On Fri, Dec 19, 2025 at 1:42 PM Luise Flesch <[email protected]> wrote:
> Hi Aleksa! Thanks for the answer :) > > I've been digging into everything (and I MEAN everything. I think the only > place I didn't look for information was in the freaking subreddit) lilypond > online to see if there was anything I could do, and I think I've found the > issue. See this regtest (beam-quant-32nd.ly): > [image: image.png] > "Stem lengths take precedence over beam quants" (I think I understand what > beam quanting is? The quants are like, the points of the staff that the > beam ends go on. so forbidden quants are beam ends ending up in the middle > of a staff space. might be misinterpreting though). So this is on > *purpose*, to prioritise good stem lengths over good beam slants (and > indeed, you can see beam ends in staff spaces like at least 3 times in this > very regtest). Why, I don't heckin know lmao. > I think the point of this regtest is to demonstrate that for triply-beamed stems, lilypond chooses to put beams in bad locations in order to avoid a worse problem of stems that go too far outside of the staff. This is showing a design decision. I'm not saying it's a proper decision, but that it is a design decision. Given that, I think the design needs to be changed to solve the problem you are having. I have done some looking in old versions. This same regtest in lilypond 2.2 has no beams end in spaces, as far as I can see. https://lilypond.org/doc/v2.2/input/regression/out-www/collated-files.html Perhaps we missed a change in a regtest a LONG time ago and need to fix a bug that was introduced some time after 2.2. Between 2.2 and 2.4, the beaming in the last pair of measure 2 changed from horizontal to 1/2 staff space, and that ended up with a beam in the center of a space. Or perhaps some of the stem lengths in 2.2 that prevent unallowed 32nd beam positions are worse than having the beams end in the spaces. Note that beam-quanting-horizontal shows that lilypond respects Gould's rules when beams are horizontal. When the slope between two notes is a second (one half staff space), then it's impossible to follow Gould's rules with a beam that reflects the slope between the heads. It must either be a horizontal beam, or a full-staff-space slanted beam to maintain Gould's rules. It should be possible to bisect if needed and find the change that shifted from horizontal to half-space-slanted beams between 2.2 and 2.4 HTH, Carl >
