gabriele renzi
5/1/2006 8:01:00 AM
harsh@computer.org ha scritto:
> The Lisp version runs pretty quickly on CMUCL. Run times for "fib(32)"
> on my machine are:
>
> - 0.8 seconds in the translated-to-Lisp version
> - 11.5s for vanilla Ruby 1.8.2
> - 3.8s for YARV 0.4.0/Ruby 1.9.0
> - 3.3s for Python 2.3.5
> - 5.0s for Perl 5.8.4
I guess this code is easily translatable in binary even via YARV's
ruby->C translator and/or ruby2c, it would be nice to see comparisons :)
> It's just a couple days' work, and doesn't yet support much of the
> language (basically just method definitions, instance variables, and
> method calls; and even support them fully). But if there's enough
> interest I'd probably keep working on it, and/or publicly post the
> soure code.
interestingly, the PyPy guys have an (outdated) CMUCL backend, so I
think at least the idea of running $language on a lisp implementation is
not so strange :)
It would be interesting to see what you've done, and I wonder if you
used some of the previous work such as ripper to build the AST or
ParseTree or metaruby's type inferencer or whatever.