I'm sure this is a minority opinion, but...
It is hard to find a paper on any interesting topic that does
not introduce a new syntax. The business of thinking about
specific issues seems intimately connected to finding a good
representation for the issue in print. It seems clear from this
that hacking syntax is a *good* thing, and everyone
does it because it is useful in problem solving.
This is the long way around of saying that I don't agree
that rigid syntax is a good thing. I don't believe that one set
syntax for a language is a good thing.
It is *very* true that freeing up syntax opens the door to all
kinds of abuses, but we're all adults. Rigid syntax does
not prevent abuses, it just prevents me from developing a syntax
that fits my problem domain.
On the low level, then, I am all for rebinding +, - and anything else.
The standard operator symbols are valuable for rebinding because
they have so many different meanings in different contexts. I would
be more willing to prevent the rebinding of obscure operator symbols
because they have so few meanings!
On another level, what about a language that could handle multiple
symbol tokens (if-then-else, [-], while-do) as well as all the
different fixities? It is available now in OBJ3, but the parsing
algorithm is backtracking -- expensive and difficult to predict for
certain constructs.
My wish? A language that allowed syntax modifications and warned me
about bad ones.
Then constructs like n+k wouldn't be the headache of implementers,
just the people who use them. Hopefully the basic language would be
simpler and maybe syntax would disappear from discussions like this
because everybody could have the one they liked.
Ken
