Keith Devens .com

Entry 1179

More cool comments on the topic that was spawned by my question (thanks again Ehud). Specifically:

"More generally, Paul Graham wants to write a variety of Lisp because that's the language he likes most, but he wants to change it in various ways. The last major attempt to revamp Lisp resulted in Scheme, which in turn resulted in languages like Dylan and REBOL, both interesting, useful, powerful, and different languages. And that's the point. Arc will presumably be a different language from Dylan, REBOL, and Scheme, with different strengths and weaknesses."

Great comment. I'm not aware of how much Dylan took from Scheme. Well, while I'm here, here's a neat article on Dylan from MacTech, here's the Dictionary of Programming languages entry for Dylan, and here's Cetus Links on Dylan.

He also writes this: "Anyone asking a question which amounts to "why isn't my favorite offshoot of Scheme taking over the world" should probably learn Scheme, to find out the answer. SICP might be a good place to start."

Well, I do have the book, although I haven't read the whole thing yet. However, my question didn't amount to "why isn't my favorite offshoot of Scheme taking over the world". Although I think it does have some nifty features, I don't even particularly like REBOL. The thing is, some very smart people over at REBOL Tech took Scheme, tore it apart, and put it back together to come up with a language which they think is better. My question was asking Graham where he thinks they failed and where his philosophy differs (where will he do things differently). I was asking about his ideas about language design. I wasn't saying that REBOL is the best language ever, or complaining that it hasn't taken over the world yet.

Finally, the first part of my question was related. I can speak (relatively) intelligently about the merits of Java, C, C++, Pascal, Perl, Python, etc. etc. for solving problems. But I'm not very well versed on the functional area of language design. What sort of problems are Haskell or ML good for that Lisp isn't? Why did people feel the need to try different approaches to functional language design if Graham thinks that Lisp is the best basis to have? Did the other approaches succeed?

Hey, someone else thought that comment was a little weird too.

Permalink | Comments? | id 1179

16:49 (utc-5)
edited: 16:54 (utc-5)