Part combiner & Stem direction, w/patch

[Ted Ashton]

Well, having gotten a build environment put together, I've spent the day
working over the part combiner.  In specific, I changed:
  1) When soloADue is off, the stems of part "two" in split intervals now get
     pointed down, they didn't before.
  2) When one part has a long rest and the other has note during it (for
     example, an r4 in part "one" and r8 d16 r16 in part "two"), the smaller
     rests were getting cut, leaving the note up against a long rest.  The
     patch makes UNISILENCE the equivalent of UNIRHYTHM (that is, not only do
     both parts have to have silence, they must have it in the same rhythm).  I
     had to decide between doing that and making changes to the -devnull
     engravers to check for UNISILENCE and UNIRHYTHM both being set.  The docs
     I found didn't indicate which was better and this makes for a smaller
     patch and, I think, a more useful definition of UNISILENCE.
  3) I also did considerable cleanup to make things work when changeMoment is
     set to something other than the default.  This work is still not complete.
     Under known bugs should go that if you set it to 1/8,1/8 and you have, 
     for example, four 16ths barred together, weird things can happen including
     losing notes.  Fixing that is beyond my currently available time and
     knowledge.  It does work exceedingly well if you set it to 1/4,1/4, though
     and, so far as I can tell, still works as before (with the exception of
     fix 2 above) when set to the default.
I would be happy to help write docs/regression tests/samples on the changes,
but need guidance as to format/location/etc.

I'll attach the patch which is against the stable release source (the cygwin
lilypond source package).  Three files are affected.  The changes to
a2-engraver are just to get the stems pointed correctly.

Ted
-- 
Ted Ashton (address@hidden) | From the Tom Swifty collection:
UGA Graduate Mathematics         | "We need a 10-gauge needle", Tom
Deep thought to be found at      | hypothesized.
http://math.uga.edu/~ashted      |

patch1
Description: Text document