00:01 travis-ci joined
travis-ci Rakudo build passed. Elizabeth Mattijsen 'Disable failing test for now' 00:01
travis-ci.org/rakudo/rakudo/builds/648657065 github.com/rakudo/rakudo/compare/5...abf8d5cf25
00:01 travis-ci left 00:04 vrurg joined 00:24 SqrtNegInf left 00:36 lucasb left 00:50 AlexDaniel joined 00:51 AlexDaniel left, AlexDaniel joined 01:34 sena_kun left 01:43 titsuki joined 01:45 epony left, epony joined 01:47 sena_kun joined 01:56 Kaiepi left, Kaiepi joined 02:12 ufobat__ joined 02:16 ufobat_ left 03:04 finsternis left 03:06 finsternis joined 03:14 finsternis left 03:15 finstern1s joined 03:22 finstern1s is now known as finsternis 03:34 sena_kun left 03:39 Geth left 03:48 sena_kun joined 03:59 moon-child left 04:04 moon-child joined 04:17 hungryd87 joined 04:20 hungrydonkey left 04:40 moritz left, moritz joined 04:53 squashable6 left 04:55 squashable6 joined 05:34 sena_kun left 05:46 sjn joined 05:48 sena_kun joined 06:00 ufobat__ left 06:34 titsuki left 07:27 stoned75 left 07:33 stoned75 joined, sena_kun left 07:38 domidumont joined 07:41 hungrydonkey joined 07:43 hungryd87 left 07:47 hungryd16 joined, sena_kun joined 07:49 hungrydonkey left 07:53 hungrydonkey joined 07:54 stoned75 left 07:56 hungryd16 left 08:09 patrickb joined 08:20 Ven`` joined 09:01 hungryd15 joined, hungrydonkey left 09:06 Voldenet left 09:11 Voldenet joined, Voldenet left, Voldenet joined 09:33 sena_kun left 09:46 sena_kun joined
jnthn lizmat: But that makes it harder for other implementations, because to be a drop-in replacement they have to emulate something with RAKUDO in its name. 10:03
MasterDuke theses are the changes i made to related to rand_I for the ltm v1.2.0 update. github.com/MoarVM/MoarVM/pull/1221...1151-L1206 and github.com/MoarVM/libtommath/commi...c210deac64 10:32
i'm not entirely sure how srand was effecting it before 10:34
maybe the ltm functions were calling rand?
but if they aren't now, then maybe we need to not use them at all, since it doesn't look like they can be seeded 10:35
ah ha. the ltm v1.1.0 changelog has: "mp_rand() has now several native random provider implementations and doesn't rely on `rand()` anymore" 10:37
nine Could we use mp_rand() by default and switch over to rand() once srand has been used? 10:46
MasterDuke is there an easy way to tell if srand has been used? or do we need to do something like set some flag hanging off tc? 10:51
or should we just also use our tinymt for rand_I (which we use for rand_n)? 10:53
github.com/MoarVM/MoarVM/blob/mast...082-L1090, and then we can not use srand() at all 10:54
11:17 timotimo left, timotimo joined 11:34 sena_kun left 11:42 squashable6 left 11:45 squashable6 joined 11:47 sena_kun joined
patrickb jnthn: I think the underlying question is whether the WHATEVERLIB env var is a rakudo specific feature / implementation detail or something we consider part of Raku in general. If it's something Raku should have in general then it should be called RAKULIB. If we consider it to be an implementation detail, then a replacement shouldn't need to 12:08
implement it.
lizmat Files=1302, Tests=109841, 209 wallclock secs (28.36 usr 8.27 sys + 2922.59 cusr 275.88 csys = 3235.10 CPU) 12:15
jnthn patrickb: PERL6LIB is covered by the spectest, and so was part of the language specification 12:16
Well, is part of...
So at a language design level, we'd be saying that we plan to deprecate it and not replace it
And that it's implementation-defined
We can, but that probably warrants discussion of why 12:17
patrickb It it the thing corresponding to `-Isomelibfolder`? If yes then my gut feeling tells me it's very much a Raku thing and every implementation should support it.
jnthn Yes, it's the env var version of -Ifoo
My gut feeling matches yours, I'm just aware I may have missed a discussion that had reasons to do otherwise :)
patrickb I have had several such discussions in the past months, but IIRC all about other parameters (e.g. PERL6_HOME -> RAKUDO_HOME). 12:19
jnthn RAKUDO_HOME is perhaps reasonable in so far as iiuc it's a detail of how Rakudo deals with relocatability, and we'd probably not want to tie every impl into that 12:21
patrickb If I interpret git history correctly, both PERL6LIB and RAKUDOLIB have been supported for a long time. I'd guess the corresponding RAKULIB alias for PERL6LIB was missed in the past rename refactors.
jnthn Ah, interesting. 12:22
patrickb jnthn: WRT RAKUDO_HOME That's how I see it as well. Having that as a general Raku variable would mean we'd force different implementations to install into the same folder. That's just provoking collisions. We don't want to do that. 12:25
jnthn Yeah, makes sense. 12:28
OK, so it looks like it's simply that nobody added a RAKULIB yet, not that RAKUDOLIB was chosen over it. 12:29
patrickb++
|Tux| Rakudo version 2020.01-118-gffabf8d5c - MoarVM version 2020.01.1-33-g571c731ff
csv-ip5xs0.694 - 0.698
csv-ip5xs-205.732 - 5.970
csv-parser22.671 - 23.398
csv-test-xs-200.364 - 0.365
test7.226 - 7.303
test-t1.746 - 1.766
test-t --race0.792 - 0.794
test-t-2029.773 - 29.966
test-t-20 --race8.464 - 8.509
12:31 MasterDuke left 12:42 SqrtNegInf joined
lizmat hmm... had just t/spec/S10-packages/basic.t fail in 47-48 and 50 and crash on 51 13:10
oddly enough 47 and following tests issue warnings for uninitialized values of $repo-id in string context 13:11
so feels to me we still have some gremlins in there related to throwing exceptions
13:12 lucasb joined
lizmat hmmm... looks like we lost Geth again, pinging tyil 13:16
tyil o/ 13:17
13:17 Geth joined
lizmat just pushed github.com/rakudo/rakudo/commit/25...ae5430ebae 13:17
13:17 rypervenche left 13:21 rypervenche joined 13:24 Ven`` left
Geth nqp: e8afe42a73 | (Elizabeth Mattijsen)++ | docs/ops.markdown
Document nqp::decodelocaltime
13:25
13:33 sena_kun left
lizmat wonders where nqp::p6decodelocaltime is being used 13:40
nine lizmat: probably not at all any more 13:44
lizmat: which you seem to have discovered already about a year ago :)
lizmat yeah, I had a vague feeling of deja vu :-)
13:48 sena_kun joined 14:15 AlexDaniel left
Geth rakudo: b19c4eec8e | (Elizabeth Mattijsen)++ | 2 files
Make R:I.get-local-timezone-offset about 19x as fast

