#raku-dev on 7 December 2020 - Raku Programming Language Log

00:26 lucasb left 01:03 ggoebel_ left 01:24 ggoebel_ joined 03:24 leont left 03:53 maggotbrain joined 06:17 MasterDuke left

[Tux]

Yesterday:

07:02

Rakudo v2020.11-101-gd1d2546e6 (v6.d) on MoarVM 2020.11-57-g51cc76c16

csv-ip5xs	0.812	-	0.843
csv-ip5xs-20	7.797	-	8.252
csv-parser	24.894	-	26.065
csv-test-xs-20	0.388	-	0.394
test	7.544	-	8.334
test-t	1.860	-	1.909
test-t --race	0.805	-	0.864
test-t-20	32.072	-	32.441
test-t-20 --race	9.346	-	10.092

07:38 domidumont joined 07:47 patrickb joined

patrickb

.tell rba raku.org is down. Can you have a look?

07:54

tellable6

patrickb, I'll pass your message to rba

patrickb

ping rba

tellable6

2020-12-03T21:30:43Z #raku <samcv> patrickb no I don't use Raku at work, sadly

2020-12-03T22:05:51Z #raku <tony-o> patrickb i'm interested in it, for sure. ot

2020-12-03T22:06:34Z #raku <tony-o> patrickb it's been kind of a pain in my side, though. i have a working prototype and the code to run it. i need to figure out how to make it work in zef.

2020-12-03T22:06:56Z #raku <tony-o> patrickb in response to your ecosystem inquiry.

07:54 patrickb left 08:14 MasterDuke joined

nine

lizmat: Thanks for finding those! That's exactly what I wanted to look into today as raku -e '(^10000).race.map({EVAL "q:to/END/;\nEND\n"}).sink' still fails very rarely with some message about a cursor

08:53

Though it looks like those are not actually the reason for the failures I see.

09:25

nqp: class Foo { }; role Bar { has $!bar; method set-bar() { $!bar := 1; } }; my $i := 0; my @threads; while $i < 16 { my $thread := nqp::newthread({ while 1 { my $foo := Foo.new; Foo.HOW.mixin($foo, Bar); $foo.set-bar } }, 0); nqp::push(@threads, $thread); nqp::threadrun($thread); $i++ }; nqp::threadjoin(@threads[0]);

09:26

camelia

(signal XCPU)

nine

Locally this immediately gives me "Unhandled exception: P6opaque: no such attribute '$!bar' on type Foo+{Bar} in a Foo+{Bar} mixin when trying to bind a value"

Which is the same as during compilation: P6opaque: no such attribute '$!named' on type QAST::IVal+{QAST::SpecialArg} in a QAST::IVal+{QAST::SpecialArg} mixin when trying to bind a value

09:27

10:02 frost-lab joined

nine

Oh, it seems like the error happens either immediately or not at all. This suggests that the race condition is in generating the mixin. Once all threads get the type from the parametrics cache it seems to be fine

10:09

10:47 domidumont1 joined 10:49 domidumont left 11:33 domidumont1 left 11:47 domidumont joined 11:48 domidumont left 11:54 domidumont joined 12:06 domidumont left

lizmat

Files=1346, Tests=117144, 227 wallclock secs (30.55 usr 7.58 sys + 3150.20 cusr 291.29 csys = 3479.62 CPU)

12:16

12:26 domidumont joined 12:43 domidumont left

Geth

nqp: 84ff7e5edd | (Elizabeth Mattijsen)++ | tools/templates/MOAR_REVISION
Bump MoarVM to get nwc10's extensive hash work

13:02

rakudo: 6792cc425f | (Elizabeth Mattijsen)++ | 2 files
Use still slightly faster explicit nqp::add_i vs prefix ++

Looks like the ++$i codegens to a bare IVal, whereas the nqp::add_i codegens to a WVal containing an IVal. Yet the WVal version turns out to be about 1% faster in benchmarks still.