDM audio driver) - the OS still is perfectly good but restricted to 100 kHz and 16 bit
  
• Bit perfect MME even if card drivers do not circumvent kernel mixer; low latency, too
  
• Bypassing Aureon Sky/Space 24 bit recording driver bug (no workee in RMAA or Winamp line-in plugin, works via native ASIO and even ASIO via ASIO4All (<-- workaround in here?))
  

Stephan

[Michael Tippach]

I believe the way these vendors that _do_ bypass kmixer in their MME drivers do not replace major Windows components, as would be required for a "fake" MME implementation, but rather make use of the fact that - if they provide hardware mixing capabilities (or implement software-mixing in the miniport) - Windows simply would not instanciate kmixer at all - for as long as there is at least one more possible client instance left on the device, IOW kmixer would be spawned only on the last remaining instance of the device. For the usual single-client stuff, therfore, kmixer is always instanciated.