Re: [Denemo-devel] Playback asynchronous

[Richard Shann]

On Sun, 2017-08-13 at 10:14 +0200, Andreas Schneider wrote:
> Am 13.08.2017 um 10:10 schrieb Andreas Schneider:
> > Am 12.08.2017 um 20:33 schrieb Richard Shann:
> >> On Sat, 2017-08-12 at 19:07 +0200, Andreas Schneider wrote:
> >>> Working on a score in which some measures are still without notes, I
> >>> encounter an asynchronous playback, i.e. during playback some staffs lag
> >>> one measure behind the others. This is illustrated in the attached
> >>> screenshot. Notes from one measure sound together with notes from the
> >>> next measure from other staffs, which is not so good for checking it.
> >>
> >> The problem is that the first measure is not empty - it contains the
> >> repeat start directive, so it is *not* filled with a space, as Denemo
> >> does not know what that directive is doing. Empty measures must be truly
> >> empty if you want Denemo to tacitly insert a whole measure rest. If you
> >> put a whole measure rest in the first measure, all is well.
> > 
> > What does that mean in general? That Denemo plays back as the music is
> > exported, e.g., to lilypond? I had expected Denemo to play back as shown
> > in the editor, not taking into account directives that might skip or
> > fill anything for the print output. Moreover, as Joe showed, the
> > playback can also be brought out of sync by incomplete measures. This
> > may not be what an average user expects, but is in accordance to how
> > such music is shown in the print output. That behaviour can also be
> > handy, e.g. for ancient music where notes may extend into the next
> > "measure", resulting in an overfull and a following underfull measure in
> > the Denemo editor, but a correctly typeset score in the print and a
> > correctly sounding playback. The question is: Which behaviour is best
> > for the majority of use cases?
> 
> Our mails have crossed in the ether, Richard has answered most questions
> already in his last post.

I've fixed the bug too! Now the Print View and the MIDI output should
match in all cases. The will both be "async" for Joe's example.
 Please look out for possible unintended side-effects, I think it's a
safe change, I just lifted the test out of exportlilypond.c and put it
into exportmidi.c and all seems well.

Richard