Re: regression? [was Re: BeatLength and BeatGrouping]

[Carl Sorensen]

On 9/29/09 7:25 PM, "luis jure" <address@hidden> wrote:

> 
> 
> thank you carl for your answer. i still think that 2.13.4 has some
> beaming issues.
> 
> 
>> We've had a serious discussion about whether eighth note beams should
>> end on every beat or only on beats 2 and 4.
>> 
>> We came down on the side of ending eighth note beams on 2 and 4,
>> because it seemed to fit Stone's rules.
> 
> OK, i agree that in 4/4 eight notes can be beamed in groups of two
> every quarter note or in groups of 4 every half note. i accept as
> correct beaming in groups of four.
> 
> *** but only as long as you only have 8th notes, and not when smaller
> values like 16ths are involved ***
> 
> in fact, the message you directed me to suggests the same:
> 
>> http://article.gmane.org/gmane.comp.gnu.lilypond.devel/23515
> 
> (see last paragraph by Trevor Daniels)
> 
> now, 2.13.4 correctly beams every quarter note when 16th notes are
> present, **except in one case:**  a 16th rest followed by a dotted 8th
> (see second bar in the attached image).

There's actually another case that can lead to failure.  That is when the
16th notes come in the 2nd or 4th beat of the measure, e.g.

a8 a a16 a a8

which will beam as

a8[ a a16 a a8]

instead of 

a8[ a] a16[ a a8].


> 
> this is inconsistent with the other cases and i see it as a bug. i
> think the dotted eight should be detached from the following eight
> notes.

Please file a bug report containing only the second measure of your example.
It should have both the autobeamed result and the desired correct beaming.

This will not be a simple bug to fix; the reason the bug exists is because
there is no 1/16 stem in the beam; all three stems are 1/8, so the 1/8
beaming rule applies.  I'm not saying it should; I'm just describing the
fact that the current LilyPond infrastructure doesn't support making the
correct decision in this case, so the infrastructure will need to be changed
somehow.

As a temporary workaround, please use manual beaming to achieve your desired
beam structure when the autobeaming fails.

Thanks,

Carl