Re: GOP2-3 - GLISS or not

[Graham Percival]

On Fri, Jul 27, 2012 at 08:54:47AM +0100, Joseph Rushton Wakeling wrote:
> On 26/07/12 19:19, Graham Percival wrote:
> >I should add some more context.  I've just remembered that we have
> >a tutorial (don't ask me how I forgot), and that covers pretty
> >much what I was thinking about "really simple music".
> 
> Thanks for that clarification -- it's actually a narrower subset
> than I thought you might be considering.  I do think it's very wise
> to exclude beaming, articulation and slurs/ties/phrase marks --
> there are a number of potential syntax changes I can think of that
> might be useful here.

Yes, both in terms of user input and simplifying the parser.  I'm
not saying that we'll never stabilize those; just that I think
they're too complicated for the first round.

> >I think we're talking about different things.  Let's put it this
> >way: do you think that we'll ever move away from
> >   c'4
> >being a quarter note for middle C ?  That's the basic question
> >here.  It doesn't matter how lilypond represents c'4 internally
> >(whether it uses grobs or contexts or lilypond rationals or scheme
> >rationals or bits or trits or qubits).
> 
> This is kind of the nub of the issue.  I agree that the notation for
> staff pitches (and octaves) is going to remain stable -- but I'm
> _not_ convinced that you can guarantee stability for accidentals or
> durations.

Think of the stable notation as a subset, not the complete set.

>       c + 1/2 - 1/4 + 1/8 + 1/16 ....   [NOT a suggested notation!!]

Suppose we adopt this notation in the future.  We can still let
users write
  cis   c+1/2   ces  c-1/4-1/4
and produce c sharp c sharp c flat c flat.  Interally, the
"stable" cis would be "translated" into c+1/2.

Granted it would be somewhat annoying to maintain that kind of
"pre-translation", so I don't want to open the floodgates and
blithly assume that we can do that for everything.  But in the
case of note pitches, I think it's simple enough that we can
handle this.  Ditto for durations -- even if it's desirable to
specify note lengths in a more advanced fashion, I think that the
basic "1 2 4 8 16" could be "pre-translated" to that advanced
fashion if necessary.

> So in a way, this may be the
> part of Lilypond that it's least urgent to commit to stabilizing
> (which doesn't mean that it won't turn out to be stable in
> practice).  That's why I suggested doing a careful empirical study
> of the principal sources of syntax problems when trying to convert
> from one Lilypond version to another.

Hmm.  I'll have to think about this more.  My first thought is
that sources of syntax problems is changing the syntax -- but
rather being a "duh" moment, my point is that we've never changed
the note-names and basic rhythms, so there's no data about how
hard that would be.  Users would need to memorize a new set of
note-namess (not hard, but annoying), programs which export
lilypond would need to be updated, etc.

I'll wrap your concern into the next version of this proposal and
send it out later today or tomorrow.  That should give it extra
visibility and somebody may have a good suggestion about it.

- Graham