00:04 pamplemousse joined 00:22 hungrydonkey joined 00:42 finsternis left
vrurg timotimo: around? 01:00
timotimo asquare 01:01
vrurg I was one complaining about not having access to plugin marketplace in CT. I think I found what's wrong. 01:02
Here is the request URL from CT: plugins.jetbrains.com/api/search?o...&max=9 – the server returns empty array on this 01:03
timotimo: most likely it's the build parameter which is the cause. 01:04
timotimo sorry, what is CT? 01:05
vrurg timotimo: Comma CT :) 01:06
timotimo is that CommuniTy?
or CompleTe?
vrurg Community for now. 01:07
timotimo ah i see it mentioned
sorry, i know next to nothing about jetbrains' plugin repository thing 01:08
vrurg You participate in Comma development or am I mistaken?
timotimo you are correct 01:12
i do not, however, do releases
so unfortunately i have no idea how to help you :(
but the others read here, too
vrurg At least I have where to point at if I lost the logs where I got the URL. :) 01:13
anyway, thanks! 01:14
02:01 hungryd14 joined 02:02 hungrydonkey left 02:06 hungryd14 left 02:10 hungrydonkey joined 02:26 hungryd28 joined 02:30 hungrydonkey left 02:41 hungryd28 left 02:42 hungrydonkey joined 03:12 pamplemousse left 03:37 moon-child joined, hungrydonkey left, hungrydonkey joined 04:48 hungrydonkey left 04:58 hungrydonkey joined 05:12 hungrydonkey left 05:19 hungrydonkey joined 05:26 hungrydonkey left 05:31 hungrydonkey joined 05:32 hungrydonkey left 06:05 hungrydonkey joined 06:07 hungrydonkey left 08:34 sena_kun joined
lizmat Files=1306, Tests=111242, 214 wallclock secs (28.43 usr 8.39 sys + 3002.33 cusr 280.96 csys = 3320.11 CPU) 08:50
08:56 Altai-man_ joined 08:59 sena_kun left
Geth rakudo: 31ffc2bd2d | (Stefan Seifert)++ | 2 files
Make timing based performance tests opt in via RAKUDO_RUN_TIMING_TESTS

It's easier to tell a hand full of core developers to set an environment variable to opt into performance regression tests than to tell an unknown number of unknown people to opt out of them for automated builds.
Fixes #3065
linkable6 RAKUDO#3065 [closed]: github.com/rakudo/rakudo/issues/3065 [intermittent][severe] Flopping test: 99-misc.t (2019.07)
nine Now what do we do about roast? 10:36
10:57 sena_kun joined 10:59 Altai-man_ left
lizmat marking them todo I guess? 10:59
sena_kun nine++ 11:04
Geth rakudo: 8dcfa849f6 | (Ben Davies)++ | src/Perl6/Metamodel/ParametricRoleGroupHOW.nqp
Add a parents metamethod to Metamodel::ParametricRoleGroupHOW
rakudo: 033b23bd82 | (Elizabeth Mattijsen)++ (committed using GitHub Web editor) | src/Perl6/Metamodel/ParametricRoleGroupHOW.nqp
Merge pull request #3649 from Kaiepi/role-group-parents

Add a parents metamethod to Metamodel::ParametricRoleGroupHOW
nine lizmat: I meant what do we do about timing based tests in roast? There's ROAST_TIMING_SCALE in use for some of them, but that makes them equally opt-out instead of opt-in.
lizmat ah, I guess we make them all opt-in
nine That would get us one step further on running them on CI 11:22
lizmat nine: so all test files with ROAST_TIMING_SCALE should only run their test if what env is set ? 11:26
11:27 ggoebel left
nine Originally I thought, we'd use a general RAKU_RUN_TIMING_TESTS variable. But then I realized that this might tempt module authors to use that variable as well and people opting in might opt into more than they anticipated 11:32
11:51 Voldenet left 11:54 Voldenet joined, Voldenet left, Voldenet joined
lizmat m: my @a; @a>>.map: * + 2 11:55
camelia elems requires a concrete object (got a Mu type object instead)
in block <unit> at <tmp> line 1
lizmat m: say () =:= () # why isn't this True ? 12:48
camelia False
Geth rakudo: 9b503292e5 | (Elizabeth Mattijsen)++ | src/core.c/metaops.pm6
Minimal change to make my @a>>.foo not die

