Interesting idea. It seems like building this on top of REPA would save a
lot of work, since it has a native notion of "rank" encoded in the type
system. I'd then see the APL-like combinators as a "niche" API for REPA,
rather than as a library of their own. And of course, you'd get
parallelization for free, more or less. I think some of the combinators on
that wiki page already have counterparts in the REPA API.
On Thu, Mar 8, 2012 at 8:44 AM, Simon Peyton-Jones
Friends****
** **
Many of you will know the array language APLhttp://en.wikipedia.org/wiki/APL_%28programming_language%29. It focuses on arrays and in particular has a rich, carefully-thought-out array algebra. ****
** **
An obvious idea is: *what would a Haskell library that embodies APL’s array algebra look like*? In conversation with John Scholes and some of his colleagues in the APL community a group of us developed some ideas for a possible API, which you can find on the Haskell wiki here: http://www.haskell.org/haskellwiki/APL****
** **
However, we have all gone our separate ways, and I think it’s entirely possible that that the idea will go no further. So this message is to ask: ****
**· **Is anyone interested in an APL-style array library in Haskell?****
**· **If so, would you like to lead the process of developing the API?****
** **
I think there are quite a few people who would be willing to contribute, including some core gurus from the APL community: John Scholes, Arthur Whitney, and Roger Hui. ****
** **
Simon****
_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe