Urs Liska wrote :
I've always found that beams with three or more beams look too tight,
but now reading Gould, p. 19 confirms that, and I consider this a
type-ugly bug:
According to Gould outer beams starting with the third should move away
from the notehead, and "beams should never be closer to the notehead
than the correct position of the semiquaver beam (2 1/2 spaces)".
Attached you'll find examples for three and four beams, and I expect
this to continue with more beams. I have colored the ones too close in
red and adjusted them (blue) to Gould's minimum distance.
It seems:
Thanks for adding the tracker item.
The discussion at https://lists.gnu.org/archive/html/bug-lilypond/2016-02/msg00000.html contains further considerations about the proper distance. Maybe the to-be-extended length should be modified and made slightly smaller in some cases. However, currently it seems impossible to achieve the best results, even with manual overrides. It seems the beam quanting code prevents several constellations, among them the desired one.
In general I'd summarize:
The distance of a semiquaver beam is usually correct by default. But with a third beam (and more) the inner beam should be identical to the semiquaver, and additional beams have to be added to the outside (I haven't inspected the code yet, but it seems this should be the way to go).
Maybe I will work on that if I should successfully finish the beaming pattern rewrite.