Paul Graham wanted to release the design of Arc, his new lisp, at the Lisp conference this week. He got a little hung up on continuations and floating point numbers, so he presented his current draft language specification instead.
Paul started by identifying the irreducible primitives that are necessary to build a lisp sufficiently powerful to build and eval. His list is currently at 9, but is likely to expand to include a few more primitives. (McCarthy's lisp started with seven irreducible primitives: car, cdr, cons, if/cond, equal, lambda and set).
Continuations are going into Arc, as are macros. Paul is currently taking a very minimalist approach to naming his keywords. Lambda is currently spelled fn, function definitions use the def form, and macros are created with the mac form. Scheme's call-with-current-continuation or call/cc will likely enter Arc as ccc.
Yes, that means that the predicate to determine whether something is a continuation would be called cccp, if you are of a Lisp-y instead of a Scheme-y bent.
On a more serious note, is there more publically available detail on arc yet? archub.org is kind of slim on the ground.
-Dom
Re:Sov Jokes
ziggy on 2003-10-16T16:54:37
No, he's still working on the Arc language spec, written in Arc.John McCarthy tried to impress upon Paul that this was not a good approach, since there will be required elements in the language design that are not representable within that language. But that's where Paul is now. He has ideas on where he wants to go after the core language kernel is finished, but he's not talking about any of that.
Is this the conference in question? (I don't presently move in LISP circles, so I was just wondering.
Re:conference?
ziggy on 2003-10-20T21:47:06
Yes, that was the one.
