Re: horizontal spacing regression

[Carl Sorensen]

On 1/13/11 6:36 PM, "Keith OHara" <address@hidden> wrote:

> On Thu, 13 Jan 2011 16:22:53 -0800, Carl Sorensen <address@hidden> wrote:
> 
>> \override Accidental #'extra-spacing-height = #'(-0.5 . 0.5)
> 
> That works nicely, on both the original music example we looked at on -user,
> and my new favorite example:
> 
> \paper { ragged-right = ##t }
> \version "2.13.43"
> { \clef bass
> %  \override Accidental #'extra-spacing-height = #'(-0.5 . 0.5)
>    \repeat unfold 8 {f8 bes bes f }
> }
> 
> The rising line of notes is usually spaced a little further apart, when
> possible, to compensate for the optical effect oval shape of the heads, as you
> and Janek have been discussing recently.  The accidental seems to cause
> LilyPond (all versions) to revert to the spacing for a constant-pitch series.
> This makes some sense because the accidental destroys the oval-shape illusion,
> but it looks strange when the accidental is crowded and nearby notes are
> spaced with the optical correction in place.

It's in the code, I think:

from lily/note-spacing.cc

317   /*
318     Only apply same direction correction if there are no
319     accidentals sticking out of the right hand side.
320   */
321   else if (stem_dirs[LEFT] * stem_dirs[RIGHT] == 1
322            && !acc_right)
323     correction = same_direction_correction (me, head_posns);
324 
325   *space += correction;

> 
> The extra-spacing-height seems to do just the right thing.  True, it gives no
> extra space when the interval is larger {c be be f } but neither did 2.12.3
> and I think we don't need it there.

Have you explored its affects with multiple accidentals?  Does it cause
problems, e.g. with a key signature?

In your opinion, should we make it part of the Accidental default
properties?

Thanks,

Carl