Re: Distance of a grob from its reference point

[Aaron Hill]

On 2020-01-14 7:16 am, Kieren MacMillan wrote:
> My mind was a little blown by the result of changing that to
>
>     \override $grob-path .before-line-breaking = #proc
>
> Doing so made it *deeply* and *immediately* clear that, despite being
> a serious power-user of Lilypond for almost twenty years, I know
> almost nothing of how it actually works.

You and me both.  I am still trying to understand what is happening, but 
just querying for relative coordinates is enough to cause problems with 
the bracket's extents as well as the Beam of all things.  I suspect that 
since ly:grob-relative-coordinate caches information behind the scenes, 
that calling it during before-line-breaking is possibly not 
well-supported.

Here is a hopefully minimal example for filing an issue:

%%%%
\version "2.19.83"

{
  % Break things by calling ly:grob-relative-coordinate.
  \override Staff.OttavaBracket.before-line-breaking
    = #(lambda (grob)
         (ly:grob-relative-coordinate grob
           (ly:grob-parent grob Y) Y))

  % Highlighting the beam only for clarity.
  \override Beam.color = #red

  \ottava #1 g''8 8 4
}
%%%%

> Hence the ability to offset (e.g.) an OttavaBracket precisely X staff
> spaces while still [re-]engaging the spacing engine is neither an
> unreasonable request, nor something that Lilypond could (or even
> should!) try to manage automagically.

I suppose my sticking point is the need to be so precise with 
positioning.  I am not connected to the world of modern notation, but I 
cannot envision any musical meaning for the exact vertical position of 
an OttavaBracket.  Rather, an engraver should simply desire it 
positioned so the performer is able to notice it without distraction.  
If it sits too close to other elements, then additional padding may be 
necessary.  And vice versa, perhaps the bracket needs to sit very 
snuggly in some cases to look right.

Ultimately, if an appropriate tool exists but is broken in some way, it 
is best to fix that tool than invent a new one.  Moving an offending 
bracket as mentioned above *should* be doable with Y-offset or any one 
of the various padding properties.  That said, I have noticed that 
slur-padding seems to have no effect on OttavaBrackets.  That is either 
yet another bug or just a figment of my misunderstanding.


-- Aaron Hill