beaming regression

[Andrew Hawryluk]

Attached is the output of LilyPond 2.7, which was benchmarked as a
near match against Baerenreiter BA320. The current beam positions do
not match: the stems are now shorter, on average. (The third measure
also has a beaming error of another kind: the first two beats should
be grouped separately from each other.)

\relative c {
  \clef "bass"
  \key d \minor
  \time 3/4
  \mergeDifferentlyDottedOn
  << {\slurDashed d8.-\flageolet( e16) e4.-\trill( d16 e)}
     \\ {d4_2 a2}
  >>
  \slurDashed
  <f' a, d,>4. e8( d c)
  \slurSolid
  bes g' f e16( f g_1 a_2 bes_3 d,_2)
  \slurDashed
  cis4.-\trill b8_3( a g)
  << {\slurDashed d'8.( e16) e4.-\trill( d16 e)}
     \\ {<f, a>4 a2}
  >>
}

Andrew

lily-2.7-sarabande.pdf
Description: Adobe PDF document

lily-2.13.4-sarabande.png
Description: PNG image