Re: Proposal: Changing tremolo beam gap implementation

[Torsten Hämmerle]

Thomas Morley-2 wrote
> P.S. I crossposted to devel, not all are subscribed to the user-list

I first wanted to ask what users in general think about the gap size.


Thomas Morley-2 wrote
> Thus I think 'gap should always specify the _white_ space.

That's exactly what I had in mind, but the current implementation behaves
differently.


Thomas Morley-2 wrote
> If 'gap <= 2 then there is white space added, exactly as specified
> (half of gap-size to left/right), measured from NoteHeads to Beam.

Yes, exactly.  But this "half of gap-size" is the answer to the following
question of yours:


Thomas Morley-2 wrote
> (1) Why should it be different if 'gap is greater than 2.0?

Let's consider the right side, it's less confusing direction-wise.
In case of semibreves (whole notes), i.e. when we have "invisible stems",
LilyPond considers the following two beam starting point alternatives:
(1) Stem x-position - gap + stem-thickness/2  (currently)
(2) Left edge of notehead - gap/2
The stem x-position for invisible stems is in the centre of the notehead
(!), that makes things a bit weird.

LilyPond will pick the alternative that's closer inside.
For small gaps, alternative (2) will be chosen, but, as gap increases,
alternative (1) is proportional to gap, but alternative (2) is only
proportional to gap/2 and for bigger gaps, alternative (1) will be picked.

The workaround with large stem widths just has the effect that LilyPond will
always pick alternative (2) and thus behave as expected.


Elaine Gould says: /"For semibreves, position the tremolo beams as if the
semibreves were stemmed […]"/ (Behind Bars, p. 226)
That would be kind of alternative (1), except that for an unknown reason
LilyPond's calculation is based on a stem starting in the centre of the
notehead (???).


Gould continues with "/It's acceptable to move semibreve beams closer to the
noteheads […]/"
That's basically alternative (2) and LilyPond's (intended) standard
behaviour. 


But first, we'll have to agree that a gap is generally supposed to be free
space. This change will also influence the whole-note beaming mechanism.
And I'd like to change the centre-stem calculation that does not make any
sense to me. Or am I missing something?

Cheers,
Torsten




--
Sent from: http://lilypond.1069038.n5.nabble.com/Dev-f88644.html