Re: TimeSignature with note in denominator

[Carl Sorensen]

On 11/13/21, 4:05 PM, "David Kastrup" <dak@gnu.org> wrote:

    Kieren MacMillan <kieren@kierenmacmillan.info> writes:
    
    > Hi David,
    >
    >> It doesn't answer the question.
    >
    > Did my explicit answer in the other email (i.e., “one
    > quintuplet-sixteenth-note”) not suffice?
    
    No.  You propose replacing (cons 3 4) as a time signature designation
    with (cons 3 (ly:make-duration 2)).  You have failed to give any
    indication of what you want to see (cons 8 20) replaced with.

What if the time signature description were (cons 3  '(2 0 1/1)) or '((rep 3) 
(dur_param  (2 0 1/1))?
    
    LilyPond does not have something like a 1/20th duration.  Regarding
    durations, what occurs inside of \tuplet 5/4 and \tuplet 10/8 is
    completely indistinguishable: "tupletism" is not a part of durations.
    
    So I repeat: what duration in LilyPond do you want to use to represent
    the denominator in 8/20 ?  1/20th here is neither a 5-tuplet nor a
    10-tuplet: it represents a fraction of a whole note, not a particular of
    several possible note values.
    
    It will likely end up as (ly:make-duration 4 0 4/5) but that has no
    unique printed representation different from (ly:make-duration 4), and
    (ly:make-duration 4 0 4/5) and (ly:make-duration 4 0 8/10) are
    absolutely indistinguishable.

The alist I proposed above would be able to distinguish between 4/5 and 8/10.  
Even if I don't have the details right yet, I'm certain that they can be made 
right.

Instead of storing the duration, it stores the parameters used to create the 
duration.

I don't know if it's a good idea, but it is an idea.

Carl