Re: [GLISS] turning strings to symbols

[David Kastrup]

"Trevor Daniels" <address@hidden> writes:

> There are a few areas of the documentation that will need manual
> change, partly to explain the new syntax, partly to correct the
> indentation of lines following changed lines, and maybe one or two
> others.

_Very_ much so.  However, a major documentation rewrite for a change
that is going to get canned is not something I want to indulge in, so I
first want to see some getahead on this.

> These can be addressed after the event.
>
> X-offset and friends.  I'd prefer to change these to x/y-offset, to
> unify the letter-casing of properties.  Can also be addressed later.

"addressed later" implies that this is a related issue, but in my book,
it is quite independent.

> I'm less concerned than Werner about the inconsistency of the
> tweak syntax.  The context needs to be specified only rarely, and
> it is a small price to pay for the enormous gain.

Well, strictly speaking we are getting hosed at the latest when \tweak
supports tweaking nested properties.  The problem with \tweak is that
the syntax really leaves no good place for an optional grob spec.

We basically have

\tweak property-path value music

We can't fit an optional argument before "value" as value can take on
_any_ value and thus will fit the optional argument.  We can't put it
_separately_ before property-path since property-path will fit fine
there as well.  Fitting it before music currently means a restricted
form of music, and then we have

\tweak color #red Accidental cis

which is not all too hot.  So I see really only two reasonably
consistent solutions that both involve _not_ using \tweak for the
grobbed variant:

\tweakGrob Accidental color #red cis

or

\single \override Accidental color = #red cis

since the latter is now available.  It is just more effort both for
LilyPond and the user.

> I'm also encouraged by the hints you've dropped that #4 and 4
> can also be made equivalent in the majority of cases.  In a later
> patch, of course.

Have you tried in the last half year?  I should be surprised if you find
many places where they are not already perfectly interchangeable.  The
problem is more making 4 mean a _duration_ when you need one.

-- 
David Kastrup