[parenscript-devel] parenscript on parenscript?

[Daniel Gackle]

< It would also be nice to have the full-featured JS environments (V8,
Rhino, SpiderMonkey) as another target for CL deployment. >

I think this is going to become increasingly important as the JS
implementations improve. The recent release of Clojurescript is an
indicator.

We've actually switched Skysheet to using PS server-side as well as
client-side so we can deploy to V8 on the server. There are three
reasons for this:

1) it eliminates the impedance mismatch between our CL code and our PS
code, which was a drag on development (example: JS's hash objects
don't have a very good counterpart in CL; neither CL hashtables nor
plists/alists are a great substitute);

2) we get better performance from V8 than we do from SBCL. (That's for
unoptimized CL. Optimized CL would do much better of course, but that
would worsen the impedance mismatch described above;

3) interoperability with JS gives some desired conveniences, like V8's
fast JSON support and Node.js for non-blocking i/o.

A full CL-in-JS would open the door to an all-CL system rather than an
all-JS one. Either would be better than straddling the two. Would we
switch? Hard to say.


On Wed, Jul 27, 2011 at 1:56 PM, Vladimir Sedach <vsedach at gmail.com> wrote:

> Currently Parenscript uses some things from Common Lisp that aren't in
> Parenscript. The three approaches to make PS self-hosting are:
>
> 1. Rewrite PS code in a DSL that can be compiled with Parenscript. For
> example, replace car/cdr/etc. with destructuring and templating macros
> that translate to car/cdr for CL and array access in JS. Then on the
> JS side you'd construct your code from arrays instead of conses. The
> only problematic thing is symbol representation.
>
> 2. Implement the missing CL features as a run-time library (add a cons
> prototype with car/cdr methods).
>
> 3. Make a Common Lisp implementation that runs in JavaScript.
>
> I think the last option is the most overall useful, and the most fun
> to do. Parenscript then would (along with an existing CL
> implementation) act as a "syslisp"/bootstrapping dialect and
> cross-compiler. Red Daly already has large parts of the needed CL
> runtime implemented in his PSOS project
> (https://github.com/gonzojive/paren-psos).
>
> How useful would Parenscript be in the browser? My guess is not very.
> After all, it will just be a macro pre-processor over JavaScript's
> eval, and anything you can do with that you're better off doing with
> closures.
>
> OTOH having a full Common Lisp on the browser (*with* a working ASDF)
> would be pretty sweet. It would also be nice to have the full-featured
> JS environments (V8, Rhino, SpiderMonkey) as another target for CL
> deployment.
>
> Vladimir
>
> On Wed, Jul 27, 2011 at 7:24 AM, Anton Vodonosov <avodonosov at yandex.ru>
> wrote:
> > Hello.
> >
> > I am new in parenscript and only glanced through the documentation on the
> project page.
> >
> > As I understand parenscript source is compiled to javascript using a CL
> implemented compiler.
> >
> > It's interesting, how difficult it is to have the same compiler
> implemented on javascript (i.e. on parenscript itself).
> >
> > As parenscript is a subset of CL, in theory it mitght be not difficult.
> >
> > The goal - is to have a browser only Lisp implementation, that can work
> without hunchentoot. To include parenscript
> > scripts onto browser directly, redefine functions interactively and so
> on.
> >
> > What do you think?
> >
> > Best regards,
> > - Anton
> >
> > _______________________________________________
> > parenscript-devel mailing list
> > parenscript-devel at common-lisp.net
> > http://lists.common-lisp.net/cgi-bin/mailman/listinfo/parenscript-devel
> >
>
> _______________________________________________
> parenscript-devel mailing list
> parenscript-devel at common-lisp.net
> http://lists.common-lisp.net/cgi-bin/mailman/listinfo/parenscript-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.common-lisp.net/pipermail/parenscript-devel/attachments/20110727/da740d19/attachment.html>