))))))))))))))))))))))))))))))))))))))))))))))))))

[JoeP]

Creator Of Lisp, John McCarthy, Dead At 84 | TechCrunch

Another giant of the computer industry.

(if (= nil 'john-mccarthy) (print "Goodbye world"))

[S.Vashti]

good title, JoeP.

[Ensign Steve]

Rest in peace and all that, but god damn lisp was the most boring and stupid language I ever did learn. Prolog wasn't much better. Maybe I just hate AI.

Still, it's awesome and we wouldn't have had Watson without this guy. So rest in peace, again, sincerely.

[Corona688]

AI people will forgive any language limitations for being able to mathematically prove their program does what they think it should, and LISP is, theoretically, mathematically perfect.

Not an easy achievement, that.

[Demimonde]

[Chatter]

Originally Posted by JoeP

(if (= nil 'john-mccarthy) (print "Goodbye world"))

All the more profound since (= nil 'john-mccarthy) will always be nil.

[specious_reasons]

Originally Posted by Demimonde

"My God, it's full of 'car's!"

Sheer awesome, that.

[Chatter]

Originally Posted by Ensign Steve

Rest in peace and all that, but god damn lisp was the most boring and stupid language I ever did learn.

Did you learn Common Lisp? Which implementation? CMUCL? SBCL? Allegro? Lispworks? Corman? CLISP?

There's a lot of confusion about Lisp (especially concerning the capitalisation). There have only been two main dialects for the last 30 years: Common Lisp (or just Lisp) and Scheme. Common Lisp is supposed to be the industrial language. Scheme is the academic language. The former was standardised in the mid-80s.

I don't know what dialect of Lisp Corona688 learned, but it isn't Common Lisp or Scheme. You can't prove very much about a Lisp program, because there are no guarantees about what a piece of Lisp code will do. The language is a mess. It allows more pernicious side-effects than any other language, allowing extreme behaviour such as being able to reflect into arbitrary namespaces and rebind symbols on the fly, thereby forcing code to be recompiled and uninlined all over the place and dangerously changing a program's behaviour.

This sort of thing means that Lisp is probably the most hackable language out there. And it has been hacked to death. Common Lisp itself was the result of standardising arcane language hacks across numerous dialects. Hackery is pretty necessary in computers that are Lisp all the way down to the machine-level (Lisp machines). But it means that guarantees and proof go out the window.

No, the folklore selling point for the elegance of Lisp lay in how easy it was to write a Lisp interpreter in Lisp. The principle data-structure is the symbolic expression, one designed by John McCarthy to represent parsed code. Parsing itself is trivial. The language barely has syntax. And besides, there is a library function "read" which will do parsing for you. After that, it's just a matter of doing a big switch statement and working out how to compose a continuation (a first-class higher-order function, a kind of object that was for a long time unique to Lisp), and then finally throwing the continuation the identity function.

The ability to do this sort of thing means that it's very easy to write a custom Lisp interpreter and so experiment with new language features. This was called the "Art of the Metacircular Evaluator". It led to such things as Brian Smith's infinite reflective tower which went on to become the basis for reflection in many object-oriented programming languages (including Python).

This isn't a big deal anymore, as there are languages which do the metacircular stuff much better and more sanely, and which really do allow you to prove properties of programs.

[Ensign Steve]

Yeah, I was probably just confused is all.

[Corona688]

Originally Posted by Chatter

I don't know what dialect of Lisp Corona688 learned, but it isn't Common Lisp or Scheme. You can't prove very much about a Lisp program, because there are no guarantees about what a piece of Lisp code will do. The language is a mess. It allows more pernicious side-effects than any other language, allowing extreme behaviour such as being able to reflect into arbitrary namespaces and rebind symbols on the fly, thereby forcing code to be recompiled and uninlined all over the place and dangerously changing a program's behaviour.

Well, goddamn. That was certainly the entire point of lisp as described in college. Now the whole thing just looks like an exercise in machismo.

Come to think of it I don't think they taught us any modern functional languages, just several old/broken/disused ones.

[Chatter]

Just came across this while randomly surfing.

Peter Norvig is currently the director of research at Google. He is also a legendary AI and Lisp hacker. The above preface explains why Lisp was important to AI.

I'm not sure how seriously I can take what he says about macros, though, now that he's such a Python fan.

Originally Posted by Corona688

Well, goddamn. That was certainly the entire point of lisp as described in college. Now the whole thing just looks like an exercise in machismo.

What does, Common Lisp or what you did in college?

Come to think of it I don't think they taught us any modern functional languages, just several old/broken/disused ones.

I think the term "functional language" is just a buzzword, anyway. Python has been called a functional language simply because it supports first-class functions (though not the Lisp kind).

Here at Edinburgh, the first language students use is Haskell. The things I like to emphasise about this language have nothing to do with Lisp, as far as I know.