Re: [yoshimi-user] Yoshimi Fails on Xrun!

  • From: cal <cal@xxxxxxxxxxxx>
  • To: Ivan Hernandez <ihernandez@xxxxxxxxxx>
  • Date: Thu, 22 Jul 2010 18:39:32 +1000

Just to clarify a point in my previous blurb - where I said there was code in
the
jack process callback to handle the early return, it's actually all in
Master::MasterAudio() and Master::actionLock(). From various points of
effect/midi
controller change processing, the call is to actionLock(lockmute), which not
only
waits for the lock, but also sets a muted flag (allegedly atomic in nature). At
the top of MasterAudio, the muted flag is checked, and if it's set,
MasterAudio()
returns immediately.

This is the stuff that I currently consider seriously inelegant, and I'm not
even
convinced that it's a wholly valid approach to the problem. Maybe it is, but
then
there's the question of just which effect/control changes actually do need to
apply the mute function. The results suggest that I might have got a couple of
them right, but there's possibly more that could use such treatment (again, if
and only if that basic approach is valid. This is where the mind boggle part
comes into play.

You did ask for this :-).

cheers, Cal


Other related posts: