Re: ! Please answer interesting functionality question for PhD Diss

[Jean Abou Samra]

Hi Harm,

Le 25/10/2020 à 11:40, Thomas Morley a écrit :
> Hi Jean,
>
> Am Sa., 24. Okt. 2020 um 11:55 Uhr schrieb Jean Abou Samra <jean@abou-samra.fr>:
>
> > The problem so far is that I didn't find a proper way to set the chord 
> > configuration at a given musical moment (note heads placed at the left or at 
> > the right of the stem). For now I've used an extra-offset, which leads to 
> > inconsistent spacing. There is \override Stem.note-collision-threshold = 50, 
> > but that doesn't let you achieve the fourth chord before the end, where the 
> > upper note is on the left and the lower note is on the right, since the usual 
> > placement is the other way around.
> Stem.note-collision-threshold gives you some possibilities, but to
> freely position note-heads at a Stem, one needs to tackle
> Stem.positioning-done.
>
> #(define (distribute-note-heads-around-stem note-head-shifts)
>    (lambda (grob)
>    "Takes the note-heads from a stem-grob and applies offsets in X-direction
> taken from @var{note-head-shifts}."
>      (let* ((nhds-array (ly:grob-object grob 'note-heads))
>             (nhds-list
>               (if (ly:grob-array? nhds-array)
>                   (ly:grob-array->list nhds-array)
>                   '()))
>             (sorted-nhds
>               (sort
>                 nhds-list
>                 (lambda (p1 p2)
>                   (ly:pitch<?
>                     (ly:prob-property (ly:grob-property p1 'cause) 'pitch)
>                     (ly:prob-property (ly:grob-property p2 'cause) 'pitch)))))
>             (stem-starting-nhd?
>               (lambda (nhd)
>                 (if (positive? (ly:grob-property grob 'direction))
>                     (equal? nhd (car sorted-nhds))
>                     (equal? nhd (last sorted-nhds))))))
>
>        ;; Move note-heads in X-direction, looking at 'note-head-shifts'
>        ;; If a stem starts at a moved note-head, adjust this note-head's
>        ;; 'stem-attachment
>        (for-each
>          (lambda (nhd nhd-shift)
>            (if (and (stem-starting-nhd? nhd) (not (zero? nhd-shift)))
>                (let ((default-stem-attach
>                        (ly:note-head::calc-stem-attachment grob)))
>                  (ly:grob-set-property! nhd 'stem-attachment
>                    (cons (car default-stem-attach)
>                          (- (cdr default-stem-attach))))))
>            (ly:grob-translate-axis! nhd
>              (* nhd-shift
>                 (-
>                    (interval-length (ly:grob-extent nhd nhd X))
>                    (interval-length (ly:grob-extent grob grob X))))
>            X))
>          nhds-list
>          note-head-shifts)
>        #t)))
>
> distributeNoteHeads =
> #(define-music-function (nhd-shifts)(number-list?)
> "Returns an override to distribute note-heads around a stem, according to
> @var{nhd-shifts}."
>    (if (not (every ly:dir? nhd-shifts))
>        (let ((signs (map (@@ (lily) sign) nhd-shifts)))
>          (ly:warning
>            "Only directions like -1, 0, 1 are accepted, transforming ~a to ~a."
>            nhd-shifts
>            signs)
>          ;; sign should really be public
>          (set! nhd-shifts (map (@@ (lily) sign) nhd-shifts))))
>    #{
>      \override Stem.positioning-done =
>        #(distribute-note-heads-around-stem nhd-shifts)
>    #})
>
> {
>    %% Test the warning
>    \once \distributeNoteHeads #'(1.1 0)
>    < ees' gis' >
>
>    \voiceTwo
>    \once \distributeNoteHeads #'(0 -1)
>    < e' g' >
> }


Wohoo! Even when dealing with complex and important merge requests
you are still available for impressive things like this!

Thanks! Now I shall try to understand…

This sort-of looks like a missing feature in LilyPond to me. I'll
take a look at the original callback to see if it's easy to make it
read some property which has a callback similar to what is calculated
at present in positioning-done, so you could override it.

> > Code attached, with an output that resembles the image you (Michael) sent 
> > earlier. Note that you need a development version of LilyPond 
> > (lilypond.org/development). Naturally, this is only a start since at this point 
> > more details are needed about the specification of what you want to achieve. 
> > Nevertheless, I wanted to share this code to avoid duplication of effort in 
> > case someone else were willing to take over this topic.
> To your comment:
> % Not defined?
> #(define pi (acos -1))
>
> It's defined in lily-library,scm as
> (define-public PI (* 4 (atan 1)))

Thank you for the tip!

Cheers,
Jean