Re: Stem lenghts

[Werner LEMBERG]

> My point is that the stem lenghts should lead the eye: if the note
> goes down the stem goes down, if the note stays the same the stem
> stays the same.  Ross book maybe a good starting point to this.

Noone is questioning this.

> My original example would be fine if the phrase ended on the first
> beat, but it is supposed to end on beats two and four.  The stem
> lenghts suggest different pitches on the ending note.

I haven't looked at the code but I can imagine that there is still
room for optimization (at the cost of speed, however).

> I have attached an example of Grieg, edition Eulenburg that shows
> what I mean.

Not really.  The problem with this sample is that this Eulenburg
edition is far too small to really show high typography.  While it
leads the eye well, it exhibits far too much other typographical
defects, for example the second and last beam, both presenting those
small white triangles which should *never* be present in former times
since they can be flooded with ink as soon as the printing plate gets
used a bit.  If you want to show us good typography please refer to a
large score for a piano solo piece, say.

> I can quote a dozen examples just by opening scores at random. The
> lilypond_output.png leads the eye differently every time (at least
> my eye). :)

You can help us a lot if you define this behaviour algorithmically,
casting the rules into something which can be implemented in C++, then
comparing the results with lilypond's output, pointing out the
differences in an analytical way.


    Werner