Hacker News Reader

Comment by sketerpot | Hacker News Reader

Inspired by this, I've started writing a compiler for a subset of Matlab. I'm sick of Matlab being an interpreted language year after year, and while I don't hope to change that, I can at least do a proof-of-concept to lend weight to my derision.

I've got the lexer and parser done, thanks to Parsec, and some basic C code generation as a sanity check. Now for the runtime LLVM code generation, to make it feel like an interpreter! Thanks, HN!

View on HN

I've been fooling around with this idea for a good year now. I've never found the time, but I think about doing something like it all the time. If I come across free time after my thesis, I hope to write a simple typed matlab-like clone that targets llvm code (which supports vectors). To me it seems that there is a vacuum for a simple typed fast numerics language with native vector support. Writing numerics in Matlab is great, but if you ever have to loop, it ends up dog slow. Porting the code to C is usually not terribly difficult, but is always a big pain, but usually yields a factor 10x or more speedup which is pretty significant. It would seem that applied mathematicians and physicists around the world would love a new, modern and fast numerics language to replace coding in C/C++/Fortran.

Probably frontend can be taken from GNU Octave? Claims to be similar to Matlab...

Could you post a link to Parsec. Quick googling didn't come up with any compiler tools. Thanks!

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

It's a Haskell library: http://hackage.haskell.org/package/parsec

Here's a tutorial that uses it to parse a simplified Scheme: http://en.wikibooks.org/wiki/Write_Yourself_a_Scheme_in_48_H...

Reply on news.ycombinator.com