Re: Does the optimal-page-breaker work?

[Keith OHara]

Werner LEMBERG <wl <at> gnu.org> writes:

> >> I'm successfully using ly:page-turn-breaking with an organ piece,
> >> fitting about 200 bars on 11 pages.
> 
> I'm inserting \allowPageTurn commands: In virtually all places where a
> turn could happen, one hand plays while the other hand can turn the
> page.

This example included 14 \allowPageTurns, from which LilyPond chose the
5 best page-turns.  The music would fit on 10 pages with no regard to
possible page turns, so by using just 11 pages with good page-turns,
LilyPond does quite a good job.

ly:page-turn-breaking works, so we do get some benefit form LilyPond's
efforts to estimate how tall the systems will be when choosing the line-
breaks.

The default page-breaker still seems to get no benefit (such as smarter
choices of line-breaks) from the estimates of how many systems will fit
on a page, but suffers loose or over-full pages when those estimates are
wrong.

Shevek <saul <at> saultobin.com> writes:

> I recommend playing with system-count and systems-per-page. 
> I have found this to be the best way to get Lilypond's page breaking
> algorithms to give good results.

That does work.  By setting systems-per-page manually, we are making
the page-breaking decisions ourselves.