Originally created by: *anonymous
Originally created by: benko....@gmail.com
Originally owned by: benko....@gmail.com
opening a new issue from the discussion in
https://code.google.com/p/lilypond/issues/detail?id=2470
would be great if someone could quote standards about how to beam tuplets, or, more specifically, whether there are cases when beaming can (should) be different from tuplet spanners.
beaming together tuplets of eighthnotes may be particularly confusing if the tuplets involve quarters (which can't be beamed).
a simple (but perhaps too drastic) way would be to set all beamExceptions in time-signature-settings.scm to have ((1 . 12) . (3 3 ...)) as smallest setting, but this will certainly change behaviour of sixteenth patterns in, say, 2/2, 3/2 or 9/4 times.
Originally posted by: colingh...@gmail.com
Labels: Type-Ugly
Originally posted by: k-ohara5...@oco.net
If you are talking about the beaming in
{ \time 6/4
%\set Timing.beamExceptions = #'((end . (((1 . 12) . (3 3 3 3 3 3)))))
%\set Timing.beamExceptions = #'((end . (((1 . 8) . (2 2 2 2 2 2)))))
\times 2/3 { g'8 a' b' } \times 2/3 { g' a' b' }\times 2/3 { g' a' b' }
g'8 b' g'8 b' g'8 b'
}
that is because LilyPond beams 6/n time as two beats of three 1/n-notes. That structure is appropriate for 6/8 and 6/16, but maybe not for 6/4. Similarly, maybe 9/4 time should not be considered to have 3 beats.
If that is what this issue is about, then try the settings in the comments and see if it looks wise to make either one default.
Originally posted by: k-ohara5...@oco.net
Simpler, maybe, would be to define a beatStructure for 6/4, to replace the default (3 3), if tuplets should not be beamed across quarter notes.
\set Timing.beatStructure = #'(1 1 1 1 1 1)
Originally posted by: benko....@gmail.com
my (fully theoretic) preference is to beam a 6/4 bar like two 3/4 bars (and a 9/4 one like three 3/4 ones).
Are there any counterexamples out there (either in engraving books or in actual engraved music)?
Originally posted by: PhilEHol...@googlemail.com
Gould says that a beam in a 6/4 bar must not cross the middle of the bar, so it is effectively 2 3/4 bars for beaming purposes.
Originally posted by: benko....@gmail.com
make beams in 6/4, 9/4 and 12/4 behave like those in 3/4
http://codereview.appspot.com/6389044
Labels: Patch-new
Originally posted by: ColinPKC...@gmail.com
Patchy the autobot says: Autobeam-show-defaults has major problems: subdivided groups disappearing, tuplet brackets missing in completion-heads-unit
Labels: -Patch-new Patch-needs_work
Originally posted by: benko....@gmail.com
Patchy the autobot says:
> Autobeam-show-defaults has major problems: subdivided groups disappearing,
just like in 3/4. do we want a 6/4 bar behave like two 3/4 bars? if so, do we want to make that the other way?
> tuplet brackets missing in completion-heads-unit
not needed anymore as distinct triplets are not beamed together - that was the starting point of the issue.
Labels: -Patch-needs_work Patch-review
Originally posted by: ColinPKC...@gmail.com
Set to new to recheck my Patchy run.
Labels: -Patch-review Patch-new
Originally posted by: ColinPKC...@gmail.com
Setting back to needs-work: the test for 1/8ths in 3/4 has two bars in the original and shows 3 bars after the patch: an extra bar of 16ths beamed in 4 has appeared. Attached comparison for autobeam-show-defaults, and I see that I was wrong about subdivided groups disappearing: I would hazard a guess that each example is producing longer lines, causing the regtest to fold and then compress them to fit in the same size image as the original.
Forgive me if I'm being dense, but that quantity of change in regtests seems problematic.
Labels: -Patch-new Patch-needs_work
Owner: benko....@gmail.com
Originally posted by: ColinPKC...@gmail.com
Here's the compare image.
Originally posted by: benko....@gmail.com
that's comparing the output of master on the old version to the output pf the patch on the new version. comparing master and patch both on the new version gives me the attached.
when I change regtest (including adding new files), I do the comparison like that:
$ git checkout origin/master
$ git checkout my-branch input/regression
$ make all test-baseline
$ git checkout my-branch
$ make all check
Originally posted by: PhilEHol...@googlemail.com
Attachments of this size can start to cause us problems. Google are strange in that they allow almost infinite disk space for almost everything, but we currently have only 49 megs left for attachments - so this thread has taken over 1% of our available space. If you can find a better way of posting - smaller PNGs, links to other sites, etc., it would be helpful.
Originally posted by: benko....@gmail.com
I don't mind removing my attached pictures - they are generated from
either regtests or attaches small .ly files.
p
Originally posted by: PhilEHol...@googlemail.com
Removing them once they're attached does not affect our quota, so please don't bother.
The considerations on beam subdivision in [#5547] are possibly pertinent.
Related
Issues: #5547
Last edit: Simon Albrecht 2019-07-31