Reintroduce beaming "begin" rules back to solve issues?

[Xavier Scheuer]

Hi Carl, dear developers,

Carl's recent comment in issue #2246 , introducing  beamHalfMeasure ,
make me think, because I have never heard of such concept as "half
measure beaming".  Is it explained that way in references like Ross,
Read or Gould?  Or is it a concept that have no musical meaning but
that is purely a "programming commodity" variable to use in LilytPond's
beaming "system" code?

LilyPond beaming rules system is really great —which is mainly due to
Carl's work— thanks and congratulations.  It is also really nice to see
Carl improving it again for the rare cases where some users want
different specific rules to achieve their expectations.

But I'm a little bit afraid that "strictBeatBeaming", "beamHalfMeasure"
introduced to fix respectively issues #2228 and #2246 might
"overcomplicate" the automatic beaming rules system (which is already
not easy to understand, especially for lambda users) and could
potentially produce unwanted/unexpected *new* issues in circumstances
other than what was the original fix for (at least that is my impression
as a user).

IIRC 2.12 beaming system made the distinction between rules allowing a
beam to _start_ ("begin" rules) and rules allowing the beam to _end_
("end" rules) at specific moment in the measure.
AFAIK the rules implemented in the new 2.14 beaming system are only
"end" rules.
Can't we (I mean, probably you, Carl  ;-P) introduce "begin" rules
again and wouldn't it potentially permit to "solve" situations like the
ones expressed in issues #2228 and #2246 in a more generic way?
Or are the concept of "strictBeatBeaming", "beamHalfMeasure" really
necessary because they would cover cases that would not possibly be
covered with "begin" beaming rules?

For example, IMVHO, the issue expressed in issue #2228 could be solved
if we forbid the start ("begin") of 16th beaming on the 4th and 10th
semiquaver in 6/8 time.
And issue #2246 could be solved by forbidding 8th beaming to start
("begin" rules again) on the 4th quaver of 3/4.

Is there something wrong with this approach?

Sorry for "late awakening", I'm following LilyPond development more and
more scarcely and the code usually goes way over the top of my head.

Thank you for your attention (and your wonderful job).

Cheers,
Xavier

-- 
Xavier Scheuer <address@hidden>