This will probably see some refactoring by yours truly in the near future anyway.
12:56 Altai-man_ joined 12:59 sena_kun left
Altai-man_ Good news are: no serious changes for Blin (gist.github.com/Altai-man/8f77b711...2696eab80) for revert branch. 13:02
I'm now checking two modules if these are stable regressions or not. 13:03
Can anyone please confirm the branch has performance back?
lizmat Altai-man_: testing now 13:08
timings for my machine: 1.786 -> 1.211 , --race: .833 -> .662 13:11
so yes, I'd say that gets our performance back
Altai-man_ lizmat++ # thank you! 13:12
Altai-man_ tests LibXML and Lumberjack modules
Altai-man_ .oO ( so maybe we can even release today )
seeing a stresstest failure in t/spec/S02-types/version.t 13:16
lizmat Altai-man_: I see S06-advanced/dispatching.t failing consistently
"No such method 'is_wrapped' for invocant of type 'Method'" 13:17
Altai-man_ lizmat, you need a roast with PR where this test is skipped.
lizmat, it is still WIP github.com/Raku/roast/pull/638
lizmat ah, ok, then all is otherwise well!
Altai-man_ >t/spec/S02-types/version.t (Wstat: 256 Tests: 44 Failed: 0)
>Parse errors: Bad plan. You planned 45 tests but ran 44.
lizmat works ok for me 13:18
Altai-man_ I don't like it. The test checks something which should be stable. I am having an impression 1)dispatcher revert made something fragile/racey; 2)it was there always I just didn't notice them. 13:19
lizmat can't get it to fail even running 30times or so 13:22
Altai-man_ I'm running stresstest with TEST_JOBS=24.
And in around 10 times I did yesterday and started to run them today, it failed once. 13:23
I'll probably check files manually as was planned.
lizmat also runs a stresstest
Geth roast/dispatcher-revert: 11791284bb | Altai-man++ | S06-advanced/dispatching.t
Mark dispatcher tests as todo
lizmat Altai-man_: t/spec/S17-procasync/kill.t flapped for me 13:29
Altai-man_ lizmat, this one is known
lizmat then again, all clear from this side :-)
Altai-man_ submitted a ticket to Lumberjack... 13:32
Geth rakudo: f115a8a556 | (Elizabeth Mattijsen)++ | src/core.c/Code.pm6
Revert 360232a54fadc83efdafc29dfe808677bcde2865

