Re: 2.17.16 vs 2.17.17 Page Breaking

[Keith OHara]

On Wed, 01 May 2013 21:34:46 -0700,  wrote:

> On Wed, 01 May 2013 21:07:52 -0700, Jay Anderson <address@hidden> wrote:
>
> > On Wed, May 1, 2013 at 12:15 AM, Keith OHara <address@hidden> wrote:
> > > Nothing was intended to change page-breaking between those versions.
> > >
> > > However, almost anything changing layout will change the page-breaking
> > > when pages are close to full.
> > >
> > > If one of these things puts the page-breaking back to the old way
> > >   \markLengthOff
> > >   \override Score.RehearsalMark #'Y-offset = 0
> > >   \override Score.MetronomeMark #'Y-offset = 0
> > > then something in my change to give space to tempo marks has caused trouble.
> >
> > (moving to the bug list)
> >
> > Git bisect led to commit b6f94447415dded7c6e146b41b6139fe76cb84c4
> > (http://git.savannah.gnu.org/gitweb/?p=lilypond.git;a=commitdiff;h=b6f94447415dded7c6e146b41b6139fe76cb84c4).
> > I do have a '\tempo 4=108' in the score (for midi purposes), but it is
> > hidden with

I cannot reproduce the full problem.  Has anyone seen LilyPond print 
'annotate-spacing' output where both the limits for the top staff are negative 
numbers?

When the top staff ends at 0, the commit cited above gives generally better 
page-breaking estimates of the height of the staff (see  
<http://code.google.com/p/lilypond/issues/detail?id=3342>).  The estimate is a 
bit worse in the case where it allows space for a tempo that is completely hidden.  
The reason given for hiding the tempo completely was that it is for midi only, but 
the manual says in that case to use  \midi {\tempo...}

I think I have an implementation of \markLenghtOn that will avoid this problem, 
but if we don't confirm that, then people setting score and parts will space 
the tempo marks in the parts by hand
<https://lists.gnu.org/archive/html/lilypond-devel/2013-02/msg00075.html>