Re: MIDI dynamics parsing error

[David Kastrup]

Thomas Morley <address@hidden> writes:

> Sorry if I was not clear enough.
> I was looking for a method to circumvent the midi-error, but keep the
> visual output.
>
> Let me be a bit more verbose:
>
> The following gives nice visual output, though triggers an error with midi:
>
>   { c'1\> c' c'\! }
>
> Inserting an expicit Dynamic resolves the midi-error, though, the
> Hairpin starts at the right of the DynamicText, ofcourse.
>
>   { c'1\mf\> c' c'\! }
>
> Trying to remove the printed DynamicText via \hide or \omit or
> #point-stencil, leaves the Hairpin untouched, i.e. starting to the
> right of the now invisible or non-existent DynamicText-stencil.
> (Little different with \hide)
>
>   { c'1-\hide\mf\> c' c'\! }
>   { c'1-\omit\mf\> c' c'\! }
>   { c'1-\tweak #'stencil #point-stencil \mf\> c' c'\! }

The question then is what the right look is.  If the right look is the
one with the Midi error, and apparently it is what we get in other
situations, then arguably the code for typesetting the hairpin should
explicitly check for an _omitted_ stencil (one being #f) and in that
case revert to the anchorless behavior.

In short: if \omit does not do the trick now, I think it should (bug
report?).  We don't have a separate override shortcut for point-stencil
in case one still wants an anchor around.  The still-discussed
appearance override would allow using something like

\behold \markup \null \mf

in this function if it is really necessary for some case.  But I think
that in the use case of \omit\mf, the desire really is to only have an
acoustic representation without a visual impact.

-- 
David Kastrup