cocoon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Judson Lester <jud...@irev2.com>
Subject Re: "Continuations" vs. "Stateful Execuation" (was "Preserving Rhino State")
Date Thu, 07 Mar 2002 19:33:38 GMT
On Wednesday 06 March 2002 05:35 pm, Ovidiu Predescu wrote:
> On Wed, 6 Mar 2002 18:05:42 -0500, Jason Foster <jafoster@uwaterloo.ca> 
wrote:
> > > I looked on the Rhino code, and also spoke with Christopher Oliver
> > > (Rhino serialization) and Norris Boyd (Rhino main developer) about
> > > continuations support in Rhino.
> >
> > You *have* been busy :)
> >
> > <snip content="full continuations are out"/>
>
> Yep ;-)
>

Would this really be so bad?  Certainly the set of utility that continuations 
supplies minus the set of utility from CPS is not null, (well...I'll concede 
the point as OT) but isn't the set of utility we need for Cocoon from 
continuations intersected with the set that CPS provides is identical to the 
first set, no?  I mean, don't we get everything we want from CPS?

<snip>
> > Near as I can tell, the really big thing, other than transparent states,
> > will be the abilities to call a Cocoon pipeline for output and to
> > construct pipelines on the fly.  <-- deliberate Functionality Syndrome
> > provocation, but let's be honest... it *will* happen regardless of
> > whether we use Rhino or FlowScript.
>
> As I said in an earlier message, we should not have a language
> designed for the average programmer, but for the hackers we all
> are. It would take a lot of words to explain why we should do this,
> and other people have done much better explaining this. Therefore I
> ask you to go read some interesting stuff:
>

I think both your references could as easily be read for either side of this 
argument.

>   http://www.dreamsongs.com/WorseIsBetter.html (long version, highly
>   entertaining and funny)

In which it's argued that Lisp is a bad solution because its better.  Or 
worse.  Or not.  He doesn't know.  In essence though, the Worse is Better 
suggestion is that less structured systems are prefered...which tends to 
attack the purity suggestion later.
>
> Five questions about language design:
>   http://www.paulgraham.com/langdes.html

I thought the suggestion that CPS in servers would be great, real 
continuations better "if it isn't too costly." 

> > I'm not trying to take anything away from continuation-style programming,
> > Scheme, or any other language.  I'd actually like to learn about this
> > stuff.  What I'm wondering is whether we are pursuing purity at the
> > expense of maintainability and comprehension?
>
> Probably this is controversial, but maintainability and comprehension
> usually come from purity. The whole apparent complexity that you see
> in Lisp like languages is built on top of only 5 (five) operations:
> quote, variable reference, lambda, if, set variable and procedure
> call! Everything else is syntax only, which is translated to above 5
> operations.

And to some extent, the obtuseness and confusion of Lisp results directly 
from the concision of its basis.  Is there a real reason to pass up Rhino 
just because it won't do real continuations?  Isn't CPS sufficient for our 
purposes?







---------------------------------------------------------------------
To unsubscribe, e-mail: cocoon-dev-unsubscribe@xml.apache.org
For additional commands, email: cocoon-dev-help@xml.apache.org


Mime
View raw message