Call-with-current-continuation

Abdulaziz Ghuloum writes:
> Jussi Piitulainen wrote:
>> I think Abdulaziz meant them to be the same, but when I tried to
>> use them for something, I noticed the difference.
>
> Obviously, I completely messed this up. Sorry. Mostly, I was

Ok, thanks. No problem.

> But now I'm confused for why this needs call/cc at all. Why would
> one not just do:
>
> (define (generate-one-element-at-a-time ls)
> (lambda ()
> (if (null? ls)
> 'last-element
> (let ([a (car ls)])
> (set! ls (cdr ls))
> a))))

So it falls flat. But perhaps the for-each call inside the original
generate-one-element-at-a-time was struggling to get out, in the way
that becomes so much more apparent in Michele Simionato's interesting
post. It did what I wanted, at least.

On Oct 21, 4:32*am, leppie <xacc....@gmail.com> wrote:
> PS: Continuations confuse the crap out of me. For a beginner I would
> not worry about it too much, there are much nicer things to learn

Leppie, you implemented a R6RS interpreter and still you say this, is
there hope for the rest of us?!

Grant Rettke <grettke@gmail.com> writes:

>On Oct 21, 4:32=A0am, leppie <xacc....@gmail.com> wrote:
>> PS: Continuations confuse the crap out of me. For a beginner I would
>> not worry about it too much, there are much nicer things to learn

>Leppie, you implemented a R6RS interpreter and still you say this, is
>there hope for the rest of us?!

Implementing continuations is probably much harder than using them. I
know that I am able to use continuations fairly easily if I take a little
time to think about the problem.

Aaron Hsu

--
+++++++++++++++ ((lambda (x) (x x)) (lambda (x) (x x))) +++++++++++++++
Email: <arcfide@sacrideo.us> | WWW: <http://www.sacrideo.us>
Scheme Programming is subtle; subtlety can be hard.
++++++++++++++++++++++++++++++++++++++++++++++++++ +++++++++++++++++++++

On Oct 21, 8:53*am, the.brown.dragon.b...@gmail.com wrote:
> Hello,
>
> I have just started learning scheme using and got pretty confused when
> I reached call-with-current-continuation.
>
> After reading up on Wikipedia (http://en.wikipedia.org/wiki/Call-with-
> current-continuation) I think I've got some idea of how what call/cc
> is. But I'm still a bit confused by the example given:
>
> * * *1 *;;[LISTOF X] -> ( -> X u 'you-fell-off-the-end)
> * * *2 *(define (generate-one-element-at-a-time lst)
> * * *3 * *(define (generator)
> * * *4 * * *(call/cc control-state))
> * * *5
> * * *6 * *(define (control-state return)
> * * *7 * * *(for-each
> * * *8 * * * (lambda (element)
> * * *9 * * * * (set! return
> * * 10 * * * * * * * (call/cc
> * * 11 * * * * * * * *(lambda (resume-here)
> * * 12 * * * * * * * * *(set! control-state resume-here)
> * * 13 * * * * * * * * *(return element)))))
> * * 14 * * * lst)
> * * 15 * * *(return 'you-fell-off-the-end))
> * * 16
> * * 17 * *generator)
>
> Basically, what's confusing me is line 9 - why is "return" getting
> set! at this point? And how can it be? Won't the continuation jump out
> of the current context so it no longer continues?
>
> Help?
>
> BD
18 (define x (generate-one-element-at-a-time '(1 2 3)))
19 (x)

The set! first evaluates the call/cc call at line 10, and stores the
value only after call/cc has returned. But call/cc does not return
yet.
In line 13 it returns an element to the continuation of the call to x
in line 19. So line 19 produces element 1.
However, in line 12 variable control-state is updated such as to
contain a continuation that finishes the assignment of line 9.

20 (x)

So now the assignment of line 9 is completed. The value is that of
(return element) which is the continuation of line 20 as passed to
control-state by procedure generator.
After the assignment of line 9 is completed, for-each does the next
cycle while return points to the continuation of line 20.

You may want to have a look at: http://schemecookbook.org/Cookbook/CoRoutines

Jos

Grant Rettke skrev:
> On Oct 21, 4:32 am, leppie <xacc....@gmail.com> wrote:
>> PS: Continuations confuse the crap out of me. For a beginner I would
>> not worry about it too much, there are much nicer things to learn
>
> Leppie, you implemented a R6RS interpreter and still you say this, is
> there hope for the rest of us?!

I'd say yes - they are easier to implement than understand.

--
Jens Axel Søgaard

On Oct 26, 7:13*pm, Jens Axel Soegaard
<findrealaddresswithgoo...@soegaard.net> wrote:
> Grant Rettke skrev:
>
> > On Oct 21, 4:32 am, leppie <xacc....@gmail.com> wrote:
> >> PS: Continuations confuse the crap out of me. For a beginner I would
> >> not worry about it too much, there are much nicer things to learn
>
> > Leppie, you implemented a R6RS interpreter and still you say this, is
> > there hope for the rest of us?!
>
> I'd say yes - they are easier to implement than understand.
>

I am just now implementing them (full re-invokable continuations), but
both implementing and understanding is hard, but I have had some good
help from #scheme on freenode.

Hopefully within the next few weeks I will have something working

Stay tuned!

On Oct 26, 7:13*pm, Jens Axel Soegaard
<findrealaddresswithgoo...@soegaard.net> wrote:
> Grant Rettke skrev:
>
> > On Oct 21, 4:32 am, leppie <xacc....@gmail.com> wrote:
> >> PS: Continuations confuse the crap out of me. For a beginner I would
> >> not worry about it too much, there are much nicer things to learn
>
> > Leppie, you implemented a R6RS interpreter and still you say this, is
> > there hope for the rest of us?!
>
> I'd say yes - they are easier to implement than understand.
>

I am busy implementing them (full re-invokable continuations) for
IronScheme at the moment.

Hopefully I will have something working in a few weeks

Stay tuned

leppie

On Oct 24, 8:45*pm, Grant Rettke <gret...@gmail.com> wrote:
> On Oct 21, 4:32*am, leppie <xacc....@gmail.com> wrote:
>
> > PS: Continuations confuse the crap out of me. For a beginner I would
> > not worry about it too much, there are much nicer things to learn
>
> Leppie, you implemented a R6RS interpreter and still you say this, is
> there hope for the rest of us?!

Full continuations are coming soon to IronScheme

leppie

PS: this is a duplicate message as my replies do not seem to come
through...

On Oct 28, 10:33*am, leppie <xacc....@gmail.com> wrote:
> I am busy implementing them (full re-invokable continuations) for
> IronScheme at the moment.

Awesome!

On Oct 26, 3:23*pm, jos koot <jos.k...@telefonica.net> wrote:
> On Oct 21, 8:53*am, the.brown.dragon.b...@gmail.com wrote:
>
> > Hello,
>
> > I have just started learning scheme using and got pretty confused when
> > I reached call-with-current-continuation.
>
> > After reading up on Wikipedia (http://en.wikipedia.org/wiki/Call-with-
> > current-continuation) I think I've got some idea of how what call/cc
> > is. But I'm still a bit confused by the example given:
>
> > * * *1 *;;[LISTOF X] -> ( -> X u 'you-fell-off-the-end)
> > * * *2 *(define (generate-one-element-at-a-time lst)
> > * * *3 * *(define (generator)
> > * * *4 * * *(call/cc control-state))
> > * * *5
> > * * *6 * *(define (control-state return)
> > * * *7 * * *(for-each
> > * * *8 * * * (lambda (element)
> > * * *9 * * * * (set! return
> > * * 10 * * * * * * * (call/cc
> > * * 11 * * * * * * * *(lambda (resume-here)
> > * * 12 * * * * * * * * *(set! control-state resume-here)
> > * * 13 * * * * * * * * *(return element)))))
> > * * 14 * * * lst)
> > * * 15 * * *(return 'you-fell-off-the-end))
> > * * 16
> > * * 17 * *generator)
>
> > Basically, what's confusing me is line 9 - why is "return" getting
> > set! at this point? And how can it be? Won't the continuation jump out
> > of the current context so it no longer continues?
>
> > Help?
>
> > BD
>
> * * * 18 (define x (generate-one-element-at-a-time '(1 2 3)))
> * * * 19 (x)
>
> The set! first evaluates the call/cc call at line 10, and stores the
> value only after call/cc has returned. But call/cc does not return
> yet.
> In line 13 it returns an element to the continuation of the call to x
> in line 19. So line 19 produces element 1.
> However, in line 12 variable control-state is updated such as to
> contain a continuation that finishes the assignment of line 9.
>
> * * *20 (x)
>
> So now the assignment of line 9 is completed. The value is that of
> (return element) which is the continuation of line 20 as passed to
> control-state by procedure generator.
> After the assignment of line 9 is completed, for-each does the next
> cycle while return points to the continuation of line 20.
>
> You may want to have a look at:http://schemecookbook.org/Cookbook/CoRoutines
>
> Jos

Thanks Jos. I re-read your message 3 times and I think I get it.

So if I understand correctly - a continuation (cont value) returns the
next "continuation context". In the example, this is saved in the
"return" variable and used in the next iteration.

cheers,
BD

On Oct 29, 7:26*am, the.brown.dragon.b...@gmail.com wrote:
> On Oct 26, 3:23*pm, jos koot <jos.k...@telefonica.net> wrote:
>
>
>
> > On Oct 21, 8:53*am, the.brown.dragon.b...@gmail.com wrote:
>
> > > Hello,
>
> > > I have just started learning scheme using and got pretty confused when
> > > I reached call-with-current-continuation.
>
> > > After reading up on Wikipedia (http://en.wikipedia.org/wiki/Call-with-
> > > current-continuation) I think I've got some idea of how what call/cc
> > > is. But I'm still a bit confused by the example given:
>
> > > * * *1 *;;[LISTOF X] -> ( -> X u 'you-fell-off-the-end)
> > > * * *2 *(define (generate-one-element-at-a-time lst)
> > > * * *3 * *(define (generator)
> > > * * *4 * * *(call/cc control-state))
> > > * * *5
> > > * * *6 * *(define (control-state return)
> > > * * *7 * * *(for-each
> > > * * *8 * * * (lambda (element)
> > > * * *9 * * * * (set! return
> > > * * 10 * * * * * * * (call/cc
> > > * * 11 * * * * * * * *(lambda (resume-here)
> > > * * 12 * * * * * * * * *(set! control-state resume-here)
> > > * * 13 * * * * * * * * *(return element)))))
> > > * * 14 * * * lst)
> > > * * 15 * * *(return 'you-fell-off-the-end))
> > > * * 16
> > > * * 17 * *generator)
>
> > > Basically, what's confusing me is line 9 - why is "return" getting
> > > set! at this point? And how can it be? Won't the continuation jump out
> > > of the current context so it no longer continues?
>
> > > Help?
>
> > > BD
>
> > * * * 18 (define x (generate-one-element-at-a-time '(1 2 3)))
> > * * * 19 (x)
>
> > The set! first evaluates the call/cc call at line 10, and stores the
> > value only after call/cc has returned. But call/cc does not return
> > yet.
> > In line 13 it returns an element to the continuation of the call to x
> > in line 19. So line 19 produces element 1.
> > However, in line 12 variable control-state is updated such as to
> > contain a continuation that finishes the assignment of line 9.
>
> > * * *20 (x)
>
> > So now the assignment of line 9 is completed. The value is that of
> > (return element) which is the continuation of line 20 as passed to
> > control-state by procedure generator.
> > After the assignment of line 9 is completed, for-each does the next
> > cycle while return points to the continuation of line 20.
>
> > You may want to have a look at:http://schemecookbook.org/Cookbook/CoRoutines
>
> > Jos
>
> Thanks Jos. I re-read your message 3 times and I think I get it.
>
> So if I understand correctly - a continuation (cont value) returns the
> next "continuation context". In the example, this is saved in the
> "return" variable and used in the next iteration.
>
> cheers,
> BD

Correct, Jos