whiteknight/imcc_interfaces branch ready for testing

[Peter Lobsinger]

On Mon, Jan 24, 2011 at 11:49 AM, Andrew Whitworth
<wknight8111 at gmail.com> wrote:
> The whiteknight/imcc_interfaces branch is ready for general testing.
> It implements a series of cleanups for the IMCC interface functions,
> and should preserve most existing functionality. These changes are the
> basis for additional work I am doing in this area. Major changes
> include:
>
> * the PIR and PASM compreg compilers use NCI signatures "PJS" instead
> of "PJt". This should be a transparent change since NCI was handling
> S->t translations automatically. From the PIR-user perspective,
> there's no change here.

Great to hear. More Parrot-ish interfaces!

> * The PASM emitter for IMCC was ripped out. It's no longer possible to
> input PIR code and output PASM to stdout like used to be possible.
> This feature was always broken, and the old RT ticket that described
> it was closed as WONTFIX back in 2006.
> * Along with the above, all  "pir_2_pasm*" functions in Parrot::Test
> were ripped out, and I deleted all tests that relied on those. Most of
> those were extremely esoteric tests, or tests for the IMCC optimizer

Good ridance.

> There are also a handful of other smaller cleanups that happened in this branch.
>
> I've tested this branch on my machines with building Parrot and NQP. I
> will run Rakudo tests tonight. Any other testing would be much
> appreciated.

t/compilers/imcc/syn/regressions.t#14 (die in immediate) fails testr
(linux x86_64, gcc 4.5.2, --optimize).

> Thanks,
>
> --Andrew Whitworth
> _______________________________________________
> http://lists.parrot.org/mailman/listinfo/parrot-dev
>