News for hackers. Stuff that matters.
Re:Lisp is Sin (2006) by jerf on Thursday August 12, 2010 @10:57AM (#1598230)
Assuming by Lisp we mean either "Common Lisp" or "what somebody would have called Lisp in 2006", there is now an exception to "All roads lead to Lisp" that has moved from total obscurity to all-language-geeks-should-know-about-this: Haskell. You can not take a 2006 Lisp and macro your way to purity. You could build a pure subset, but that would be a new language in every way that matters because you couldn't use non-pure libraries without a lot of wrapping.

You can of course also start a new Lisp-like language with new primitives that do include immutability, but I would suggest that that shows Lisp moving forward by taking a feature from another heritage, which means you've no longer got all roads leading to Lisp/2006, you've got at least one leading away.

If there's anything strange about Lisp, it is that it really was a local optimum posing as the global optimum for so long. We should end up with a few more local optima before all is said and done; there aren't enough yet.

Re:Lisp is Sin (2006) by ohyes on Thursday August 12, 2010 @11:51AM (#1598380)
The whole point of lisp is that it rolls around picking up new ideas. Haskell was originally implemented in lisp.

Examples: Lexical scoping (ALGOL)

The 'format' macro(I believe Fortran)

The 'loop' macro (COBOL? Pascal? not sure?)

Type systems ala Qi and Haskell (ML and typed lambda calculus influenced)

Optional type declaration are probably to approximate C.

You can implement prolog it...

I believe lisp machines even had GC'd versions of C, Pascal, and other languages written for them.

Personally, when I read "all roads lead to lisp," it is taken with tongue firmly in cheek. Lisp isn't really the 'best' language, it just ends up being able to subsume all of the ideas from them. In RPG terms, Lisp is the red mage.

This is why I like it a lot; given a certain problem domain, you can approximate the global optima that is reached in another language.

Re:Lisp is Sin (2006) by zephyrfalcon on Thursday August 12, 2010 @11:54AM (#1598389)
"The whole point of lisp is that it rolls around picking up new ideas."

Kind of like the Katamari Damacy of programming languages... :-)

Re:Lisp is Sin (2006) by frou_dh on Thursday August 12, 2010 @12:46PM (#1598539)
When I played that game I was ruining and picking up junky city debris! That sounds a scary language.
Re:Lisp is Sin (2006) by gaius on Thursday August 12, 2010 @11:31AM (#1598324)
Not only Haskell: F#, OCaml, Erlang and Scala too. While Lispers were basking in the 50-year-old glow of being 30 years ahead once, all of these languages leapfrogged them and they don't even know it.
Re:Lisp is Sin (2006) by cageface on Thursday August 12, 2010 @02:03PM (#1598730)
One killer feature remains exclusive to Lisp: macros. However, in languages as flexible as those you mention, with higher order functions and good type inference, you don't need macros as much as you might in less flexible languages. I'd love to see a serious evaluation of the cost/benefit of s-exprs and macros in real-world production code but I'm not sure how exactly how you'd measure this.

My own cursory examination of the clojure-contrib library's use of macros, for example, suggested to me that many of them wouldn't be necessary in a language like Scala, but I'd love to see a more rigorous study.

Re:Lisp is Sin (2006) by gaius on Thursday August 12, 2010 @02:24PM (#1598776)
One killer feature remains exclusive to Lisp: macros

Lisp can't rest on its laurels here either:

http://www.haskell.org/haskellwiki/Template_Haskell