Re: Continuations

["Steven Shaw" <address@hidden>]

> But now I have another question about continuations: how useful they are
for
> *masses* of developers?

Generators (like in Sather) are something I've always wanted in a
language-I-get-to-use-at-work. I've been  reading how they can be defined in
Scheme using first-class continuations.

People have already mentioned web applications. I remember when doing web
applications with
Java servlets/jsp, I was frustrated because the user could (in IE) do
File->New Window and get
a duplicate of the current page. IIRC, this window shared the same
session-id. With Netscape/Mozilla,
doing a similar operation did not share the session. Learning about
continuations gives me a different
way to think about web interactions. I haven't "got it" yet, though.

Workflow seems to be getting alot of attention. It can be attacked in a
similar way to web applications
are with continuations. Surely inside all workflow applicaitons lives an
implementation of continuations!
Maybe of some "lesser" sort like recording position in a finite state
machine or something (but this probably
makes it easier to save the current position to the database). Transparent
persistence might be a savier here (but very long lived
workflows/continuation may need to survive/evolve through multiple
system/schema upgrades). I think some of the problems with persisting
continuations (making them small) have been solved as I was refered to some
papers awhile ago from this list that talked about saving a continuation and
placing it inside dynamically generated html.

> Does typical programmer need to know about and use continuations?

Maybe not if they can take a leap of faith.
I hate doing that (but I still don't really understand continuations despite
efforts).

> What kind of advantages continuations bring into development process?

 No idea.

> Even reading archive of this list I can see that there is some level of
> confusion about continuations.
> What can we expect from "simple" developer if we give him continuations as
> part of "mainstream" language (such as Java, C#, etc)?

Port some Scheme-based prolog/rule engine? Port seaside web framework? At
least wait for ports...

If continuations were part of standard platforms like Java and .NET, I think
people would search out Scheme for the answers and SICP would be read by
many. Is there hope of this happening? Is there any way to implement
continuations more efficiently than putting activations frames on a gc-heap?

Steve.