Re: Nested transposition

[David Wright]

On Sat 13 Mar 2021 at 11:18:52 (+0100), Lukas-Fabian Moser wrote:
> Am 13.03.21 um 02:16 schrieb Valentin Petzel:
> > As far as I know transpose should only really be used with absolute pitches,
> > as it would be quite impossible to do it otherwise. Think of it like this:
> > Say we are in relative mode and have c' f. Then f will be above c. If you 
> > then
> > do c' \transpose f a f, then the transposed a would be above the c – so it
> > would need an octave indication. But the octave the music has to be put into
> > depends of the previous note. So a single transpose function cannot provide
> > this.
> While that is true, I think your next two remarks are overgeneralisations:
> > So, always use absolute pitch with transpose!
> 
> \transpose and \relative live happily together as long as \transpose
> is _outside_, so it's fine to use
> 
> \transpose c d \relative e' { ... }.
> 
> In particular, it's no problem to \transpose a complete score, no
> matter what combination of absolute and relative note entry is used in
> the score.
> 
> So, I propose to rephrase your rule to:
> 
> - "Avoid using \transpose inside of \relative" ("and if you have to,
> remember to issue a new \relative block inside the \transpose").

The problem with inventing arbitrary rules is that they don't
necessarily teach the real reason behind them.

It's perfectly correct to say that what is given to transpose,
its "argument", ie the … … … in   \transpose X Y { … … …  },
has to be absolute.

It doesn't matter whether … … … is any or all of:

.  a sequence of absolute pitches (ie deliberately written that way),

.  \relative [P] { some music in relative pitch notation }

.  \fixed Z { some music in fixed octave pitch notation }

and that is because each of these constructions yields an expression
with absolute pitches.

But we were being pedantic, apparently, so I would point out one thing.
Although you might write   music = { a c' e' a' }   and say
that the expression is absolute music, it *is*, but only if
you avoid placing \music in inappropriate positions in the score.

If you want to exercise an abundance of caution, as we say today,
you should really write   music = \absolute { a c' e' a' }
so that you can place \music anywhere.

> And of course there's
> 
> - MacMillan's strengthening: "Avoid using \relative altogether" :-).
> (I don't agree but understand his rationale; I think it's a matter of
> weighing up the pros and cons.)
> 
> In my experience, there are few use cases for transposition inside
> \relative except for precisely the case situation the OP described:

Even here, there is no reason not to close the \relative
section so that the \transpose command is not inside it.
There is no trailing material (shown by ••• •••
in https://lists.gnu.org/archive/html/lilypond-user/2021-03/msg00118.html)
which might conceivably depend on not closing the section. Thus:

Elaine's final:

\transpose c a, {
  \relative {
    c'4 d e f g a b c
    \transpose as bf \relative {
      cs'4 ds es fs gs as bs cs
    }
  }                         ← remove from here
}

restructured:

\transpose c a, {
  \relative {
    c'4 d e f g a b c
  }                         ← to here
  \transpose as bf \relative {
    cs'4 ds es fs gs as bs cs
  }
}

> Enharmonic respelling of a complete passage, most often needed in
> situations where a whole-score \transpose creates awkward key
> signatures for some sections of the score. This happens quite often
> when engraving Kunstlieder, for example, because with them it's common
> to issue the same piece in various transposed versions. I did an
> edition of Norbert Burgmüller's "Wie der Tag mir schleichet" last
> year,
> https://imslp.org/wiki/5_Lieder%2C_Op.12_(Burgm%C3%BCller%2C_Norbert)
> last year, and I remember having to wrestle quite a lot with
> enharmonic respellings for that one...

Cheers,
David.