Static Typing and Functional Languages | Hacker News Reader

Static Typing and Functional Languages

etherplex.org

14 pointsapotheon10 comments

That's really an odd post. The Hindley-Milner type inference algorithm is from ~1969. Advanced static typing systems have been around for ages, it's not like we suddenly have the capability of building languages with them.

Mark me down as confused about what the author was trying to say here.

The article is very similar in spirit to my hit 2008 research paper "Forward Motion in Automobiles".

All kidding aside, he probably should have named the article "Static Type-checking in Lisp", and we would have still been just as unimpressed because he forgot to mention Qi; the single most innovative Lisp from the last 20 years.

http://en.wikipedia.org/wiki/Qi_%28programming_language%29

Or any of the other soft/partial/gradual typing work that's been ongoing since the 80s. Check out Jeremy Siek's papers on gradual typing for some of the latest. Also Tobin-Hochstadt and Felleisen's work on Typed Scheme.

On the one hand, this is ridiculous. On the other hand, I'm happy to see people outside of PL thinking about programming languages. There are a number of texts that can help a novice learn more about the topic, with Pierce's Types and Programming Languages being the foremost.

Interesting... I hadn't heard of Siek's work before. A link: http://ecee.colorado.edu/~siek/gradualtyping.html

> So, suddenly it all came together for me.

He seems new to this stuff. Nothing really advanced here, but he brings up a big unresolved issue in language design in a way that probably makes sense to a Java-educated college student.

BTW, Hindley said the concept predates computers; there's been a lot of refinement before and after 1969. (http://www.cis.upenn.edu/~bcpierce/types/archives/1988/msg00...)

Could you supply an example of an "older advanced static typing system"?

The first ML implementation by Luca Cardelli was done at the same time as K&R C was fully implemented. 36 years ago!

http://en.wikipedia.org/wiki/ML_(programming_language)

er - type inference in ocaml is descendent from ML, which had the feature in the 70s - hell, hindley-milner was published in 1969. This has nothing to do with how recently the language has been developed, it is a simple tradeoff between type safety and developer effort. There is really no room for speculation on a subject that a 15 second visit to wikipedia would clear up.

Not sure what the moral of this piece is. But that piece of Java code is both impressive & scary as hell at the same time.

Reply on news.ycombinator.com