On Wed, 21 Feb 2001, Ben Tilly wrote:
> The idea is, of course, much older than that.  The
> first interpreted language that I know of was Lisp.

Yes, yes, but I didn't want to go into details. Basically, COBOL didn't
even exist at the time.

> > > There was also talk of attaching various interperters (translators?) to 
> > > a single VM,
> I suspect that the original idea was mine.  I suggested
> that if the front end and back end for Perl were being
> made independent anyways, then the front end should be
> made pluggable.  This is an extension of hacks like
> Lingua::Romana::Perligata (accept source-code that looks
> like Latin).  I am not sure it is a good idea (in fact I
> called it a bad idea whose time may have come), but it
> certainly is an interesting one.

Attaching multiple parsers to a same language is the kind of thing you get
in Pliant. Multiple languages for the same backend is similar to the
JVM's current situation (something like 150-200 languages). 

> Perl's scalars are hideous beasts.  Basically a scalar can
> switch between types essentially at will, and an SV will
> store what the value is as a string, number, etc at the same
> time to speed this up.  Scalar values in Ruby are far, far
> simpler since everything is a straightforward object.

Perl Scalars are more like the 0-dimensional container I proposed in one
previous email, the way they are bound to the symbol table. Ruby has no
equivalent, as the symbol table is not typed and you can put anything in
it directly.

> >BTW, I don't think "virtual machine" is an appropriate word to describe
> >anything in what I want to do.
> Virtual machine can be used to mean virtually anything.
> Can't you stretch the term to fit what you want? :-)

Probably, yes. :-)

matju