RFC: Mechanism for splitting/joining autobeams

[Lukas-Fabian Moser]

Folks,

it has bothered me for some time now that if, at some individual point in a piece, I'm not satisfied with LilyPond's beaming decisions, I have to resort to specifying the full beams manually.

Example: If for some reason, the manuscript I'm copying has exceptional beams of 4+2+2 instead of the usual 4+4 as in the second bar of

from Scarlatti's Partimenti (and I want to reproduce that beaming faithfully), I have do to:

c8 c' b a e[ e'] cis'[ d']

So basically I have to enter redundant information: I have to tell LilyPond to

1) start a beam on e,
2) end a beam on e',
3) start another beam on cis',
4) end that beam on d'.

Out of these, only 2) & 3) should be necessary, and actually, those could be condensed further to

0) Split the autobeam between e' and cis'.

I implemented this, and a corresponding mechanism for joining auto-beams, in a draft merge request to LilyPond proper: https://gitlab.com/lilypond/lilypond/-/merge_requests/2668

With that addition, I can typeset the second bar above as:

c8 c' b a e e' \| cis' d'

Unfortunately this is not available for testing at the moment for people who do not self-compile LilyPond.

Nevertheless, I'd like to ask for feedback from the community: Would such a feature be welcome?

I, for one, have stumbled over this missing feature (that's basically equivalent to MuseScore's "Beam start" and "Beam middle" features) so often that I finally sat down and implemented it. But of course it's not completely obvious that such an addition would actually be beneficial, as it introduces new syntax elements (at the moment: \| for splitting and \_ for joining auto-beams) for something that can, in effect, be created by current LilyPond as-is.

Lukas