Re: Accidental-restructuring (major)

[Rune Zedeler]

Han-Wen Nienhuys skrev:

> I have another request though. If you are going to rewrite the thing,
> can you think of a representation which is less fragile than the
> arcane nesting of lists?

I am not sure what you mean with 'nested lists'.
The current implementation uses lists of symbols and pairs:

'(Voice (same-octave . 0) (any-octave . 0) (same-octave . 1) Staff 
(same-octave . 0) (any-octave . 0) (same-octave . 1))

My suggestion for the new system is to replace the pairs with functions:

`(Voice ,(lambda ...) ,(lambda ...) ,(lambda ...) Staff ,(lambda ...) 
,(lambda ...)  ,(lambda ...))

I think that the one-level lists are okay, because the algorithm 
literally traverses the list from left to right, doing all the stuff in 
the list in sequence.

> Perhaps you can capture some part of the
> rules in either smobs or Scheme records.

I have never heard of scheme records, and R5RS does not mention them.
What do you mean?

-Rune