  - by bypassing DateTime object creation altogether
  - directly using the nqp::decodelocaltime / time_i ops
  - changed API to accept the epoch value to use
  - adapted *TZ setting to give the epoch value to use
... (5 more lines)
14:22
lizmat Working on this, I realized two things: 14:26
- $*TZ is a misleading name, it should probably be $*TZ-OFFSET
as it does not contain any actual (named) timezone information
- is constant once set
so would be incorrect after a DST change during the run of a process
jnthn I thought the idea was that $*TZ can be any object, so long as it numifies appropriately? 14:30
So you could use a module that provides more enriched values, but CORE only needs its numeric nature for actual calculations. 14:31
lizmat the documentation currently does not reflect that 14:32
but if that's the intent, then I will document as such :-)
nine And another one: [ 139s] Unhandled exception in code scheduled on thread 4 14:39
[ 139s] push_o requires a concrete object (got a ThreadPoolScheduler::Queue type object instead)
build.opensuse.org/package/live_bu...ctory/i586
Geth rakudo: f4ad0c2efd | (Mikhail Khorkov)++ | lib/Test.rakumod
#2364 Make throws-like throw an exception on Bool:D matchers

Now, throws-like and fails-like Test subs throw an Exception in case the matcher uses Bool:D as matcher value instead of explicit Bool:D check. The behavior is similar to .grep(True).
15:03
linkable6 RAKUDO#2364 [open]: github.com/rakudo/rakudo/issues/2364 [Test.pm][easy to resolve] throws-like should die on Bool:D matchers
Geth rakudo: 50f084d834 | (Mikhail Khorkov)++ | lib/Test.rakumod
#2364 Fix review

Now check uses smartmatching over identity operator.
rakudo: f2b3091976 | (Vadim Belman)++ (committed using GitHub Web editor) | lib/Test.rakumod
Merge pull request #3464 from atroxaper/2364-throws-like-bool

Make throws-like throw an exception on Bool:D matchers
roast: c923109df6 | (Mikhail Khorkov)++ | S24-testing/fails-like.t
Add tests for fails-like with Bool:D matcher
roast: 5aca1aa55d | (Vadim Belman)++ (committed using GitHub Web editor) | S24-testing/fails-like.t
Merge pull request #616 from atroxaper/2364-throws-like-bool

Add tests for fails-like with Bool:D matcher
15:34 sena_kun left 15:47 sena_kun joined
Geth rakudo: vrurg++ created pull request #3473:
Correct the order of checks in Bool condition
16:11
rakudo: 5696dbd1d4 | (Vadim Belman)++ | lib/Test.rakumod
Correct the order of checks in Bool condition

DEFINIED must go first.
rakudo: cdc907b205 | (Vadim Belman)++ (committed using GitHub Web editor) | lib/Test.rakumod
Merge pull request #3473 from vrurg/test-matcher-bool

Correct the order of checks in Bool condition
roast: ebf9054656 | (Vadim Belman)++ | S05-metasyntax/regex.t
Make correct Bool matchers

  :attr(? *) instead of just :attr
16:12
rakudo: 8dcf1a8e2b | (Jeremy Studer)++ | src/Perl6/Grammar.nqp
Change err msg for pos after named in signature

If putting a positional paramater after a named parameter in a signature, the error previously stated that you cannot have a positional after a variadic. Change to be more accurate.
16:15
rakudo: ab9c1ffa64 | (Jonathan Worthington)++ (committed using GitHub Web editor) | src/Perl6/Grammar.nqp
Merge pull request #3467 from jstuder-gh/named_param_lta_error_msg

Change err msg for pos after named in signature
rakudo: ca1e51c83f | (Daniel Green)++ | src/core.c/Stash.pm6
Add the Stash's name to the Failure when trying...

to look up a non-existent symbol in the Stash.
16:17
rakudo: 774a839c4d | (Jonathan Worthington)++ (committed using GitHub Web editor) | src/core.c/Stash.pm6
Merge pull request #3465 from MasterDuke17/add_some_more_context_to_could_not_find_symbol_failure

