Re: status of spanner IDs across voices?

[Werner LEMBERG]

>> Thanks for the hint.  This works, sort-of, but it isn't a practical
>> solution for complicated piano music, unfortunately.  In
>> particular, lilypond no longer knows where to start or end a slur
>> vertically, sometimes assigning it to the wrong voice in parallel
>> music.
> 
> You won't get anything better than "starts at right note column"
> from cross-voice spanners either.

If the right solution with the slur engraver in Voice context was

     __________
    /          \
               oo
               |

lilypond suddenly did

     ___________
    /           \
               oo
               |

with the slur engraver in Score context.  It looks to me that Voice
context information is lost in Score context (which I can understand),
making lilypond select an anchor point arbitrarily.  However, in
complex piano scores, such situations are far too common, making
manual anchoring of every slur very tedious, just to be sure lilypond
doesn't select the wrong position by some hidden algorithm.

In other words, it is completely acceptable to manually anchor
cross-voice exceptions, but the context shift of the slur engraver
produces far too much wrong positions.

> Anchor to individual notes (inside of chords) if you want that.

This will be certainly needed for cross-voice slurs.


    Werner