Re: grace synchronization

[David Kastrup]

Reinhold Kainhofer <address@hidden> writes:

> On Fr., 26. Aug. 2011 23:32:39 CEST, David Kastrup <address@hidden> wrote:
>
>> I checked around, and Rationals (which are used for time) have a
>> representation of -inf.   I think it would be a good start if the default
>> grace time component (namely when no grace is present) was not set to 0,
>> but to -inf, meaning everything comes before graces.   Except those
>> things that are actually preceded by graces.
>
> Sounds reasonable. 
> But of course, then we run into the opposite problem: What if one
> voice has a time sig between the grace and the note (grace moment 0),
> while another voice has the corresponding time sig before the note
> without any grace (i.e.-inf)...

Where is the problem?  Time sig -inf comes first, then grace, then time
sig 0.  Specifying a time sig at grace 0 is decidedly weird.  If you
want to have the other time sig indeed at grace 0, just put \grace { }
before it.

> The only thing I would change is that in the output of a moment the
> grace component would only be available if >-inf. But then, that's
> also how 0 is currently handled, so this would clearly be changed,
> anyway.

I think that grace time -oo and grace time 0 are conceptually different
enough to warrant a different presentation also in print.

Suppressing -inf output and printing 0 would work.  SCM_UNSPECIFIED
would also be natural for suppressing print and in semantics, it just
needs more treatment in sorting and comparisons.

-- 
David Kastrup