Add the Stash's name to the Failure when trying...
16:18 patrickb left
Geth rakudo: 6fbfc4a0c3 | (Konrad Bucheli)++ | src/core.c/Channel.pm6
properly close Channel Supply
16:23
rakudo: 8390016495 | (Jonathan Worthington)++ (committed using GitHub Web editor) | src/core.c/Channel.pm6
Merge pull request #3471 from kbucheli/propberly_close_channel_supply

properly close Channel Supply
japhb (merge fest)++
Geth roast: b458b1db40 | (Konrad Bucheli)++ | S17-channel/basic.t
channel elements not yet consumed in the supply should still be accessible
roast: e4ffee22ec | (Jonathan Worthington)++ (committed using GitHub Web editor) | S17-channel/basic.t
Merge pull request #617 from kbucheli/properly_close_channel_supply

channel elements not yet consumed in the supply should still be accessible
16:25 domidumont left 16:29 patrickb joined 16:47 hungryd15 left 16:50 Phi joined
Phi Hello, has anyone tried to compile Raku on Termux 16:51
? 16:52
nine Phi: you mean on Android? 16:59
Then the answer is yes
17:00 Phi left
Geth rakudo: 50ca2ed986 | (Jonathan Worthington)++ | src/Perl6/Grammar.nqp
s/// parts should be thunks, not lexical scopes

This difference became observable when a placeholder parameter was used in either the match or the substitution part, and then tried to attach to the wrong place. Fixes #3358.
17:09
linkable6 RAKUDO#3358 [closed]: github.com/rakudo/rakudo/issues/3358 Odd behavior when using `$^a` within a substitution regex
Geth roast: f48ebe0159 | (Jonathan Worthington)++ | S05-substitution/subst.t
Test placeholder params used in S/// construct

Covers github.com/rakudo/rakudo/issues/3358.
17:10
17:33 sena_kun left
Geth rakudo: 4a954cbc2b | (Jonathan Worthington)++ | src/core.c/ShapedArray.pm6
Fix clone of multi-dim array to be deep enough

We could potentially optimize this with a STORE that has to do a bit less checking, because the shape is already good. However, in the meantime, better to actually be correct!
17:37
roast: 87b4553e3a | (Jonathan Worthington)++ | S09-multidim/methods.t
Test clone of a shaped array

It should retain the shape and have independent Scalar containers. Covers github.com/rakudo/rakudo/issues/3334.
17:38
nqp-configure: patrickbkr++ created pull request #16:
Rename `--git-reference` to `--git-cache-dir`
17:42
nqp-configure: 03b3757066 | (Patrick Böker)++ | lib/NQP/Config.pm
Rename `--git-reference` to `--git-cache-dir`

This more accurately reflects what it does as the option automatically fills the cache dir and keeps it up to date.
17:44
nqp-configure: f6b9606d91 | (Patrick Böker)++ (committed using GitHub Web editor) | lib/NQP/Config.pm
Merge pull request #16 from patrickbkr/git-cache-dir-option

Rename `--git-reference` to `--git-cache-dir`
17:48 sena_kun joined
Geth rakudo: 3a86272ed5 | (Jonathan Worthington)++ | src/Perl6/bootstrap.c/BOOTSTRAP.nqp
Ensure Foo(...) coercers make Raku objects

Str($*USER) was producing a BOOTStr because we return an unboxed `str` in `IntStr.Str` (which is fine), and since the caller was NQP code it then ended up boxed as a BOOTStr. Fixes #3308.
17:58
linkable6 RAKUDO#3308 [closed]: github.com/rakudo/rakudo/issues/3308 [data types] BOOTStr issue
Geth roast: 2eb6c1814b | (Jonathan Worthington)++ | S02-literals/allomorphic.t
Cover bug where Str($*USER) gave a bad result

Covers github.com/rakudo/rakudo/issues/3308.
17:59
jnthn Well, there's some PRs and issues down... 18:03
sena_kun jnthn++ 18:05
Geth rakudo: 8c995febda | (Patrick Böker)++ | tools/build/binary-release/build-debian.sh
Add a binary release build script that works on Debian

This can / will be used when building for ARM architectures. The decision fell on Debian 9, as it is the distro with the oldest glibc, while being mainstream and still supported on armhf / arm64 architecture.
18:15
rakudo: patrickbkr++ created pull request #3474:
Implement `--git-cache-dir` Configure.pl option
18:29
18:49 camelia left 18:56 vrurg left 18:58 jmerelo joined
jmerelo Hi, I'm trying to compile Rakudo in a Debian docker container, doing this perl Configure.pl --gen-moar --gen-nqp --backends=moar 18:59
It says this: "Can't locate NQP/Config.pm in @INC (you may need to install the NQP::Config module)"
It's likely the README.md in Rakudo is outdated, but I can't figure it out...
patrickb jmerelo: did you put "./" before "Configure.pl"? 19:00
jmerelo patrickb: hum 19:01
patrickb thinking about it this shouldn't make a difference... 19:02
jmerelo patrickb: No, I use directly the line above, so perl Configure.pl gets the file in the current directory
patrickb ah! 19:03
do you have git installed?
that Config.pm is part of the nqp-configure submodule
that has to be cloned first
Configure.pl should do that automatically first thing, but I suspect it somehow didn't on your machine 19:04
jmerelo patrickb: it's a fork of the official Rakudo docker container, it downloads the tarfile directly
patrickb: from here github.com/rakudo/rakudo/archive/$...on}.tar.gz
patrickb: should I try and use the build-debian.sh you just uploaded? 19:05
patrickb is the 3rd-party/nqp-configure folder in the rakudo folder empty?
lizmat hmmm... t/spec/S07-hyperrace/basics.t started failing for me (test 84) 19:06
can anyone confirm ?
jmerelo patrickb: no, it's not
patrickb uh... Then I'm out of clues. Configure.pl line 28 should add the needed includes just fine... 19:08
Just double checking, there is a rakudo/3rdparty/nqp-configure/lib/NQP/Config.pm file? 19:09
jmerelo patrickb: no, that's empty 19:10
patrickb: the tarfile does not include the submodule...
patrickb I think you got the wrong file
jmerelo patrickb: It's this one github.com/rakudo/rakudo/archive/$...on}.tar.gz 19:11
patrickb on github there are "rakudo-2020.01.tar.gz" files. Those are the right ones.
Those are the ones that the releaser actually created.
jmerelo patrickb: OK, I'll try that
patrickb The 2020.01.tar.gz are automatically created archives of the repository. Don't use those. 19:12
Geth nqp: 9084b28c78 | (Elizabeth Mattijsen)++ | tools/templates/MOAR_REVISION
Bump Moar to get --git-cache-dir support

Patrick Böker++
19:13
lizmat patrickb ^^ take it away!
patrickb whoot!
jmerelo patrickb: it's likely that the rakudo-star dockerfiles are busted, then...
patrickb: seems to be working fine now. Thanks!
Geth nqp: e77b938802 | (Patrick Böker)++ | 3 files
Implement `--git-cache-dir` Configure.pl option

This option makes the configure process create / update git repositories for all of its submodules and MoarVM in the given folders and use those as reference repos. The option is forwarded to a potential MoarVM sub-build.
Also remove stale documentation for a not anymore implemented
  `--git-reference` option. I suspect it was lost during the great build
system refactor.
nqp: c46b54e309 | (Patrick Böker)++ | 3rdparty/nqp-configure
Pull in new nqp-configure with support for --git-cache-dir
19:14
nqp: 849971e2ee | (Patrick Böker)++ (committed using GitHub Web editor) | 4 files
Merge pull request #595 from patrickbkr/git-reference-option

