Parrot 5.1.0 "Zombie Parrot" | parrot.org/ | Log: irclog.perlgeek.de/parrot | #parrotsketch meeting Tuesday 19:30 UTC
Set by moderator on 3 March 2013.
00:51 kurahaupo joined 04:45 preflex_ joined 08:38 woosley joined 09:35 bouncy joined 09:57 woosley left 10:28 donaldh joined 11:31 bbb joined 12:51 woosley joined 12:55 darbelo joined 13:54 bluescreen joined 14:16 PacoAir joined 14:18 PacoAir joined 14:23 Psyche^ joined
dalek p/vmarray-list: 47e8fee | (Arne Skjærholt)++ | src/QAST/Operations.nqp:
Refactor nqp::list_b to create an nqp::list instead of an RPA.
15:59
p/vmarray-list: 1ee8aa8 | (Arne Skjærholt)++ | src/ops/nqp.ops:
Make nqp::islist allow VMArrays.

RPAs are still allowed as well, since there may be some of those hanging around in the serialized bits of the bootstrap compiler.
p: cfdcc70 | jnthn++ | src/QAST/Operations.nqp:
Add nqp::resume(...) for resuming an exception.
18:02
p: 38a4fd9 | jnthn++ | t/nqp/44-try-catch.t:
Update test to use nqp::resume(...).

This should make it possible to try and get working in JVM land.
18:32 darbelo joined 19:07 PacoAir joined 19:15 darbelo_ joined
dalek p: cfd60cb | jnthn++ | src/QAST/Operations.nqp:
Better handling of repeat -> $x { ... } code-gen.

Still need a way to let Nil be specified, but at least this doesn't leave it unknown exactly what will be passed in.
20:00
kudo/md-nqp: 5c49111 | jnthn++ | src/core/Numeric.pm:
Fix a proto.
p: b94f739 | jnthn++ | src/guts/multi_dispatch. (2 files):
Make the cache a bit smarter about nameds.
21:48
kudo/md-nqp: 495550a | jnthn++ | src/Perl6/Metamodel/BOOTSTRAP.pm:
Allow common postcircumfix calls to be cached.

Before, they could not be because some candidates accepted named args. Now, the common named-argumentless case can be cached.
21:53
kudo/md-nqp: 3b39d8b | jnthn++ | src/Perl6/Metamodel/BOOTSTRAP.pm:
Remove an optimization that made things wrong.
22:26
kudo/md-nqp: f94fb8b | jnthn++ | src/Perl6/Metamodel/BOOTSTRAP.pm:
Remove bogus decontainerization.
22:27
p: ed35b15 | jnthn++ | src/ops/nqp.ops:
Ensure we don't overflow fates buffer.

A recent Rakudo change used fates without having a fate array in the first slot. This led to a buffer overflow, which showed up (at least on Win32) as spectest failures. This fixes it not to overflow.
23:06
kudo/md-nqp: 150bdc9 | jnthn++ | src/Perl6/Metamodel/BOOTSTRAP.pm:
Don't recalculate candidate lists without need.
23:29
kudo/md-nqp: 19d9ae0 | jnthn++ | src/Perl6/Metamodel/BOOTSTRAP.pm:
Fix logic for finding all possible candidates.
kudo/md-nqp: c9e3ae1 | jnthn++ | src/ (2 files):
Use NQP-based dispatcher for deferral also.
kudo/md-nqp: c4e8af0 | jnthn++ | src/ (3 files):
Rip out most of the C-based multi-dispatcher.

Only leftover bits are the compile-time dispatch analysis used by the optimizer, which still needs porting.