By popular demand.
14:07 ggoebel joined
Altai-man_ Ok, revert is correct. 14:31
releasable6, status 14:32
releasable6 Altai-man_, Next release will happen when it's ready. 1 blocker. 166 out of 377 commits logged (⚠ 3 warnings)
Altai-man_, Details: gist.github.com/2d18882b56662aa4b2...e06d8f7b43
Altai-man_ So what we have for Rakudo 2020.05 release: 1)1 blocker is remaining to be decided/fixed; 2)We're also waiting for an explicit vrurg's approval; 2)Blin is nice, spects are nice, revert looks ok to me; 4)The commits to log are very "interesting" due to how many re-reverts we have, so I'll take it on me, so please do any logging only if you do really fancy it. Thanks. :) 14:45
14:57 sena_kun joined, travis-ci joined
travis-ci Rakudo build failed. Elizabeth Mattijsen 'Revert 360232a54fadc83efdafc29dfe808677bcde2865 14:57
travis-ci.org/rakudo/rakudo/builds/682304528 github.com/rakudo/rakudo/compare/9...15a8a556cf
14:57 travis-ci left 14:58 Altai-man_ left
AlexDaniel sena_kun: can you file a blin ticket? 15:46
sena_kun: so that I can later go through the weird ones that fail for no good reason
sena_kun AlexDaniel, sure, a bit later~
AlexDaniel CSS::Grammar needs a PR for sure 15:47
sena_kun AlexDaniel, you mean the fact the issue was fixed on master long time ago and Blin somehow tests older version? 15:49
AlexDaniel oooh
yeah why the hell does that happen x) 15:51
I know for sure that if you bump the version then it'll end up fetching it
so what's going on there? Zef sees that the locally available version is identical to the one in the module list and doesn't attempt to fetch it? 15:52
15:54 ggoebel left, ggoebel joined
timotimo lizmat: it's probably not a very useful optimization opportunity, but you know how the dwimmy operator can repeat the last generated value if it gets a Whatever? 15:55
in some cases, really mostly just reified iterator, we can check 1) does the array in question have a typecheck that prevents Whatever from showing up? and 2) can we just restart iterating from the beginning without trouble and don't have to buffer iterator results? 15:56
16:10 AlexDani` joined 16:12 AlexDaniel left
dogbert11 m: my $a = abs(-4711) for ^10000000; say now -INIT now 16:30
camelia 0.212389
dogbert11 m: my $a = (-4711).abs for ^10000000; say now -INIT now
camelia 1.66452763
dogbert11 what's up with this?
quite a big speed difference no? 16:31
what say lizmat? 16:32
sena_kun AlexDani`, github.com/Raku/Blin/issues/24 <- 16:49
16:49 AlexDani` is now known as AlexDaniel, AlexDaniel left, AlexDaniel joined 16:56 Altai-man_ joined
lizmat dogbert11: abs(constant) is optimized to a constant fetch by the static optimizer 16:57
(constant).abs is not
16:58 sena_kun left 17:15 ws7ws joined
AlexDaniel CyclicDependency section here is interesting: github.com/Raku/Blin/issues/24 17:17
some LHFs
half of them seem to be modules that list themselves in build-depends 17:18
Altai-man_ AlexDaniel, not even themselves, but particular files inside a distibution. Zef can install a module by something from a distribution as well as actual distribution name. 17:21
AlexDaniel but hmm… is it meant to be this way? 17:22
Altai-man_ AlexDaniel, what exactly? 17:23
The cycle detection or installation of a distribution by name of its module? 17:24
AlexDaniel I mean, how does zef handle it?
Altai-man_ AlexDaniel, no idea, but IIRC ugexe has said it to me, e.g. Cro::HTTP can be installed by zef install Cro::HTTP::Client, and I never saw him being wrong / having unclear reasoning in anything in this field. 17:27
timotimo i wonder if we can make constructing our nodes a little bit faster 17:28
AlexDaniel yeah, that's relatively easy, but if Cro::HTTP had a build-depends on Cro::HTTP::Client, then what?
timotimo because the constructor relies on calling methods by name 17:29
but it could be that there is almost no actual cost to this 17:30
AlexDaniel m: my @a = 4,8,15,16; @a[1]:delete; .say for @a 17:32
camelia 4
AlexDaniel m: my @a = 4,8,15,16; @a[1]:delete; .say for @a.reverse
camelia 16
AlexDaniel 6c: my @a = 4,8,15,16; @a[1]:delete; .say for @a.reverse
committable6 AlexDaniel, gist.github.com/5dc9a5bd35e570858f...922a51e024
AlexDaniel bisect: my @a = 4,8,15,16; @a[1]:delete; .say for @a; .say for @a.reverse 17:33
bisectable6 AlexDaniel, Bisecting by output (old=2015.12 new=f115a8a) because on both starting points the exit code is 0
AlexDaniel 6c: my @a = 4,8,15,16; @a[1]:delete; .say for @a; .say for @a.reverse
committable6 AlexDaniel, gist.github.com/c6d842ae060d99f2fc...cf6054ab9e
bisectable6 AlexDaniel, bisect log: gist.github.com/37c9fa24d518a5e7dd...06336596ad
AlexDaniel, (2017-03-01) github.com/rakudo/rakudo/commit/be...d919c695d0
AlexDaniel 6c: my @a = 4,8,15,16; @a[1]:delete; .say for @a; say ‘----’; .say for @a.reverse 17:34
committable6 AlexDaniel, gist.github.com/ae0c6598a5bd214e77...3b25aed44f 17:35
dogbert11 lizmat: thx for the explanation, I get a bit confused when the speed of methods vs subs differ too much 17:36
AlexDaniel dogbert11: yes but subs are not always faster 17:37
sometimes it's other way around
dogbert11 AlexDaniel: I have noticed that :) 17:41
timotimo i see that ival, sval, and nval already pull the value argument out directly to construct themselves faster 17:47
and op and var and such
17:54 ws7ws left
timotimo isn't it superfluous to bind $!flags to 0 every time we create a QAST node? they should be nulled out on creation already, i would assume? 17:57
commented them all out, i guess i'll just put up a pull request for discussion purposes 18:04
the bindattr should be turned into a super cheap mov assembly instruction anyway, but *shrug* 18:05
i guess leaving the instruction out lets the moarvm file for QASTNodes be a little bit smaller
not like a kilobyte smaller
AlexDaniel lizmat: hah: 18:09
m: my @a = 4,8,15,16; @a[*-Inf]:delete; say @a
camelia [4 8 15 16]
AlexDaniel m: my @a = 4,8,15,16; say @a[*-Inf]:delete; say @a
camelia Can not remove elements from a Array
in block <unit> at <tmp> line 1
18:12 ggoebel left
timotimo 556 bytes vs 514 bytes "before" in just one example (QAST::Var.new) 18:13
18:13 ggoebel joined
timotimo 216 vs 196 frame size after spesh, 1 fewer spesh slot, and jitted machine code went from 1259 to 1179 bytes 18:15
m: say "qastnodes.moarvm changes size by $(98952 - 99496) bytes" 18:31
camelia qastnodes.moarvm changes size by -544 bytes
Geth nqp/qast_constructor_no_zero_init: 5cf4a289f6 | (Timo Paulssen)++ | 18 files
Don't bind an int attribute to 0 directly after construction

