Can't load iterator.slate

Brian Rice water at tunes.org
Fri Nov 5 23:47:27 PST 2004


On Nov 5, 2004, at 12:42 PM, Todd Fleming wrote:

> Brian Rice wrote:
>
>> Yep
>
>
> So I guess it's time for my free feature request. Mua-hahahaha
>
> How about this: it would be really, really helpful if Slate could 
> solve NP-complete problems on deterministic machines in polynomial 
> time, in such a way that no other language could repeat 
> (Turing-completeness theory not withstanding :) ).

Oof! Uh, can I take a pass on that one? Good, thanks. ;-)

> On a more serious note, I have a question about where Slate is going. 
> Does/will Slate support first class continuations? This isn't a 
> feature request because these are so controversial. They are both a 
> means to produce amazing features (coroutines, threading, prolog-style 
> backtracking, easy stateful http services, ...) and a method of 
> blowing goto-sized confusion into code.

I've had a few conversations with Avi Bryant in person about this (I'm 
sure if you mention http services and continuations you either have 
heard of him or you should), and basically the concurrency setup I've 
worked on and have let languish a bit due to a day-job and lack of 
focus should allow it pretty easily. It won't be continuation-based, 
but more like E (http://erights.org/elib/concurrency/) where you have 
processes with very little shared memory and coordination based on 
asynchronous message-passing and promises. You can roll continuations 
on top of this fairly easily, but the E folks are rabidly concerned 
with language security and distribution (someone has to be), so they 
haven't explored this.

I'm also interested in "user interface continuations" which has come up 
in the context of MIT's Haystack information console UI project. You'll 
also note (well, assuming that you looked) that Slate stack frames have 
continuation parameters in them, which is a bit misleading, but 
indicative.

So yes, I am interested in this and will try to focus my efforts on 
that since Slate desperately needs easily-used concurrent operation to 
scale up.

--
Brian T. Rice
LOGOS Research and Development
http://tunes.org/~water/




More information about the Slate mailing list