Re: ugly ties in crossed partcombined voices

[David Kastrup]

Dan Eble <address@hidden> writes:

> Does anyone understand why the partcombine output is so awful?
>
> \version "2.19.16"
>
> \score {
>   \new Staff \with { instrumentName = "<< >>" } <<
>     \new Voice \relative { a'2~2 a2~2 }
>     \new Voice \relative { c''2~2 f2~2 }
>   >>
> }
>
> \score {
>   \new Staff \with { instrumentName = "partcombine" } <<
>     \partcombine \relative { a'2~2 a2~2 }
>                  \relative { c''2~2 f2~2 }
>   >>
> }

\voiceOne / \voiceTwo imply tie directions.  If you write

\version "2.19.16"

\score {
  \new Staff \with { instrumentName = "<< >>" } <<
    \relative { a'2~2 a2~2 } \\
    \relative { c''2~2 f2~2 }
  >>
}

\score {
  \new Staff \with { instrumentName = "partcombine" } <<
    \partcombine \relative { a'2~2 a2~2 }
                 \relative { c''2~2 f2~2 }
  >>
}

instead (which uses \voiceOne and \voiceTwo internally) you'll get the
same output above and below.  In your example, LilyPond is allowed to
choose tie directions in spite of the voice crossing.

Usually you'd want to have the higher pitched voice first _unless_ you
need to have a visual inversion (like when alto II dips below alto I).

-- 
David Kastrup