Why Parrot Matters

[Manny Swedberg]

Someone wrote (original msg lost from my archive):

What if Ruby also runs on Perl6's VM (Parrot)? Then you wouldn't need to
switch at all. I susupect that it's quite possible that we'll have Ruby
running on Parrot by the time Perl6 comes out (given that Perl6 is
still a couple of years away).

The Parrot team's firm intention is to have Parrot run Python and Ruby
just as well as Perl6. This is helped(?) by the fact that the plans
for Perl6 are so feature-rich (not to say -bloated that supporting
everything in it basically means supporting everything in Ruby.
Things that are in Ruby, but not Perl6, like continuations are slatted
to be added to Parrot anyways out of sheer good-neighborliness. It
should, in fact, be possible to compile *any* dynamic scripting
language into Parrot code: scheme, integer basic, befunge...whatever.

Because Perl6 is so far away, support for Ruby and Python is probably
actually going to come first. A big test, the first major public
showing of Parrot, is going to come at this year's O'Reilly
convention. Python/Parrot is going head to head benchmarking with
CPython. The loser gets a pie in the face; watch for it.

Parrot matters. To scripting-language hackers generally, to Ruby
hackers specifically, and to the Open Source movement as a whole.

Parrot promises to furnish a fast, portable environment for every
major scripting language. This will remove one of the big obstacles
to more widespread deployment: speed. Moreover, if I download a
Parrot VM to run someone's PyGame program on my machine, I already
have what I need to run your Ruby or Perl program without further
dependency worries: viral portability. Fast Ruby means more Ruby
hackers. Fast Python and Perl means more hackers in those languages
and thus more people who might take a look at Ruby; a common runtime
would make the transition even easier.

For OSS as a whole, Parrot promises a rival to Java or .Net without
corporate ownership, developed as open source, for languages that are
open source and in which tons of open source code is already written.
As the Gnome project considers a new development language, a timely
Parrot implementation could mean an in for Python, maybe even Ruby.
That would be *awesome*.

Parrot is a respectable ways along. Not by any means done, but more
than vaporware. Support for objects was recently added.

Here's the main Parrot page:
http://www.parrotcode.org/
(now 0.1.0)

Here's a project to make a Ruby->Parrot frontend:
http://rubyforge.org/projects/cardinal/
(now 0.0.3)

 

[gabriele renzi]

il 18 Apr 2004 01:46:48 -0700, (e-mail address removed) (Manny Swedberg)
ha scritto::

<snipall>
I agree with anything but 2 things: I believe perl6 will have first
class reified continuations, and I believe perl5 will run safely on
parrot before python and ruby

BTW I even hope that Dan Sugalsky wins the piethon, cause that would
mean that parrot is really on a good track .
And, I even hope that rite will blow off parrot

 

[Charles Comstock]

They aren't just supporting continuations for kicks. If I remember
correctly from my reading of there weekly dev list updates, there
fundemental model of computation is in terms of continuations.
Basically they allocate stack frames with continuations. So the entire
interpreter would fail without continuation support, allowing it to work
pretty nicely if you push it onto the actual language.

Charlie