Re: [Lilypond-auto] Issue 3518 in lilypond: Support temporary divisi staves

[lilypond]

Updates:
        Owner: address@hidden

Comment #8 on issue 3518 by address@hidden: Support temporary divisi staves
http://code.google.com/p/lilypond/issues/detail?id=3518

> > this can't work perfectly due to the rigid structure of our backend.

> Speak for yourself; LilyPond's back end had no difficulty considering  
> several hypothetical line-breaking possibilities, and figuring which  
> staves would be removed in each hypothetical case.

It's probably a good idea to actually read what you are replying to.  The  
problem is _exactly_ that LilyPond considers several hypothetical _good_  
line-breaking possibilities and figuring which staves would be removed in  
each hypothetical case.  But the best solution will often be one where the  
line breaks would be _bad_ (with overfull lines) before removing the  
unnecessary staves and respacing without the need to align to the material  
in the now removed staves.

So we'll generally get looser spacing than optimum.

I see that you changed the issue to "Started" without declaring yourself to  
be an owner of the issue.  I'm working on the issue, so I'm putting myself  
tentatively as the owner.  That does not mean that either should stop  
working/thinking on the issue, just that it makes sense to communicate the  
progress one is making.

I've stated wanting to use the \addQuote mechanism, but quotes only pick  
off a single voice, and the not-complex-enough-to-need-divisi-staves  
passages might well be complex enough to need several voices.  So that part  
of the implementation idea seems to be too restrictive.

The comment at the head of the regression test has now grown to be
% The divisi actions are multiple.  For one thing, the unisono events
% have to be diverted into the divisi staves without affecting their
% will to die without events on their own.  For another, the divisi
% staves have to be removed if empty, and the unisono staff must only
% survive if none of the divisi staves survive.

% to work with divisi, we work with a number of parallel staves, most
% of which will get removed.  One staff is the non-divisi staff which
% will get removed whenever the divisi staves don't survive.  Another
% staffgroup consists of the "true" divisi staves used for deciding
% when there is divisi material.  This group will _always_ be removed.
% They will collect the "items-worth-living" actually used for the
% decision of keeping or removing the divisi staff.  Finally, there is
% the actually typeset divisi StaffGroup that carries the same
% material as the non-divisi staff, properly formatted for use in
% divisi staves.

Most of that can be done by juggling material around in Scheme engravers.   
There probably will be the need to add a limited amount of code to  
hara-kiri-group-spanner.cc in order to access/generate the right  
termination conditions.

--
You received this message because this project is configured to send all  
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings