Music properties and naturalization/transposition style

[Joseph Wakeling]

Hello all,

A few queries related to my ongoing project to implement customizable
transposition/naturalization styles.

It was suggested to me to use music properties to set the rules.  I've
come to the conclusion that this may not be the right approach.  Why?
Because as far as I can see music properties work via an "outmost wins"
hierarchy: if I have,

  \withMusicProperty #'some-property = #1 {
      \withMusicProperty #'some-property = #2 {
          \withMusicProperty #'some-property = #3 {
              \myMusic
          }
      }
  }

... then #'some-property will have a value of 1 for \myMusic.  By
contrast I think that a naturalization style needs to work like the
\relative command -- it should be the _innermost_ one that matters.

Specifically, the naturalization style should have the following
characteristics:

   (i) possible to set a default value at Score, StaffGroup, Staff
       and Voice level, with innermost overriding outermost [sample
       use-case: in general a score will have a default style, but
       individual instruments (e.g. harp) may have special rules]

  (ii) possible to alter within the music, with innermost overriding
       outermost [sample use case: whatever the character of the piece,
       individual passages may need to be designated tonal, chromatic,
       or whatever else suits]

 (iii) ideally, possible to override in a manner that _ignores_ any
       inner modifications [sample use case: an instrument like the
       harp, where you need to be able to guarantee that the music
       will be notated according to certain rules]

  (iv) an option to employ naturalization _only_ for music that has
       been passed through the "transpose" function, or alternatively
       to apply to _all_ music where a rule is set -- so it can be
       alternatively a transposition style _or_ a means of enforcing a
       general notation style.  Again, this property can be set at
       Score, StaffGroup, Staff or Voice level.

Sample pseudo-Lilypond of use, excluding the 3rd case which I'm not sure
how best to notate.

  \new Score \with {
    % this setting turns naturalization "off",
    % i.e. uses Lilypond's defaults.
    naturalizationStyle = ##f

    % this setting determines whether naturalization
    % is applied to all music (#f) or just transposed
    % music (#t).  By default it is #t.
    naturalizeTransposedOnly = ##f
  } {
    <<
      \new Staff = "Clarinet" \with {
        naturalizationStyle = #'chromatic
      } {
        \transpose a c' {
          \someMusic
          \naturalize #'tonal {
            \someTonalMusic
          }
        }
      }

      \new PianoStaff = "Harp" \with {
        naturalization = #'harp
      } {
        \harpMusic
      }

      \new Staff = "Voices" {
        <<
          \new Voice = "Soprano" \with {
            naturalization = #'chromatic
          } {
            \chromaticVocalPart
          }

          \new Voice = "Alto" {
            \defaultVocalPart
          }
        >>
      }
    >>
  }

What I'd like: first, thoughts on the appropriateness and feasibility of
the above notation and "conceptual" model of the naturalize functionality.

Second, advice on implementing it, since I don't think music properties
are the way to go.  In particular, where should naturalization be called
if it's aimed to apply to _all_ notes?  (It's clear where it should be
called if it's to apply only to transposed music.)

Thanks in advance,

    -- Joe