problems in 1.3.115/placement of dynamics

[nel]

Hi,

I've just noticed that multimeasure rests are broken in 1.3.115;
also, there's a problem building the documentation: refman.itely
has a section like this:

-------
\include "paper16.ly"
\paper  { linewidth = -1.000000 \pt; }
\score {
 \notes { \transpose c'' {
  \chords {
    c1  c:3-       c:7     c:8
    c:9 c:9-.5+.7+ c:3-.5- c:4.6.8
  }
}

 }
  \paper { }
}
------

Lily 1.3.115 doesn't output any errors, but creates an empty file
(and this probably means something is wrong with lily, since this file
looks valid); when gs is run to generate the corresponding PNG,
pnmtopng stops with an error because of that. 1.3.114 processes this
with no problems.

While not really bugs, the placement of some things in lily could be
improved, IMHO. I checked some Henle and Dover scores, and the
first important difference I noticed is that lily seems to "left-align"
dynamics and noteheads, that is, they both "start" simultaneously. In
both these editions I see that they usually are more or less "centered"
together, that is, the dynamic mark starts earlier than the notehead
and extends further (especially true for longer marks, like pp, mf etc).
Actually, they are even close to "right-aligned"; I guess there's some
optical corrections in place here. The point is, sometimes these
marks in lily seem to refer to the next note, and it seems to me that
centering them would have a much better result (don't know how difficult
it is to implement that, though).

Another thing that annoys me is the placement of (de)crescendos; lily
starts the mark *after* the end of the first notehead and stops *after*
the end of the second notehead; the Henle edition starts them together
with the first notehead and stops them after the end of the second
notehead. If there are marks on the notes themselves, it changes this
behaviour: if there's a dynamic mark on the starting note, the crescendo
stays in the same place and the dynamic mark is shifted to the left; if
there's a dynamic mark on the ending note, the crescendo stops earlier
so that it doesn't collide with the mark.

Now, I suppose it is not too easy to mimic this behaviour, and it
probably isn't worth it before 1.4; but the problem is, the way lily
currently does things has some problems: if the score is too tight,
a crescendo between two consecutive notes appears to start and end
almost immediately, and don't seem to relate to the first note; and
worse, this can't be adjusted with spacing notes (at least I couldn't).
Second, if there's a mark on the second note, it is overwritten by
the crescendo mark; this can't be compensated with spacing notes on
the first beat of the measure, but may be alleviated (by making the
crescendo stop at the end of the previous measure).

I believe that "dragging" the crescendos to the left, making them start
at the *beginning* of the first note and end right before the last note
would be better: first, the fact that the crescendo doesn't go up to
the end of the second note would seem, at least to me, a much (MUCH) less
proeminent problem (actually, before I checked the editions, I thought
that would be what I'd find). This would also prevent the crescendos from
overwriting dynamics on the second note. At the same time, the "correct"
behaviour could be achieved with spacing notes. As for the starting note,
things would be correct for notes with no dynamics, but the crescendo
would overwrite dynamics marks on the first note; this would be alleviated
if the dynamics were centered, like I suggested above, but not solved;
however, the correct behaviour could also be achieved with spacing notes
(either by attaching the dynamic to a spacing note before the actual
note or by shifting the crescendo to the right with the spacing notes,
which is reasonably close to the "ideal"). Perhaps another option is
having the crescendo start at the "middle" of the notehead (??).

And, while this is probably harmless, what's the story with the message:

"warning: unknown spacing pair `Left_edge_item', `Breathing_sign'"

Should I really direct this kind of message to the "bugs" list?


Have fun everybody!
Nelson