Re: how to make decisions?

[Han-Wen Nienhuys]

On Tue, Sep 4, 2012 at 6:22 PM, Trevor Daniels <address@hidden> wrote:

>> If I missed your point, can you state it more explicitly?
>
> I can see now my point was not stated clearly.  It was:
>
> At this stage in the discussions it is important to be clear about what
> problems we are trying to solve, "In this discussion we must always
> consider what it is we are trying to optimise."  I take your point that
> the syntax should be unambiguous, but I don't see why this should
> dominate the discussion.  Improving the syntax from the point of
> view of usability is equally if not more important; certainly those
> considerations and discussions should come before thinking how they
> might be handled in the parser.  When you and Jan first started thinking
> about the LilyPond language I'm sure you began by considering how
> music could best be encoded.  Only later did you devise the precise
> syntax.  We should continue in the same way.

You are largely correct about historical choices, but we made our
choices on a clean slate. The complexity of the grammar has progressed
so far that ambiguity and compatibility is now a major issue.

So far, the only proposal I have seen so far that does not generate
any ambiguity concerns is making \[ postfix.

Furthermore, reiterating what David said, if something is ambiguous or
subtle to the parser, it will be to the user as well, and as such it
affects usability. Compare TeX macros. It may be obvious when you see
some examples, but there is edge case behavior which is difficult to
understand

> So what problems do the users have, exactly?  We should address this
> question first.  Janek apparently has his list, which would be a good start.
> But we should not invent problems where they don't exist.  I've probably
> read every email on the user list for the last 4 years, and inconsistent 
> parser
> rules have not figured prominently.  Another example is the considerable
> discussion so far about pre- and post-fix notation.  Again, has this been a
> problem prominent on the user list?  I don't think so.  So why try to solve 
> it?
> Especially in ways that would screw all existing code.  In fact, I don't think
> /users/ have any serious problems with the syntax as it currently exists,
> other than getting to grips with it initially.
>
> So I'd be happy to let David continue his work straightening out the
> parser, which is good, and improving the functionally, even better.
> But if we are to have a discussion about syntax let's first list the problems
> we need to solve, and reach agreement on which ones need to be tackled.
> Then we know what it is we are trying to optimise.

I completely support this analysis.

-- 
Han-Wen Nienhuys - address@hidden - http://www.xs4all.nl/~hanwen