I'm planning to write up a blog post about using list arrows for XML
processing.
Ok, I'll say tuned! Maybe a smaller example for the Haddock docs needs
less time.
Maybe this sounds weird on the Haskell mailing list, but at Silk[1]
we have a full implementation of (functional reactive) list arrows
in JavaScript. The arrows build up an AST of operations that we
statically optimize to a more efficient form. This is needed because
JavaScript itself is not going to fuse intermediate list for you.
I'm reinventing all of this in Haskell now to gain more insight in
what we've actually done in JavaScript. :-)
Sounds more interesting than weird to me. In case you blog about that
too, I'll probably read it. I'm interested to get an intuition what
you can do with arrows and what not. An intuition that goes beyond the
quite abstract "you can't choose different computation paths based on
the result of a computation". Can you, for example, define a `perm`
arrow that yields every permutation of it's input? Monadic
implementations look like they need `>>=` but there may be another
way.. Maybe you now have an example for the docs ;o)
I'm also planning to investigate whether it is possible (useful) to
perform the list computations (the map in concatMap) in parallel.
I'm not sure if someone has already tried this for list monads.
I tried something similar a while ago (for monads):
<http://www-ps.informatik.uni-kiel.de/~sebf/haskell/speedup-search-with-parallelism.lhs.html
>
Sebastian
P.S.
In
[1] https://blog.silkapp.com/2009/12/reinventing-xslt-in-pure-javascript/
there are two occurrences of the `sum` function which should probably
be `alt` instead.
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe
