On Sun, May 10, 2009 at 6:26 PM, Werner LEMBERG <w...@gnu.org> wrote: >> LilyPond does not use a hardcoded set of rules; rather, it tries to >> score different configurations, and pick the one with the best >> scores. This is a technique that works much better than hardcoding >> different rules, but in some cases there are small divergences with >> the 'prescribed' quants. We even have a regtest for it, see >> input/regression/beam-quant-standard.ly >> >> It might be possible to tune the scoring parameters to copy Ross >> exactly, but probably some other configurations will fall over. > > I can imagine that after the scoring there's a second pass to > fine-tune the result, for example, to `snap' various parts of the beam > to the staff lines where possible.
No, that would not work; the snapped position was also scored and did not win, so in many cases there is something wrong with it. -- Han-Wen Nienhuys - han...@xs4all.nl - http://www.xs4all.nl/~hanwen _______________________________________________ bug-lilypond mailing list bug-lilypond@gnu.org http://lists.gnu.org/mailman/listinfo/bug-lilypond
