syntax change discussions

[Graham Percival]

A week ago, I asked "Does anybody have any other syntax changes
they want to propose?".  This was horribly unclear, and I
apologize for completely giving the wrong impression.

I /meant/ to ask if there were any long-planned (small) syntax
changes that we all agreed on, but which I had previously vetoed
as not going into 2.13.  I didn't mean to invite lots of proposals
of vague possibilities -- even though I this was obviously what I
/wrote/.  I'm completely in the wrong here.


To restate:
- I have a vague recollection that there was one patch, similar to
  the \cresc one (i.e. well-liked and small), which I'd told the
  author(s) to wait until 3.0.  If that describes your / your
  work, please speak up.
- We're not starting a free discussion about syntax yet.
- Even if people *do* start/continue a free discussion about
  syntax, we won't be acting on any of it yet.

As Han-Wen recently noted on the bug list, syntax changes look
simple, get people very enthusiastic, but use a lot of developer
time and are fraught with peril (it's easy to get something wrong,
but if it's wrong, we're stuck with it (unless we make users very
unhappy)).


I know that there's a lot of interest in fixing certain problems
in our input syntax; I have a bunch of things as well.  However, I
also know that I don't know everything, and because of stuff I
don't know, some of my suggestions are going to be complete
bolloks and would cause huge problems if we implemented them.  The
problem is that I don't know what I don't know, so there's no way
of predicting (by myself) which of my proposals are worth doing.

We have a project set aside for this: GLISS.  It will happen when
lilypond is relatively stable (i.e. after 2.14).  It will happen
when we can afford to waste (or at least, "spend") a lot of
developer time discussing stuff.  It will be a structured debate,
wherein people with less time to discuss syntax will still have
their say.  It will be slow; people will have lots of opportunity
to think about matters, maybe do some tests, raise an objection
after two weeks, think some more, raise another objection after
two more weeks, etc.  The exact format of the discussion will be
decided later.


If you're enthusiastic about starting GLISS, the best thing you
can do is to help resolve the Critical issues.  The easiest, most
bite-sized, problems at the moment are the info that's only in the
regtests; most items can be solved in 5 minutes or less.


Again, I apologize for mispeaking / miswriting so badly.  The only
possible interpretation of my text at the end of:
http://lists.gnu.org/archive/html/lilypond-devel/2010-04/msg00308.html
is that I'm inviting free discussion, even though that wasn't my
intention.

Sorry,
- Graham