Implement `--git-cache-dir` Configure.pl option
patrickb Hm... Given we previously had an option called `--git-reference` and that option broke a year ago or so and has since then just been a noop. 19:16
Should we continue to support it, as to not break tools that pass that option?
lizmat if it was a noop, I guess it should stay a noop for the foreseeable future, with maybe a warning ? 19:17
patrickb OK, then I just broke NQP in exactly that manner. Fix coming up... 19:18
19:29 Xliff joined
Xliff o/ 19:29
Is it TEST_JOBS or NUM_JOBS for parallel spectest?
19:31 vrurg joined
lizmat TEST_JOBS 19:31
Xliff vrurg - \o
lizmat++
I am actually paying for a very large image on AWS.
Just for shiz and giggles.
vrurg Xliff: o/ :) 19:32
19:33 sena_kun left
Xliff vrurg: How are things? 19:34
Geth nqp: patrickbkr++ created pull request #598:
Don't fail with `--git-reference` Configure.pl parameter
19:35
vrurg Xliff: all the same, stuck in Vikna devel... 19:36
Xliff :)
vrurg Xliff: thinking of the clashing module problem time after time, but no time for it.
Xliff Refactoring all of my p6-GLib projects for consistency, here...
?!? 19:37
vrurg Xliff: 👍🏻
Xliff Inline::Perl5 tests are failing.
patrickb vrurg: Nice to see you!
vrurg: Mind to have a very short look at the above PR?
19:39 jmerelo left
vrurg patrickb: I'd move the warning into 'configure_from_options'. 19:39
Just to keep Configure.pl as uncluttered as only possible. 19:40
patrickb vrurg: In the nqp-configure sub-module or just the tools/lib/NQP/Config/NQP.pm? 19:41
Xliff Now they pass with --serial. Hmmm...
vrurg patrickb: if you do the same for rakudo (and I presume you're going to), then it makes sense to add it to nqp-configure. 19:42
patrickb OK
Xliff Hmmm.... my spectest run wen 109857 tests in 48 seconds and is now hanging. 19:45
Interesting.
Hung on S15-normalization/nfc-concat.t, although S32-io/lock.t was right before it. 19:46
19:46 sena_kun joined
Geth nqp-configure: patrickbkr++ created pull request #17:
Warn about no-op --git-reference option
19:47
nqp-configure: 403f0a6742 | (Patrick Böker)++ | lib/NQP/Config.pm
Warn about no-op --git-reference option.
19:48
nqp-configure: 326895a2e4 | (Patrick Böker)++ (committed using GitHub Web editor) | lib/NQP/Config.pm
Merge pull request #17 from patrickbkr/git-reference-back-compat

Warn about no-op --git-reference option
nqp: bc94cc7052 | (Patrick Böker)++ | 2 files
Don't fail with `--git-reference` Configure.pl parameter

This makes it backwards compatible. The option has been a no-op for quite some time, so we don't need to actually implement it.
19:49
patrickb And with that NQP should be good again even when --git-reference is passed again.
vrurg patrickb: BTW, it just crossed my mind that you use 'print' for the warning. There is method 'note' on NQP::Configure for this: $self->note("WARNING!", "line1\n", "line2\n") 19:50
patrickb: non-critical, obviously. :)
[Tux] Rakudo version 2020.01-136-g8c995febd - MoarVM version 2020.01.1-35-gecddb4de5
csv-ip5xs0.706 - 0.725
csv-ip5xs-205.855 - 5.970
csv-parser22.992 - 24.852
csv-test-xs-200.360 - 0.398
test7.465 - 7.765
test-t1.863 - 1.899
test-t --race0.874 - 0.946
test-t-2029.754 - 31.337
test-t-20 --race8.847 - 9.273
19:52
lizmat m: sub a(Cool:D $a) { $a.Str }; a(42) for ^10000000; say now - INIT now 20:07
ah, looks like we lost camelia :-( 20:08
p6: sub a(Cool:D $a) { $a.Str }; a(42) for ^10000000; say now - INIT now
and the fallback as well :-(
r 'sub a(Cool:D $a) { $a.Str }; a("42") for ^10000000; say now - INIT now' 20:09
0.318394
r 'sub a(Cool:D $a) { $a.Str }; a(42) for ^10000000; say now - INIT now'
0.29575214
this feels counter-intuitive
coercing a Str to a Str is slower than coercing an Int to a Str ?
shorter version: 20:21
$ r '"42".Str for ^100000000; say now - INIT now'
0.8582568
$ r '42.Str for ^100000000; say now - INIT now'
0.69737759
jnthn: ^^ any ideas ?
Geth rakudo/master: 4 commits pushed by (Patrick Böker)++ 20:30
patrickb rba: Ping. I have a bit time left today. So throw at me whatever you have to do. 20:50
rba Patrickb: through 🐪 20:58
patrickb: give me a sec 20:59
21:11 Kaiepi left 21:12 Kaiepi joined 21:27 rba_ joined
rba patrickb: Are you still around for a discussion? 21:29
patrickb rba: I'm here!
rba patrickb: Yay! 21:30
21:33 sena_kun left
rba patrickb: Do you read PM too? 21:43
patrickb Uhm... dunno. Might have to consult my IRC client where it shows them... I guess you sent one... 21:44
rba patrickb: 🤨 21:46
patrickb: like to share a tmate session with you. yet not public is the wrong place. 21:47
patrickb: In PM you are away. Strange.
patrickb grrr....
21:48 sena_kun joined 22:00 MasterDuke joined
MasterDuke m: my $a; my $b = 99.rand.Int; say $b; my $s = now; $a = $b.Str for ^10_000_000; say now - $s; say $a 22:03
oh right, camelia is down 22:04
Geth rakudo: 597288be62 | (Elizabeth Mattijsen)++ | src/core.c/Str.pm6
Support :i(ignorecase) / :m(ignoremark) with Str.index

  - on MoarVM, uses nqp::indexic / indexim / indexicim ops
  - other backends use foldcase logic for :i(ignorecase)
  - dies on other backends trying to use :m(ignoremark)
lizmat who runs camelia again ?
MasterDuke lizmat: anyway, the above example takes ~3.3s for me, but $b = 99.rand.Int.Str only takes ~1.9s 22:05
moritz or nine i think?
oh, and my numbers are for ^100_000_000
lizmat hmmm... 22:06
22:28 Kaiepi left 22:30 Kaeipi joined
jnthn lizmat: That's odd....please make an issue, otherwise I'll probably forget to investigate :) 22:34
sena_kun is it annoying when I am tagging everything as blockers these days? :) 22:36
lizmat sena_kun: no 22:37
:-)
Geth rakudo: Altai-man assigned to jnthn Issue Likely memory corruption in .hyper github.com/rakudo/rakudo/issues/3475
49aaa6ff36 | (Elizabeth Mattijsen)++ | src/core.c/Str.pm6

  - on MoarVM, uses nqp::indexic / indexim / indexicim ops
  - other backends use foldcase logic for :i(ignorecase)
  - dies on other backends trying to use :m(ignoremark)
  - refactored private die method for more general usage
22:38
Xliff Hmmm... I have spun up several VMs at both AWS and Google and can't get the performance increases I am seeing on my local machine with the .precomp lock removed. :/ 22:42
Of course, I am running a higher number of threads.
22:46 rba_ left
23:02 lucasb left
Geth rakudo: 60b0d86a11 | (Elizabeth Mattijsen)++ | src/core.c/Str.pm6
Support :i(ignorecase) / :m(ignoremark) with Str.ends-with

  - on MoarVM, uses nqp::indexic / indexim / indexicim ops
  - other backends use foldcase logic for :i(ignorecase)
  - dies on other backends trying to use :m(ignoremark)
23:15
rakudo: 7a2d68f827 | (Elizabeth Mattijsen)++ | src/core.c/Date.pm6
Fix timezone dependency

Earlier optimizations caused DateTime.new(Instant) and DateTime.today to not take the current timezone into account. This caused tests to fail only between midnight and 1am. Adding $*TZ into the mix was enough to have the tests pass all the time.
23:19
lizmat and that concludes my hacking for today 23:20
sleep&
23:34 sena_kun left 23:47 sena_kun joined 23:59 Kaeipi left