Snaps horizontally-offset fingerings to vertical column.

Snaps horizontally-offset fingerings to vertical column.

Does this via the positioning-done callback if Fingerings are within
FingeringColumn.snap-radius of each other along the X axis.

[MikeSol, 2013-03-24 07:18:46]

Adds double flat to regtest

[Ian Hulin (gmail), 2013-03-24 14:00:44]

Suggestion for regression test doc-string.
It's a bit wordier but clearer and shows you're covering cases for other wide
accidentals on the note being fingered like half-flat etc.
Cheers, Ian

https://codereview.appspot.com/7988043/diff/2001/input/regression/fingering-c...
File input/regression/fingering-column-snap-radius.ly (right):

https://codereview.appspot.com/7988043/diff/2001/input/regression/fingering-c...
input/regression/fingering-column-snap-radius.ly:6: The default being @code{0.3}
allows the snap to happen for a sharp but
"The default value of @code{0.3} allows the snap to happen when the fingering is
applied to a note with a sharp but"

https://codereview.appspot.com/7988043/diff/2001/input/regression/fingering-c...
input/regression/fingering-column-snap-radius.ly:7: not a double flat.
not to happen for notes with wider accidentals like double flat.

[janek, 2013-03-24 21:58:32]

Another patch i understood in first reading :)
LGTM
Janek

[Keith, 2013-03-25 04:42:43]

Code looks good, but we could get the same result with
  \override Accidental #'horizontal-skylines = #'()

Incorporating the detailed outline of the accidental into the skyline of the
chord costs time, then the iterative alignment of fingering costs more time, and
if that alignment does anything it has the same effect as skipping the detailed
outline in the first place.

Does anything else (ties, maybe?) benefit from using the stencil of accidentals
for horizontal-skylines ?

https://codereview.appspot.com/7988043/diff/2001/input/regression/fingering-c...
File input/regression/fingering-column-snap-radius.ly (right):

https://codereview.appspot.com/7988043/diff/2001/input/regression/fingering-c...
input/regression/fingering-column-snap-radius.ly:6: The default being @code{0.3}
allows the snap to happen for a sharp but
On 2013/03/24 14:00:44, Ian Hulin (gmail) wrote:
> "The default value of @code{0.3} allows the snap to happen when the fingering
is
> applied to a note with a sharp but"

That comment does not correspond to the behavior of the patch (nor does Mike's).
 All accidentals are wider than 0.3 spaces.  The question is the shape of the
outline of the left edge of the accidental
  <cis-1 a-1 e-1>4
  <cis!-1 a-1 d,-1>4
  <ceses-1 a-1 d,-1>4
  <ceses!-1 f,-1 d-1>4

[mike7, 2013-03-25 08:40:10]

On 25 mars 2013, at 06:42, k-ohara5a5a@oco.net wrote:

> Code looks good, but we could get the same result with
>  \override Accidental #'horizontal-skylines = #'()
> 
> 

Conceptually, I'm a fan of the snapping idea, but you're right that it takes
more time.  Would this \override fix all problems of a lack of "snapping"? 
Would it force accidentals to be that should be staggered?

If your idea gives a satisfactory result, we can go with that and keep my patch
around in case one day we need something more fine-tuned.

Cheers,
MS

[janek, 2013-03-25 08:43:15]

On 2013/03/25 04:42:43, Keith wrote:
> Code looks good, but we could get the same result with
>   \override Accidental #'horizontal-skylines = #'()
> 
> Incorporating the detailed outline of the accidental into the skyline of the
> chord costs time, then the iterative alignment of fingering costs more time,
and
> if that alignment does anything it has the same effect as skipping the
detailed
> outline in the first place.
> 
> Does anything else (ties, maybe?) benefit from using the stencil of
accidentals
> for horizontal-skylines ?

Accidentals themselves use skylines to position themselves snugly:
http://code.google.com/p/lilypond/issues/detail?id=2811

Also, in tight music there are other grobs that may go close to accidentals and
benefit from precise positioning.

best,
Janek

[Keith, 2013-03-26 02:25:29]

LGTM, after removing the erroneous part of the regtest description.

On 2013/03/25 08:43:15, janek wrote:
> Accidentals themselves use skylines to position themselves snugly:
> http://code.google.com/p/lilypond/issues/detail?id=2811
> 

Right. We decided it worth the time to have detailed skylines for 
  { % \override Accidental #'horizontal-skylines = #'()
    <es' a'!> }

> Also, in tight music there are other grobs that may go close to accidentals
and
> benefit from precise positioning.

I would have thought that Fingering was an example that would benefit, but
people seem to want them lined up in boxes.
> best,
> Janek

https://codereview.appspot.com/7988043/diff/2001/lily/fingering-column.cc
File lily/fingering-column.cc (right):

https://codereview.appspot.com/7988043/diff/2001/lily/fingering-column.cc#new...
lily/fingering-column.cc:161: while (dirty);
After the first pass, we have
     fos[i].offset == fos[i+1].offset 
     || fabs(fos[i].offset - fos[i+1].offset) >= snap
so any further passes through the loop do nothing.

[MikeSol, 2013-03-26 06:11:24]

Removes dirty loops.

[janek, 2013-03-27 21:40:06]

LGTM

https://codereview.appspot.com/7988043/diff/14001/lily/fingering-column.cc
File lily/fingering-column.cc (right):

https://codereview.appspot.com/7988043/diff/14001/lily/fingering-column.cc#ne...
lily/fingering-column.cc:140: 
I think the comment in previous patchset was helpful, no need to delete it.