Re: So, what the heck is a continuation anyway?

[Adam Turoff <address@hidden>]

On Tue, Dec 11, 2001 at 09:53:05AM -0500, Matthias Felleisen wrote:
> 2. Do you actually think that you'd notice that you're using a VM that
> executes Scheme byte codes? Think about it. Take a look at an existing
> machine. 

You keep missing the central point.  Here are Dan's words again:

>   >   [C is] both widespread enough to have a reasonably
>   > 	large group of folks competent in it for our purposes, and
>   > 	on enough platforms to be give us a chance of running
>   > 	everywhere.

How does the bytecode format enter into the equation?  

Generating Scheme bytecodes is tantamout to generating Scheme.  So
instead of creating a cross-platform VM for multiple languages with
similar behaviors (Perl, Python, Ruby, Tcl, ...) we're simple compiling
them all down to Scheme.

Simon has said this is not exactly trivial.

Dan has said that the set of programmers sufficently skilled in Scheme
and interested in Parroty things is effectively nil, while the set 
of programmers interested in Parroty well skilled in C using platforms
well supported by gcc is quite vast by comparison.

I have said on multiple occasions that Parrot isn't intended to be
the sole VM for Parrot-hosted languages, but it is intended to be 
the first as well as the baseline.  That doesn't preclude Scheme-based
VMs (given a stable interface), but practical goals necessarily prevent
a Scheme-based VM from being the one-true-everyone-else-should-take-their-
marbles-and-just-go-home-already Parrot VM.

I don't think I can make the case any more clearly than that.  This isn't
a "your (language|VM) sucks" pissing contest.  Let's not turn it into one.

Z.