🦋 Welcome to the IRC channel of the core developers of the Raku Programming Language (raku.org #rakulang). This channel is logged for the purpose of history keeping about its development | evalbot usage: 'm: say 3;' or /msg camelia m: ... | Logs available at irclogs.raku.org/raku-dev/live.html | For MoarVM see #moarvm
Set by lizmat on 8 June 2022.
Geth rakudo/main: 52bf2b4b3f | (Nick Logan)++ (committed using GitHub Web editor) | src/Raku/ast/compunit.rakumod
Remove leftover debugging code
 00:42
Geth Lingua-Conjunction: packy++ created pull request #2:
Fix bug in :str parameter		 03:57
Geth Lingua-Conjunction/main: 09e66a609b | (Packy Anderson)++ (committed using GitHub Web editor) | lib/Lingua/Conjunction.rakumod
Fix bug in :str parameter

Currently, the documentation says :str "can use special sequence [|] (e.g. octop[us|i]) where string to the left of the | will be used when the list contains just one item and the string to the right will be used otherwise," but it uses the left of the | when the list contains one OR TWO items.
This change corrects the behavior to conform with the documentation.		 09:00
Lingua-Conjunction/main: 6c860063c9 | (Packy Anderson)++ (committed using GitHub Web editor) | lib/Lingua/Conjunction.rakumod
Merge pull request #1 from packy/packy-patch-1

Fix bug in :str parameter
Lingua-Conjunction/main: 372d37c38b | (Márton Polgár)++ (committed using GitHub Web editor) | lib/Lingua/Conjunction.rakumod
Merge pull request #2 from packy/main

Fix bug in :str parameter
Lingua-Conjunction/main: ff5971b50c | (Polgár Márton)++ | 4 files
1.1.1
 09:03
lizmat m: my int $a = 0; for ^10_000_000 { $a++ }; say now - INIT now; 09:09
camelia 0.08178685
lizmat m: my int $a = 0; for ^10_000_000 { $a++ }; say now - INIT now; 09:10
camelia 0.073605638
lizmat m: my int $a = 0; for ^10_000_000 { $a++ }; say now - INIT now;
camelia 0.075644316
lizmat m: my int $a = 0; for ^10_000_000 -> int $_ { $a++ }; say now - INIT now;
camelia 0.075035974
lizmat m: my int $a = 0; for ^10_000_000 -> int $_ { $a++ }; say now - INIT now;
camelia 0.074155608
lizmat hmmm
there's a pronounced difference on an M1 09:23
I guess JIT hides the difference on Intel CPUs] 09:28
MasterDuke has anyone else noticed that the output of `make m-test m-spectest` is now interleaved a bit? it used to complete the rakudo tests and then start the spectest, but now it'll start the spectest a little before the rakudo tests finish 11:43
lizmat I never run them like that, actually: I run "make test" usually before running "make spectest" as the former finishes faster, and if it find problems, then it usually doesn't make sense to run "make spectest" anyway 11:47
*finds
MasterDuke i'm 99.9% pretty sure it'll also stop if the `m-test` fails my way also, i think they're almost equivalent 11:53
lizmat TIL :-) 12:00
Geth rakudo/main: 512f5cc467 | (Elizabeth Mattijsen)++ | src/Raku/ast/compunit.rakumod
RakuAST: better document the --doc logic
 13:44
vrurg_ MasterDuke: make -j1 would be your solution, I think. 13:46
Geth rakudo/lizmat-use-rakumod: a62d4012e7 | (Elizabeth Mattijsen)++ | 16 files
Fix mention of .pm6 in many scripts
 13:50
Geth rakudo/lizmat-use-rakumod: c3bae076b6 | (Elizabeth Mattijsen)++ | CREDITS
Restore Ingy's D:
 18:11
lizmat aah... Github webhooks working again :-)
[Coke] We used to have assign/unassign ticket notices generating alerts, yes? 18:17
(make not on raku/doc)
*maybe*
lizmat yeah, they are on 18:19
it's just that Github is about 30 minutes late pushing them to Geth at the moment
[Coke] ahhh 18:22
then I apologize for the flood to #raku-doc that's coming 18:23
MasterDuke vrurg: yeah, but i never used to have to use `-j1` 18:54
[Coke] do you have TEST_JOBS set? 18:59
MasterDuke yeah, to 12. but my env has been mostly unchanged for years 19:00
ugexe MasterDuke: i wonder if its github.com/rakudo/rakudo/commit/1d...d2c37e49f0 19:08
i dont see how it would be that commit, it just looks like the most likely out of any recent commits to that Makefile.in 19:09
MasterDuke nope, just tried with that change reverted and same behavior 19:19
ugexe you regenerated the Makefile yeah? 19:30
cause i probably wouldn't have :P
MasterDuke i assume `perl Configure.pl --prefix=/home/dan/Source/perl6/install/ --backend=moar` will, because that's what i always run first 19:34
oh wow. i just reverted the revert (so i'm on the HEAD of my use_new_stat_syscall_in_DirRecurse branch), and now i got something i've never seen before during `+++ Installing files` 19:35
gist.github.com/MasterDuke17/89654...43deda35b1 19:36
i didn't do a make clean, or clear out my precomp dirs, but i do always run Configure.pl 19:43
vrurg MasterDuke: you started having that behavior when I resolved all dependencies and allowed parallelism in Rakudo Makefile. I'm pretty sure this is the moment. But up to my memory, make doesn't care where it got its targets from, it would behave the same.
MasterDuke when was that? because i'm pretty sure this started in the past couple months 19:49
vrurg Not sure. > half a year ago, perhaps. 19:55
Found. The last December is when I commended out .NOTPARALLEL 19:56
MasterDuke: BTW, you can uncomment .NOTPARALLEL in tools/tampltes/Makefile-common-rules.in and try again. Don't forget ./Configure.pl, of course! :) 20:01
On my side (macOS, GNU Make v4.4.1) it seemingly executes both targets sequentially. 20:02
MasterDuke as is right now? or with that uncommented? 20:04
vrurg As is, but I was wrong. The illusion of sequentiality was created by m-test finishing before m-spectest managed to finish preparing and start. 20:06
MasterDuke ha
guess you have a faster machine than i 20:07
vrurg Likely so, M2 Ultra.
MasterDuke yep, uncommenting that brings back the behavior i'm used to. e.g., test finishes then spectest starts 20:09
vrurg Anyway, I don't see this behavior as a problem. There is a problem with make which doesn't allow forced-sequential targets and I do miss this on occasion. But otherwise this is not an issue. 20:10
MasterDuke yeah, i think things actually work ok, it just was a change from what i was used to 20:11
vrurg And we always have `make m-test & make m-spectest` is necessary. BTW, I tried -O and -Orecursive (guess, it's the same) but there is no fun in waiting until it collects all spectest output first. 20:13
Ah, -Oline is no difference because `prove` is the line. 20:14
OK...

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!