alpha test autobeaming

[Keith E OHara]

In "First impressions of alpha test" I wrote:
> The auto-beaming seems to beam a little less often than the
> old system.

My old scores had two situations with missing beams, neither one is a  
simple
regression:

1) swung triplets in 2/8 are not beamed.  Both 2.12 and 2.13 are missing
beams, but in different ways.  My intention is to report on bug-lilypond
after giving a bit of time to hear from other users.
\relative c'' {
  \time 6/16
  c16 c c c8 c16
  \time 2/8
  \times 2/3 {c16 c c} \times 2/3 {c8 c16}
  \scaleDurations #'(2 . 3) {c16 c c c8 c16}
}

2) Both 2.12 and 2.13 miss beams in the middle of a particular cadenza.
But my minimal example is
{ \override Fingering #'add-stem-support = ##t
  \cadenzaOn
  \repeat unfold 4 {c8 c c c \bar""}
  \cadenzaOff
}
Removing the stem-support (!) removes the problem, on the Win32 builds.
Actually, 2.12 drew all the beams in the actual score
(http://www.mutopiaproject.org/cgibin/piece-info.cgi?id=1776) but started
to drop beams as I simplified.

Number 2 does not seem to be worthwhile to chase, so I will only report at  
bug-lilypond if some other user has suffered this bug, or if developers  
want the info in bug-report form.


P.S. My advice to users using the new autobeaming: Read the manual
backwards, seriously.

In the Notation Reference, 1.2.4 Setting-Automatic-Beam-Behaviour, read
1) snippet with the measure grouping signs to learn beatStructure, simple
use of baseMoment (p74)
2) snippet with all the 32nd notes (demisemiquavers) to learn
subdivideBeams (p73)
3) section on how to apply autobeaming rules to individual voices (starts
bottom p71)
4) section on how autobeaming makes exceptions to the basic beatStructure
(p71)
5) finally, if you want, the talk-through of the code (p70)

A12.png
Description: PNG image

A13.png
Description: PNG image

cadenza.png
Description: PNG image