As you know, at present Lily assigns an initial space to each note
head (arithmetic_multiplier = 0.9 * \quartwidth;), then increases
that space by a factor (arithmetic_basicspace = 2.;) based on the
ratio of the note length to the shortest note in the measure. The
bar-by-bar approach has merit, of course, and is very useful the way
it is. But, one thing I'm seeing repeatedly with Lily setting
Scarlatti is a weakness of having only this approach to spacing.
When one bar is a cadenza-like run of 32nd notes and the next bar
only 1/4 notes or longer, the space given to the (musically minor)
32nd notes is equal to the quarter notes in the next bar (both being
the shortest in their bar). It would be easier to read, in that
spacing would match musical importance, if the space given to really
short notes were shorter than long notes, same bar or not.
One possible approach would be to have a second multiplier factor,
based on the actual length of the note rather than solely its length
relative to others in the same bar. I don't see any reason why it
couldn't work in exactly the same way as the current
arithmetic_basicspace, and blend with it.
(Of course, if you want, you could rename these factors to something
like "basicspace" and "barmultiplier", and make the new one
"durationmultiplier" so their names would match their function. But,
names don't worry me personally as long as I know what they really
mean.)
John
