Re: Issue 355 in lilypond: \repeat bar lines get messed up

[lilypond]

Updates:
        Owner: percival.music.ca

Comment #1 on issue 355 by markpolesky: \repeat bar lines get messed up
http://code.google.com/p/lilypond/issues/detail?id=355

Neither of these workarounds is fully sufficient.

The \partial command seems to have the effect of nullifying
the "reset" properties of the preceding barline, while still
printing it.  I can see at least three consequences of this:

1) The currentBarNumber value doesn't get incremented at the
   moment where the \partial command was issued.
2) The bar number will only print when the value passed to
   the \partial command is "moment 0".
3) Any notes following the \partial command behave as if
   they are part of the previous measure, eg. accidentals
   are wrongly suppressed if they are found in the previous
   measure.

To demonstrate this, I've modified the above workarounds:

To the top of each relative block, add:
  \override Score.BarNumber #'break-visibility = #'#(#f #t #t)

Change the following notes to "fis":
  1) the last note of the \repeat block, and
  2) the first note of the first \alternative block, and
  3) the first note of the second \alternative block.

So the first workaround becomes:

\new Staff \relative {
  \override Score.BarNumber #'break-visibility = #'#(#f #t #t)
  \partial 4
  \repeat volta 4 { e4 | c2 d2 | e2 fis2 | }
  \alternative {
    { fis4 g g }
    { \partial 4*0 fis4 a a a }
  }
  a b2.
}

and the second workaround becomes:

\new Staff \relative {
  \override Score.BarNumber #'break-visibility = #'#(#f #t #t)
  \partial 4
  \repeat volta 4 { e4 | c2 d2 | e2 fis2 | }
  \alternative {
    { \partial 4*3 fis4 g g }
    { fis4 a a a }
  }
  a b2.
}

Now the first one, at the start of the last ending:
1) fails to print the accidental, and
2) fails to increment the currentBarNumber sign

And the second one, at the start of the first ending:
1) fails to print the accidental,
2) fails to increment the currentBarNumber, and
3) fails to print the bar number (since the moment is not 0)

Currently the docs suggest workaround #2 above:
  NR 1.4.1 Long repeats * Normal repeats

And this may have misled at least one user recently:
  http://lists.gnu.org/archive/html/lilypond-user/2010-05/msg00037.html

Of the two workarounds, #1 (with the "\partial 4*0") is
superior, since it causes every measure to start with a "0"
moment.  Until someone makes the \partial command smarter, I
think we should make this a known issue and recommend
forcing the affected accidentals with "!".  And here's a
workaround to force the bar number to increment, using a
function (to be entered immediately after the \partial
command):

incrBarNum =
\applyContext
  #(lambda (x)
     (ly:context-set-property!
      (ly:context-find x 'Score)
      'currentBarNumber
      (+ 1 (ly:context-property x 'currentBarNumber))))

And here's the example rewritten with both workarounds:

\new Staff \relative c' {
  % \override Score.BarNumber #'break-visibility = #'#(#f #t #t)
  \partial 4
  \repeat volta 4 { e4 | c2 d2 | e2 fis2 | }
  \alternative {
    {
      % do not use \partial here
      fis4 g g  % do not use a bar check here
    }
    { \partial 4*0 \incrBarNum fis!4 a a a | }
  }
  a b2. |
}

One thing I can't figure out is how to get the bar number
for measure 1 to print.  If anyone knows, let me know.

Should I incorporate this in the docs?  Would someone like
to comment on the feasibility of making the \partial command
smart enough to render these workarounds unnecessary?

- Mark

Attachments:
        partial.png  19.0 KB