The Parrot is dead. Long live the Parrot?
christoph at mksig.org
Mon Feb 11 16:45:33 UTC 2013
On Sun, Feb 10, 2013, at 11:46, Allison Randal wrote:
> On 02/10/2013 06:15 AM, Gerhard R. wrote:
> > However, even if you remove the baggage and optimize for Rakudo (or
> > rather NQP) integration, Perl6-on-Parrot will still be a 2nd-class
> > citizen as long as no one replaces the PMC-based object system with
> > 6model, which is imo a major undertaking; it might be easier to restart
> > from scratch and fix other 'mistakes' along the way. That's where
> > Parrot2 comes in.
> Parrot barely has enough human resources to handle one implementation
> path. It certainly doesn't have enough to handle two simultaneous
> implementation paths.
> You're headed in the right direction, but still thinking too
> conservatively. And a restart from scratch is *never* easier.
Thanks, Allison. I'm glad that you're providing some opinions and
much-needed continuity of thought. Big audacious plans are valuable,
but more important is to understand why previous versions of those plans
failed and to avoid those same problems.
I had a response all written up, and while I was thinking it through a
dozen other replies flew by and made it obsolete. I was going to bring
up the idea of Parrot supporting Rakudo exclusively, but that idea is
already gaining some traction. It's a shame to see really cool things
like HLL interop fall by the wayside, but it's going to be much easier
focusing exclusively on the needs of one language than it will be to
continue trying to be all things to all languages poorly.
We've had too many good plans that didn't get fully implemented. A
reduced and more tightly focused scope could help us by giving us a
concrete goal. In my head Parrot's shortage of developer tuits over the
past year has been due to lack of focus and an inability to display
sufficient progress, in spite of heroic efforts on the part of a few
developers. If those of us who believe in this refined vision for
Parrot can get focused, I believe that other developers will find the
tuits to start contributing again.
It's clear that Parrot needs to take the first few steps here to
convince the Rakudo folks that any such plan will have enough tuits
behind it to make it viable. I don't have a fully-formed plan yet but a
good first step is to start removing PMCs, ops and compilers that Rakudo
doesn't care about and that therefore only slow down the build. Coke++
and moritz++ have already started this work and I'm excited to see how
far they can run with it.
My mission in life right now is to figure out a roadmap. It's exciting
to see the uptick in interest in Parrot. I hope we'll put that energy
to good use.
More information about the parrot-dev