22-Sep-01 Han-Wen Nienhuys wrote:
HN> I originally chose the minimum-denominator criterium, because it
HN> matches nicely with the way binary meters are built up. I have my
yes, as long as you use a binary meter with no tuplets then it works.
As soon as you insert triplets it goes wrong. like in
| c4 [\times 3/2 {c16 c c} c8] c2 |
where you get a split at 1/3 - between the 2nd and 3rd note in the tuplet.
HN> reservations about "split in the middle", since this criterium does
HN> not change when the beam is moved around in the measure. For any
HN> non-trivial beaming that is correct, you can construct an incorrect
HN> one by moving the beam right by an 8th or 16th. Could I ask you to
HN> reconsider your algorithm?
Yes, I will reconsider it and return.
I still think, however, that my algorithm produces better beams than the old
one.
(You practically never encounters beams that starts AND ends at odd positions
- and if it is only one of the ends that is at an odd position then the beam
"probably" cannot be broken at the middle)
-Rune
_______________________________________________
Lilypond-devel mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/lilypond-devel
