Re: Piano duets on facing pages - something new in 2.8.x?

[Urs Liska]

Am 02.08.2010 17:27, schrieb David Rogers:
> * Urs Liska <address@hidden> [2010-08-02 10:46]:
>
> > Am 30.07.2010 11:57, schrieb Werner LEMBERG:
> > > > > Is it possible to synchronize page breaks on facing pages of piano
> > > > > duet music with primo on the right-hand side and secondo on the
> > > > > left-hand side?
>
>
> ...
>
>
> > It might seem obvious, but of course the two scores must contain 
> > compatible music.
> > Does this mean that both sides must have the same amount of music?
> > While this is conceptually self-evident it is easy to write input 
> > files that don't comply to it.
> >
> > So, how would we expect LilyPond to deal with this situation?
> >
> >   * Should it just try to make the best of the input and spit out
> >     errors if it doesn't work out?
>
> I can imagine a duet where one of the players "falls asleep" and the
> other continues playing for a while. I guess in such a case it would be
> possible either to have the first player's part stop, or just fill it up
> with rests...
Well, I think, in this case, one would generally just use 
MultiMeasureRests - or did I miss something?
> >   * Or should LilyPond make checks before trying to typeset the music?
> >   * Should there be the possibility to manually insert checks of some
> >     kind, similar to bar checks and barnumber checks?
> >     What could they be like?
>
> I don't regularly typeset music without bar lines, but doesn't Lilypond
> already have facilities for synchronizing two players' parts in one
> score? Probably those can be re-used - as much as possible anyway.
AFAIK LilyPond has a notion of a point in time - which is also used in 
polymetric notation.
So from the perspective of keeping track of the music such a duet score 
could be implemented very similar to a normal score with two 
(potentially independent) staff groups.
> I think in the majority of cases having the two parts follow each
> other's bar checks, and give an error when that fails, should be
> sufficient.
I don't think so.
Bar checks may be inserted in any voice of any staff at any time. It 
will fail if in the actual context there is no barline. If the 
voices/staves have different meters, times etc. the bar checks will be 
independent. This doesn't have to be changed for a duet score.
> Would some new kind of "sync point" be needed, or does Lilypond
> already have such a thing for use in \cadenza sections and so on?
I think, everything needed is already there.
Now, while writing all this, I realize that from the perspective of the 
music the two scores are in fact one score. The only thing that has to 
be changed is the layout.
Therefore I would suggest the "scores" to be rather something at the 
level of a StaffGroup. I'll refine (and redefine) my example syntax from 
my previous email:

\score {
  <<
    \new LeftScore<<
      \new PianoStaff = "secondo"<<
        \new Staff = "upSecondo" {
          % ...
        }
        \new Staff = "loSecondo" {
          % ...
        }
      >>
    >>
    \new RightScore<<
      \new PianoStaff = "primo"<<
        \new Staff = "upPrimo" {
          % ...
        }
        \new Staff = "loPrimo" {
          % ...
        }
      >>
    >>

  \layout {}
  \midi {}
}

Everything that has to do with the organisation of the content, of the 
musical structure can just be done as usual.
If I want to get a normal score with both instruments, I just could 
replace the two new Contexts with \new StaffGroup.

What LilyPond would have to do in addition to its normal work is (only):
- individually determine and keep track of possible break points for 
both parts
- find a breaking solution that works well for both parts.
- correctly deal with odd and even page numbers.
> >   * In classical four hand music, both players have the same amount of
> >     bars in the same \time
> >     But of course this is no requirement for four hand music. Any
> >     constellation of polyrhythmic, -metric, -timing music that can be
> >     produced with LilyPond can be also used in four hand music.
> >   * So I think LilyPond should just try to find synchronous possible
> >     page breaking points and don't care about things like barnumbers 
> > etc.
>
> At least in those cases where the music is sufficiently simple and/or
> regular, corresponding pages should have the same number of lines, and
> corresponding lines the same number of bars. These requirements should
> only be broken when they will make the score bad for some other reason.
> (And then they should be followed anyway IMO.)
Well, this is yet another question.
It is generally a subject of discussion (or a matter of taste) whether a 
piano duet score should always have identical line breaking on both sides.
While identical page breaking is a practical necessity, the line 
breaking is more of an editor's choice.
So, if identical line breaking should be implemented at all it should 
absolutely be an option - although it could be argued whether it should 
be turned on or off by default.
>