On Mi, 2015-04-22 at 10:24 +0800, Raymond Yau wrote: > > > The Capture Volume and Mic Boost are similar to fine adjustment and > coarse adjustment of the microscope Tank you for the excellent analogy. Microscope users do the adjustment first and fine later to get a sharp picture. Commit e6051cdf inverts that to "fine first, coarse later" - obtaining a good result would be almost impossible with a microscope operated like that. The same holds for the volume control. This strengthens my point that the above commit should be reverted. > It is not easy to combine the "fine" and "coarse" adjustment into a > single volume slider which fit different scale in volume/boost Agreed. That's why I suggested not to combine them by default (but PA does this). > some user may prefer to adjust Boost first and Volume but other user > may prefer adjust Volume and keep Boost fixed If PA applies "coarse first, fine later", the volume control behaves smoothly over the entire range. The reason why users have preferred an inverted order in the past is that on some systems, high Mic Boost levels introduce a lot of noise. But (I think) the solution taken by commit e6051cdf is wrong. It would have been better to fix the Boost at a low level (0 or 1). I agree that users may have different preferences in this area. Fixing that would require to expose the choice to users in a user-friendly way. Currently it is deeply hidden in the Alsa mixer path conf files. > As pulseaudio is using floating point volume scale and combine volume > and boost to a single slider, the optimal solution may need a very > complex method of adjusting the fine and coarse adjustment Currently, PA adjusts the controls in the order specified in the path conf file. In order to ensure "smooth" volume control over the entire dynamic range, PA should instead figure out which controls are "coarse" and which are "fine", and adjust the former before the latter. Unfortunately, a proper destinction between "coarse" and "fine" is difficult with PA's current "Mixer Element" abstraction, AFAICS. However, to re-iterate my previous statement, the proper solution for this problem is not to change the order of the merged mixer elements, but not to merge them at all. Regards Martin
