Re: rests in polyphonic music - 'stems down, rests up'

[andersvi]

Keith, thanks for the pointers, and a nice solution using stemNeutral.

>>>>> "K" == Keith E OHara <address@hidden> writes:

    K> It is less clear if the use of Stem 'direction to place rests, in
    K> the notes-plus-rests case, was intentional.

Shouldn't the rests by default be placed where the voice goes in the
column - odds up, evens down - regardless of stem-direction?

The call to 'get_grob_direction isn't used in the notes+rest case inside
rest-collision.cc, and i guess thats why Rest 'direction doesn't apply
there.  Do you think it could be plunged in while running through the
rests to give priority for any explicit directions vs. rests?  

    K> For Rachmaninoff's Prelude in C-sharp minor, ....  You could let
    K> the stems take their natural direction,

Indeed.  Apart from the placement of rests leaving the defaults in place
gives very good results.
    
    K> and since rest-collision.cc insists on asking your rests what is
    K> their Stem direction, and since un-pitched rests in a stemNeutral
    K> voice answer based on neutral-direction :

I hadn't noticed this clause about un-pitched rests in stemNeutral
voices.  Is their behaviour vs. 'neutral-direction documented anywhere?

Thanks alot.

-anders