since we construct objects null-initialized anyway.
timotimo m: say "qastnodes.moarvm changes size by $(98952 * 100 / 99496) percent" 18:41
camelia qastnodes.moarvm changes size by 99.453244 percent
timotimo m: say "qastnodes.moarvm changes size by $(1 - (98952 * 100 / 99496)) percent"
camelia qastnodes.moarvm changes size by -98.453244 percent
timotimo m: say "qastnodes.moarvm changes size by $(100 - (98952 * 100 / 99496)) percent"
camelia qastnodes.moarvm changes size by 0.546756 percent
Geth nqp: timo++ created pull request #624:
Don't bind an int attribute to 0 directly after construction
timotimo i didn't run a stresstest with this just yet 18:47
18:57 sena_kun joined 18:59 Altai-man_ left 19:27 pamplemousse joined
lizmat bisectable6: my @a = ^5; @a[2]:delete; dd @a.reverse 20:15
bisectable6 lizmat, Bisecting by output (old=2015.12 new=f115a8a) because on both starting points the exit code is 0
lizmat, bisect log: gist.github.com/9e86fcd8176d6e1ead...1b3a65e8b2
lizmat, (2017-03-01) github.com/rakudo/rakudo/commit/be...d919c695d0
lizmat bisectable6: old=2020.02.1 my @a = ^5; @a[2]:delete; dd @a.reverse 20:16
bisectable6 lizmat, On both starting points (old=2020.02.1 new=f115a8a) the exit code is 0 and the output is identical as well
lizmat, Output on both points: «(4, 3, Mu, 1, 0).Seq␤»
lizmat bisectable6: old=2018.01 my @a = ^5; @a[2]:delete; dd @a.reverse
bisectable6 lizmat, On both starting points (old=2018.01 new=f115a8a) the exit code is 0 and the output is identical as well
lizmat, Output on both points: «(4, 3, Mu, 1, 0).Seq␤»
sena_kun releasable6, status 20:30
releasable6 sena_kun, Next release will happen when it's ready. 1 blocker. 181 out of 377 commits logged (⚠ 2 warnings)
sena_kun, Details: gist.github.com/f8e5f161870f28cb58...1ee4016a3d
20:35 robertle joined 20:42 ggoebel left 20:44 sena_kun left
lizmat m: my @a = ^5; @a[2]:delete; my @b is default(42) = @a.reverse; dd @b 20:45
camelia Array @b = [4, 3, Mu, 1, 0]
Geth rakudo: e9468d6a02 | (Elizabeth Mattijsen)++ | src/core.c/Rakudo/Iterator.pm6
Make List.reverse return Nil for holes

This should be slightly better than Mu, because then it will actually honour whatever you're putting the reverse into. This change will make this actually work:
   my @a = ^5;
   my @b is default(42) = @a.reverse;
   dd @b[2]; # 42, not Mu
20:53 robertle left 21:02 hankache joined 21:03 hankache left 21:04 hankache joined 21:05 hankache left, hankache joined 21:15 hankache left 21:21 ggoebel joined 21:33 cognomin_ joined 21:37 cognominal left
Geth rakudo: ca49c153db | (Elizabeth Mattijsen)++ | 4 files
Make sure @a.reverse creates proper containers for holes

  - rename R:I.ReifiedListReverse to R:I.ReifiedReverse
   as it's now also for Arrays specifically
  - add support for descriptors
  - expand nqp::null handling to creating container when needed
21:58 pamplemousse left
Geth roast: d0452c4f14 | (Elizabeth Mattijsen)++ | S32-list/reverse.t
Add tests for R#3654
linkable6 R#3654 [open]: github.com/rakudo/rakudo/issues/3654 [tests needed] .reverse on arrays with holes (Mu)
Geth rakudo/rakudo_3257: 16 commits pushed by (Vadim Belman)++
review: github.com/rakudo/rakudo/compare/6...7217acb76f
rakudo/rakudo_2993: f2c1a449aa | (Vadim Belman)++ | 4 files
Remove release and codename properties from Compiler

