I've been chugging away on Wisp, announced
in my last post, every day since I started it a few weeks ago, and
it's becoming a pretty solid system. There's now an exception system,
reference counting for dealing with garbage, and a reentrant
parser. It's no replacement for any other lisps, but I've found it to
be very fun to work on.
git clone git://github.com/skeeto/wisp.git
I wanted to show off some of the new features of Wisp, and since I was
inspired by Full
Disclojure, since it's so damn slick, I decided to make some
screencasts of Wisp in action. All of the screencast software for
GNU/Linux is pretty poor, but after a few hours of head-banging I
managed to hobble something together for you. Enjoy!
That video demonstrated the memoization function. It can be pulled in
memoize library. You give it a symbol, which
should have a function definition stored in it, and it will installed a
wrapper around it. In the video I used the Fibonacci function from
(fib 30) ; Slooooow ...
(fib 100) ; Fast!
This demonstrated the "detachment" feature of Wisp, which is similar
to "futures" in Clojure. It forks off a new process, which executes
the given function. The
send function can be used in the
detached process to send any lisp objects back to the parents, which
can receive them with the
send function can be called any number of
times to continually send data back. The
will block if there is no lisp object to receive yet.
(setq d (detach (lambda () (send (fib)))))
(receive d) ; Gets value from child process
This video shows off the point-free functions that have been defined:
function composition and partial application (I accidentally say
"partial evaluation" in the video). These are actually just simple
macros that any lisp could do.