Re: [Lilypond-auto] Issue 2583 in lilypond: Hairpins with to-barline=##f do not close at the end of a staff

[lilypond]

Updates:
        Summary: Hairpins with to-barline=##f do not close at the end of a staff
        Labels: Regression Patch-new

Comment #3 on issue 2583 by address@hidden: Hairpins with  
to-barline=##f do not close at the end of a staff
http://code.google.com/p/lilypond/issues/detail?id=2583

Version 2.12.3 got this right, but I haven't found when things went wrong.

Hairpins use a callback linked to their 'after-line-breaking' property  
(called just after line breaking) to decide if they need not be printed.   
The problem is that bar numbers and tempo marks also use  
the 'after-line-breaking', to find the top staff.  In order to find the top  
staff at the location of a given bar number, accounting for temporary  
staves, Lilypond finds the horizontal extent of each staff in the system.   
In order to find those extents, she asks each grob for its width, and to  
answer that, the grobs figure their stencils.

If the bar-number 'after-line-breaking' routine gets called before the one  
for a Hairpin, a Hairpin will be asked to create its stencil before its  
continuation hairpin has decided if it will print.

A solution is for each broken Hairpin to explicitly ask its continuation to  
complete the 'after-line-breaking' on time.  LilyPond claimed that this  
caused a cyclic dependency, but she is wrong.

I'm pretty sure there is a typo in the cyclic dependency checker, but would  
appreciate a look from someone who knows Guile.
<http://codereview.appspot.com/6449126>