Re: page breaking

[Joe Neeman]

On Fri, 2006-06-09 at 12:18 +0200, Han-Wen Nienhuys wrote:
> Joe Neeman schreef:
> > Here are the final differences between my working copy and CVS lilypond.
> > To use the new page breaker, add "Page_turn_engraver" to the staff which
> > is relevant for page turns, then add
> > 
> > #(define page-breaking ly:optimal-breaking)
> > 
> > to your (book-wide) paper block. Page turns will be considered when
> > there is a gap in notes of at least a semi-breve (whole note). To change
> > this, use the context property Staff.minPageTurnLength, eg.
> > 
> > \set Staff.minPageTurnLength = #(ly:make-moment 2 1)
> > 
> > The page turn algorithm will consider putting blank pages in the middle
> > of the score to help with page turns -- you can set the penalty for
> > blank pages in the (book-wide) paper block with the variable
> > "blank-page-force" (default is 10) or "blank-last-page-force" (default
> > is 0).
> > 
> > Also, the paper-block variable "page-spacing-weight" controls the
> > relative importance of page-spacing and line-spacing. Default is 1,
> > higher values will make the page-spacing more important.
> > 
> > (Note: this page-breaker doesn't support break-before. Use \pageBreak or
> > \noPageBreak at the end of the previous score instead)
> > 
> 
> immensely cool. I will try to review before monday. Bug me if you 
> haven't heard anything by tuesday.

There's no rush -- I have exams next week anyway.

> 
> >         }
> >       spacer.solve ((b == 0) ? line_len - indent : line_len, ragged);
> > -     force[b * breaks.size () + c] = spacer.force ();
> > +
> > +     /* add a (convex) penalty for compression. We do this _only_ in 
> > get_line_forces,
> > +        not get_line_configuration. This is temporary, for backwards 
> > compatibility;
> > +        the old line/page-breaking stuff ignores page breaks when it 
> > calculates line
> > +        breaks, so compression penalties can result in scores (eg. 
> > wtk-fugue) blowing
> > +        up to too many pages. */
> > +     Real f = spacer.force ();
> > +     force[b * breaks.size () + c] = f - (f < 0 ? f*f*6 : 0);
> >  
> 
> Am  I missing something here? Shouldn't this be f > 0 ?  It look

No, force < 0 is compression. I think this is the same as it was before
I refactored the spacer stuff.