Remove tied accidentals after line-breaking

Remove tied accidentals after line-breaking

[Keith, 2013-12-28 05:45:36]

https://codereview.appspot.com/46060045/diff/20001/lily/accidental.cc
File lily/accidental.cc (right):

https://codereview.appspot.com/46060045/diff/20001/lily/accidental.cc#newcode46
lily/accidental.cc:46: 
Probably the next three callbacks can be removed, letting LilyPond fall back to
the defaults, but I haven't figured out how the defaults work yet.

https://codereview.appspot.com/46060045/diff/20001/scm/define-grobs.scm
File scm/define-grobs.scm (right):

https://codereview.appspot.com/46060045/diff/20001/scm/define-grobs.scm#newco...
scm/define-grobs.scm:41: (Y-extent . ,accidental-interface::height)
Probably this can be simplified, to whatever now indicates that an objects
stencil is "pure"

[Keith, 2014-08-02 18:56:19]

LGTM

[dak, 2014-08-02 19:26:11]

On 2014/08/02 18:56:19, Keith wrote:
> LGTM

Cannot find the corresponding Google Code issue.

[pkx166h, 2014-08-02 20:02:41]

On 2014/08/02 19:26:11, dak wrote:
> On 2014/08/02 18:56:19, Keith wrote:
> > LGTM
> 
> Cannot find the corresponding Google Code issue.

http://code.google.com/p/lilypond/issues/detail?id=3646

[dak, 2014-08-05 07:42:05]

Does this approach give us better insights or leads how to address issue 649?

https://codereview.appspot.com/46060045/diff/60001/input/regression/accidenta...
File input/regression/accidental-unbroken-tie-spacing.ly (right):

https://codereview.appspot.com/46060045/diff/60001/input/regression/accidenta...
input/regression/accidental-unbroken-tie-spacing.ly:20: (make-fraction-markup
(number->string (numerator alt))
The fraction looks really ugly because of the horizontal alignment of the sign. 
Rather than trying to futz around with that, I'd probably punt here and just use
(make-circle-markup (number->string alt)).  That will lead to different size
circles for basically all accidentals, but so what?  Or just override with a
large font-size.

Maybe the override should be done earlier in order to demonstrate the non-break
case working as intended as well.

[Keith, 2014-08-06 05:08:37]

This clears up issues 3749 and 3646 but my motivation was to avoid the excessive
font-lookups for accidentals.  For example
\transpose c cis { cisis } generates 5 error messages, one for each attempt to
lookup a triple sharp.

The Accidental.stencil callback had responsibility to delete the Accidental grob
if a tie made the accidental unnecessary.  So any overrides to the stencil would
need to replicate that logic, and the stencil could not be safely used for
spacing purposes before line-breaking.

Better to move this responsibility to a function called once after
line-breaking.

On 2014/08/05 07:42:05, dak wrote:
> Does this approach give us better insights or leads how to 
>  address issue 649?

No real help for addressing the question of a later accidental that is
unnecessary depending on whether an accidental is printed on a tied note.  

The accidental engraver would still need to make the links between the later
accidental and any accidentals on tied notes a that affect the later accidental.
  Right now, the engraver records the answers to "what alteration would this
notename have if printed without an accidental?" which is usually a fractional
alteration, but the symbol 'tied if the is ambiguous due the possibility of
removing a tied Accidental.  That record would need to be extended to let the
later Accidental get a pointer back to the tied Accidental.  Doing this in the
general case looks like an interesting problem.

If the engraver can record in each Accidental a link to any Accidentals it
depends upon, then remove_tied() can be extended to act on that information.

https://codereview.appspot.com/46060045/diff/60001/input/regression/accidenta...
File input/regression/accidental-unbroken-tie-spacing.ly (right):

https://codereview.appspot.com/46060045/diff/60001/input/regression/accidenta...
input/regression/accidental-unbroken-tie-spacing.ly:6: \version "2.16.0"
should be 2.19.12-ish

https://codereview.appspot.com/46060045/diff/60001/input/regression/accidenta...
input/regression/accidental-unbroken-tie-spacing.ly:20: (make-fraction-markup
(number->string (numerator alt))
I'll use
\relative c' {
  \clef treble \time 3/4
  c8 b2  <g b des f>8 ~ |
  <g b des f>8 r
  % Large accidental
  \override Staff.Accidental.stencil = #
  (lambda (g)
    (let ((alt (ly:grob-property g 'alteration)))
      (grob-interpret-markup g
        (make-circle-markup (number->string alt)))))
  bes4 ~ bes ~ | bes ~ bes bes ~ | \break
  bes ~ bes bes
}