A question: AFAIK
Perl 6 will have optional type declarations
and I imagine the statically typed registers
in Parrot are there also to allow the
compiler to exploit them.
Given
my str $a="Perl" ; (is that valid Perl 6?)
the computations with $a will be
done using STRING registers?
Does that mean that STRING
operations should have
some Perl special/specific behavior?
I just wonder. Maybe it is not a problem
because STRING x STRING ops
in Perl have a quite vanilla semantics.
Will mixed-type ops always use PMCs?
*A long remark*
It seems to me that I have
some kind of understanding of how Parrot
will work. In my previous posting
I was just pointing out that the discussion
about string semantics is there because
there is a static STRING type directly
handled by the Parrot VM and also
there should be some basic helpers
for handling strings and so strings
semantics is cannot be simply delayed and
delegated to PMCs which OTOH
will be used to implement/expose the language
level string semantics.
I'm also aware that I have been spreading some
FUD about Python camp involvement ;),
(Given some of Guido's abstract statements
I think that I can't hurt more ;) ).
OTOH I'm aware that my opinions are not that
important.
I speak simply because I play regularly with
a 2nd implementation of Python. Ok it is
in Java, maybe a C reimplementation
is a lot easier. Maybe if PMCs are really
well designed it simply a matter of reorganizing
the code that define the variuos PyObjects
behaviors ;).
The FUD is: I see much work and few
added value. Clearly if some big fish
(PythonLabs, ActiveState...)
want to do much of the effort, things
could be completely different.
To add constructive (but abstract and remote)
points to the discussion:
David Simmons posting:
http://aspn.activestate.com/ASPN/Mail/Message/722505
(I mean his work and AOS project, which is OTOH
clearly a kind of very pie-in-the-sky target for
the open source language communities...)
http://citeseer.nj.nec.com/cache/papers2/cs/12671/http:zSzzSzwww.squeak.orgzSzo
opsla99_vmworkshopzSzsub.pdf/towards-a-universal-implementation.pdf
(this one is too much revealing of ;))
I find both insight-ful.
I also see that it is Ok for Parrot
to evolve as it is evolving now.
And that (for example after there is
a "JIT") it could possibly mature
other kind of implementations
philosophies for the supported languages. But on the other
hand I understand the actual relative lack
of enthusiasm outside Perl world...
regards.
