01:34 frost-lab joined 01:49 lucasb left 02:41 leont left
Geth rakudo/faster-slice-access: 08a6192046 | (Elizabeth Mattijsen)++ | 2 files
Simpler dupe module checking
05:20 SmokeMachine left, SmokeMachine joined
Geth rakudo/faster-slice-access: 1d95a5108c | (Elizabeth Mattijsen)++ | src/core.c/array_slice.pm6
Micro-opt of array slice access
06:17 Altai-man joined
Geth rakudo/faster-slice-access: 2c09f31bdd | (Elizabeth Mattijsen)++ | 2 files
Fix issue with :!exists:delete:!kv
rakudo/faster-slice-access: 74c267efb4 | (Elizabeth Mattijsen)++ | src/core.c/array_slice.pm6
Re-imagine @a[*] to use new infrastructure (60% faster)

This is on top of the 5x that was already achieved by optimizing the @a[@i] candidates (which the @a[*] was calling internally). So the whatever slice is now 8x as fast as before.
lizmat goes back to bed& 06:39
Geth ¦ problem-solving: JJ self-assigned Documentation and the ecosystem github.com/Raku/problem-solving/issues/252 08:36
09:10 sena_kun joined 09:12 Altai-man left, frost-lab left
gfldex m: class C { method ^foo {} } 09:41
camelia ( no output )
gfldex where is this form of method declaration handled in Rakudo?
I cry for your aid, there is a blogpost at peril! :) 09:42
nine What do you mean by "handled"? 10:14
10:19 frost-lab joined
gfldex the compiler is creating a mixin of Metaclass::ClassHOW and I cant find where in the rakudo source. 10:21
10:22 frost-lab left
nine look for 'method mixin' 10:23
gfldex thanks 10:28
10:34 frost-lab joined
nine Aaah NQP my dear, dear friend. How on earth is it possible to get a "Cannot find method 'qast' on object of type NQPMu" here when $block is checked to be a BlockInfo directly above? github.com/Raku/nqp/blob/master/sr....nqp#L2917 11:41
Or is it? 11:42
11:42 leont joined
timotimo NQPMu showing up somewhere is almost always an undefined identifier 11:42
nine No, it actually isn't. Because BlockInfo is not yet declared at that point, so it's implicitly NQPMu. So it's really an istype($block, NQPMu) which is always true
timotimo right
that's a not great feature of nqp
Geth rakudo/faster-slice-access: 511264b001 | (Elizabeth Mattijsen)++ | 2 files
Re-imagine @a[] (zen-slice) to new infrastructure

This moves the handling of zen-slices and its adverbs to the new infrastructure, making the bare zen-slice 12x as fast, and a zen-slice with adverbs about 6x as fast.
This breaks a small number of tests in roast because the error message ... (5 more lines)
12:38 frost-lab left
Geth nqp/getlex_fallback: 3d7cf71cf0 | (Stefan Seifert)++ | 13 files
Get rid of wrapper frames
rakudo/getlex_fallback: 8505161962 | (Stefan Seifert)++ | src/Perl6/World.nqp
Replace artificial lexical outer frame with runtime lookup
rakudo/getlex_fallback: f643e650e0 | (Stefan Seifert)++ | 4 files
First working version
13:09 Altai-man joined 13:12 sena_kun left
[Tux] Rakudo v2020.12-9-g4b6b2e070 (v6.d) on MoarVM 2020.12-7-g6bf54d784
csv-ip5xs0.803 - 1.145
csv-ip5xs-207.752 - 8.890
csv-parser27.285 - 29.101
csv-test-xs-200.381 - 0.405
test7.921 - 8.200
test-t1.979 - 2.030
test-t --race0.874 - 0.969
test-t-2031.991 - 32.683
test-t-20 --race9.582 - 10.077
14:36 linkable6 left, evalable6 left 14:37 evalable6 joined 14:38 linkable6 joined
nine win 1 15:02
Geth rakudo/faster-slice-access: e454c45440 | (Elizabeth Mattijsen)++ | src/core.c/Rakudo/Internals/PostcircumfixAdverbs.pm6
Use nqp ops in dispatch table lookup

This saves 2 IntLexRef allocations per call, making something like
  @a[@i]:p about 7% faster.
17:10 sena_kun joined 17:11 Altai-man left 17:17 epony joined
Geth rakudo/faster-slice-access: acd5a6527d | (Elizabeth Mattijsen)++ | src/core.c/Rakudo/Internals/PostcircumfixAdverbs.pm6
Some more NQPfication of dispatch table lookup

Saves about 4 BOOTCode objects per call
17:45 lucasb joined
Geth rakudo/faster-slice-access: c626ca500e | (Elizabeth Mattijsen)++ | src/core.c/Rakudo/Internals/PostcircumfixAdverbs.pm6

Handles the case where the adverbs hash is missing one named parameter already, so it needs to check them seperately. Needed for candidates such as (:$delete!, *%_)
18:11 b2gills left 18:25 b2gills joined
lizmat and that concludes my hacking for this side of Christmas Eve 18:44
best wishes to everybody, stay safe and stay well and have a good time!
[Coke] ~~ 18:46
18:54 lizmat_ joined 18:58 lizmat left 19:05 lizmat_ left 19:06 lizmat joined 19:19 lizmat_ joined 19:21 lizmat left 19:32 lizmat_ is now known as lizmat
sena_kun (not the best time, but) Apparently, I don't have enough time and knowledge to maintain whateverable bots, so please do not count on me if anyone does. Volunteers are welcome. 20:27
MasterDuke i can provide ad hoc assistance with setup and maintenance, but can't pony up to be solely responsible for them 20:31
sena_kun MasterDuke, writing a docker-based setup is what is needed, plus a bit of rewriting for hardcoded directories/urls. 20:32
Responsible is maybe not the right word. I mean, the maintainer to keep up with a lot of issues would be awesome to have, but the main goal is to keep it alive. Plus our releases depend on bots, so no bots == no releases. 20:33
MasterDuke ah, i don't know docker
i am relatively familiar with the bots, so looks like we need someone who knows docker 20:34
21:09 Altai-man joined 21:11 sena_kun left 21:57 Altai-man left
ugexe patrickb: there are cpan examples in the Appveyor file 22:14
tellable6 ugexe, I'll pass your message to patrickb
23:10 lizmat_ joined 23:13 lizmat left 23:15 lizmat_ is now known as lizmat 23:47 Kaiepi left