01:48 Xliff joined
Xliff Hi all... 01:48
01:49 Xliff left 03:52 lucasb left 03:53 Xliff joined
Xliff . 03:53
tellable6 2020-10-28T01:59:14Z #raku <[Coke]> xliff are you memory constrained on that win build?
Xliff [Coke]: No. I have 64 gigs on this machine.
.tell [Coke] No, I have 64 gig on this box. I just reran it again. Memory never grew above 10.1 total gigs usage 04:00
tellable6 Xliff, I'll pass your message to [Coke]
04:16 Xliff left 04:19 Xliff joined 05:10 Kaiepi left 05:20 frost-lab joined 05:41 MasterDuke left 06:37 nebuchadnezzar joined 07:35 Altai-man joined 07:56 Altai-man left, Altai-man joined 08:08 sena_kun joined 08:10 Altai-man left 09:58 domidumont joined
lizmat is Geth AWOL again? tyil ? 10:32
tyil kicking geth ~ 10:33
10:33 Geth joined
Geth rakudo: Altai-man assigned to lizmat Issue New Trademark font characters are right-clipped in the REPL producing an unfortunate visual result github.com/rakudo/rakudo/issues/3986
329e47f435 | (Elizabeth Mattijsen)++ | src/Perl6/Compiler.nqp

The message displayed when starting the REPL stays the same. This is a slightly hacky use of the :short-versions flag, but I guess it will do for now at least.
11:29 frost-lab left 11:34 Kaiepi joined 11:35 Kaiepi left 11:37 Kaiepi joined 11:46 Kaiepi left 11:47 Kaiepi joined 11:48 Kaiepi left 12:07 Altai-man joined 12:08 Kaiepi joined 12:10 sena_kun left 14:31 Xliff left
[Coke] lizmat: did you see the / @keywords / vs. .contains(any(@keywords)) speed diff? 15:07
tellable6 2020-10-28T04:00:21Z #raku-dev <Xliff> [Coke] No, I have 64 gig on this box. I just reran it again. Memory never grew above 10.1 total gigs usage
lizmat [Coke]: no 15:11
[Coke] lizmat: my script went from 120s to 5s.
(for the opcode tests)
lizmat ah, yes, well INTERPOLATE is a bitch :-)
[Coke] github.com/Raku/nqp/commit/d1b2c12...74bf034cb5 15:12
it also didn't help to extract the regex creation outside of the loop
in NQP, this was still very fast.
lizmat hmmm... well, I guess that's because it cannot know that @keywords didn;'t change inbetween? 15:13
[Coke] if I did my $re = / @keywords / I would not expect that to track changes in @keywords 15:15
lizmat I would have to check, but if you didn't see a difference there, I'd suspect it would be interpolating the array every time again 15:16
[Coke] raku: my @keywords = < this that>; my $re = / @keywords /; dd "this is fun" ~~ $re; dd "what is fun" ~~ $re; @keywords.push: 'what'; dd "what is fun" ~~ $re; 15:17
evalable6 Match.new(:orig("this is fun"), :from(0), :pos(4))
Match.new(:orig("what is fun"), :from(0), :pos(4))
[Coke] ... so yah, it is doing it every time.... that feels like a bug to me
15:19 domidumont left
lizmat [Coke]: please make an issue if there isn't one already :-) 15:19
[Coke] rakudo or ps? 15:20
lizmat ah... hmmm.... 15:21
rakudo for now, I'd say
[Coke] lizmat: github.com/rakudo/rakudo/issues/3988 15:27
ugexe i'm not immediately sure how it could Do The Right Thing in regards to multithreads 15:28
similar to once/state
[Coke] if it's "works as intended", I'll just make sure we update the variable interpolation regex section in the docs. 15:29
(and make sure we have a test, I guess)
ugexe although i agree this behavior is not intuitive
yea i dunno if its intended. it just feels like it might be a consequence of other design decisions
lizmat if @keywords can be proven to be a value type, it could not repeat the interpolation, I guess 15:30
but sadly, even <foo bar> is not a value type :-( 15:31
Geth nqp: b64cd90dce | Coke++ | docs/ops.markdown
Document sub_i64 (js)
nqp: 4b95f4e604 | Coke++ | t/docs/opcodes.t
only create junction once
[Coke] that shaves a second off my runtimes.
lizmat yup, not having to do the same work every time helps :-) 15:33
[Coke] :)
do we have any thoughts about being able to disable the diag output on failing tests in Raku via an env var? 15:34
(I don't need the line number of the failing test for the kind of tests I'm running, where it's 100 instances of the same test at the same line) 15:35
ugexe you could just write the TAP output yourself 15:40
although maybe it does make sense for a harness, not Test, to output that extra info 15:41
[Coke] easy enough to make a PR for folks to point at. 15:47
(working on it.)
RAKU_TEST_DIAG_LINE_ON_FAIL=0 make test... 15:48
Geth rakudo/coke/testdiag: b7e902c628 | Coke++ | lib/Test.rakumod

Defaults to 1. If set to empty/0, disables the diag output in testing that adds file/line number information on failed tests.
16:08 sena_kun joined 16:10 Altai-man left
Geth rakudo: coke++ created pull request #3989:
[Coke] ^^ 16:15
16:36 Kaiepi left 16:37 sena_kun left
[Coke] 10 more ops to go to get under 200 17:50
(for docs)
Geth rakudo: 39c5159573 | (Elizabeth Mattijsen)++ | src/core.e/array_multislice.pm6
Slight refactor of v6.e postcircumfix[;]

  - better fast pathing of simple cases
  - simplification of recursive cases
  - deleting non-existing elements returns Nil
  - add support for :!exists
  - fix several decont issues
Instigated by seeing results of writing extensive tests.
lizmat afk for a few hours&
18:29 patrickb joined 19:26 Xliff joined 20:00 camelCaser left 20:03 camelCaser joined 20:08 camelCaser left 20:10 lichtkind joined 20:22 camelCaser joined 20:25 camelCaser left 20:26 camelCaser joined 20:31 MasterDuke joined 20:45 camelCaser left 20:47 camelCaser joined 20:55 camelCaser left 20:57 camelCaser joined 21:07 ccamel joined, camelCaser left 21:59 timotimo left, sivoais left, sivoais joined, timotimo joined 22:16 Xliff left 22:23 patrickb left 23:17 Kaiepi joined 23:32 Kaiepi left