HN Explorer

6 comments

[ 3.1 ms ] story [ 25.5 ms ] thread
mruts
While sounding cool, this kind of project seems to exist solely to generate a paper, not because anyone actually has this problem.
vkazanov
I wouldn't be so sure. This thing given a syntactically valid piece of code infers all declarations required by it.

They implemented a working demo, along with some helper code so they are trying to push it.

How about IDEs?

zeotroph
And may be a a road to gradual typing, i.e. a forward looking auto. The language spec will probably never allow if (maybe someday in C++), but IDEs might fill in the type for you if it can be deduced from a call site later on.
vkazanov
Yes, that.

I find it really... Wonderful how types and declarations are inferred from actual code using 'em.

OTOH, this is something I see a lot in Hindley-Milner typed languages, although they don't usually go too far here, i.e. it's not like OCaml or Haskell generates a type declaration for me.

roywiggins
Well, they'll soon find out one way or another, as they seem to be planning to sell licenses to use it as a product:

http://cnippet.cc/

rwmj
I couldn't get the page to work, but I'm intrigued by what happens when you give it random data and it tries to infer a working C program from it.