jaleto at gmail.com
Tue Jun 29 02:12:40 UTC 2010
Thank you very much for this great work bacek++ !
I will help as much as I can in the name of replacing IMCC.
Are you using the github issue tracker or the Parrot Trac instance
for bug/issue tracking?
On Mon, Jun 28, 2010 at 2:57 PM, Vasily Chekalkin <bacek at bacek.com> wrote:
> I think it's now time to "formally" introduce PIRATE on which I,
> cotto, tcurtis and other people work recently.
> "PIRATE is PIR compiler implemented in NQP/PCT". Main reasons behind
> development (for me at least):
> 1. -Ofun. It was interesting expirience to implement compiler for
> somehow useful language.
> 2. Ability to emit PBC directly from POST.
> 3. Retire IMCC in future.
> PIRATE consists of two large parts:
> 1. PIR parser.It's mostly done. 2 things isn't finished yet: proper
> string handling (encoding/charset/interpolation)
> 2. POST pbc emitter.
> For ability to emit PBC from POST I extended POST from parrot trunk to
> have more semantically reach. E.g. trunk's POST::Sub.add_param just
> add chunk of text. POST::Op's arguments are just junks of texts, etc.
> In PIRATE I introduced various POST::* classes insted. E.g.
> POST::Register to hold registers, POST::Constant for constants, etc.
> Currently PIRATE can emit "sub-optimal" PBC for simple pir files. E.g.
> patched fib.pir (there is no "@ARGV" support atm).
> tcurtis++ work on "real-life" application for his GSoC's past
> optimization framework.
> I'm going finish PBC emiting in terms of fully support required
> features (sub modifiers, different call typess, etc).
> After this we can update PAST::Compiler to emit "new" POST and stop
> depending on IMCC for PCT-based languages.
> Using of PIRATE as drop-in replacement of IMCC is in
> "nice-to-have-but-there-is-no-dealine" todo list.
> P.S. It's on http://github.com/bacek/pir/. Anyone with signed CLA can
> ping me on IRC or via email to get commit bit.
Jonathan "Duke" Leto
jonathan at leto.net
More information about the parrot-dev