They're not relevant and are not used.
linkable6 RAKUDO#2993 [open]: github.com/rakudo/rakudo/issues/2993 [consensus needed] Are compiler.release-number and compiler.codename set?
Geth rakudo/rakudo_1202: f05bf6acb9 | (Vadim Belman)++ | 3 files
Prevent rebuild of nqp without --gen-nqp.
rakudo/attribute-package-fix: ee2f7a7022 | (Vadim Belman)++ | src/Perl6/Metamodel/ParametricRoleHOW.nqp
Alway clone attribute when specializing

Generic ones are being cloned upon instantiation. To maintain consistency and guarantee that concretization always using a copy of the original attribute, do cloning of non-generic attributes.
Provides support for fixing `.package` not being set for attributes on roles.
rakudo/attribute-package-fix: e719c7e6ee | (Vadim Belman)++ | 2 files
Fix a problem when attribute object doesn't have $!package set for roles

The fix is rather simple: insted of setting `$!package` to generic
  `$?CLASS` set it directly to $*PACKAGE as it contains the role/class
being currently compiled and which actually owns the attribute. This approach requries additional fixup later, when the attribute is getting actually composed into a class. At this point `$!package` must be set to the class typeobject. The previous commit guarantees that the action is performed upon a new copy of the attribute object and therefor won't mess up with other type object consuming the same role.
rakudo/nmake-detect-fail: e52754ca1d | (Vadim Belman)++ | 3rdparty/nqp-configure
nqp-configure submodule: switch to nmake-detect-fail branch
rakudo/nmake-detect-fail: 620ae0ecc3 | (Vadim Belman)++ | tools/lib/NQP/Config/Rakudo.pm
Remove overlooked debug print.
rakudo/nmake-detect-fail: dddb88c4bd | (Vadim Belman)++ | 3rdparty/nqp-configure
nqp-configure submodule update
rakudo/rakudo_1203_bad: 10 commits pushed by (Vadim Belman)++, (Paweł Murias)++, (Aleks-Daniel Jakimenko-Aleksejev)++
review: github.com/rakudo/rakudo/compare/5...a40eb9d46b
rakudo/issue_2250: 13 commits pushed by (Vadim Belman)++
review: github.com/rakudo/rakudo/compare/b...cdb32c4d2e
rakudo/macos-toolchain: 9cba5d7474 | (Vadim Belman)++ | Configure.pl
Follow up to PR MoarVM/MoarVM#1067

Make rakudo Configure.pl use make from XCode toolchain if the default compliler belong to it too.
linkable6 MOARVM#1067 [closed]: github.com/MoarVM/MoarVM/pull/1067 Improve check for used toolchain on macOS
rakudo/rakudo_3075: 035bd3e629 | (Vadim Belman)++ | 3 files
Fix Stash'es clashing after deserialization

When two modules are precompiled independently but then loaded by the same process it is possible that if they share common name prefix like in, say, `A::B::C::D1` and `A::B::C::D2` then the module loaded second would get its Stash on `A` overriden by the module loaded first. This would effectively result in some symbols in `A::B::...` hierarchy to be ... (21 more lines)
rakudo/rakudo_3075: ea67ef4fca | (Vadim Belman)++ | src/Perl6/Actions.nqp
Don't waste CPU cycles unless we're precompiling

There is no need for snapshotting a Stash unless current compunit is being precompiled.
rakudo/rakudo_3075: 8c0bad7ade | (Vadim Belman)++ | 3 files
[WIP] Intermediate experimental commit

Preserving the current state of the things. Need to locate a problem with macOS version failing due to SIGABRT and not allowing me to debug.
  - Don't record CORE stashes.
  - Added SC handle parameter to TAKE-SNAPHOT. Perhaps it makes sense of
   recording only symbols from current SC.
rakudo/issue_2169: 01c5caac58 | (Vadim Belman)++ | src/Perl6/Metamodel/BOOTSTRAP.nqp
Fix for rakudo/rakudo#2169

ContainerDescriptor $!default gets instantiated alongside with $!of if necessary.
Possible problem: if instantiation will be issued based on ContainerDescriptor.is_generic() return then in cases when $!of is a non-generic value but the default is set to a generic then default won't be instantiated. At this moment such situation is not possible because rakudo dies when `is default` parameter is a generic.
linkable6 RAKUDO#2169 [closed]: github.com/rakudo/rakudo/issues/2169 [tests needed][type captures] Assigning Nil to a scalar declared with captured type changes it's container metaclass to GenericHOW
rakudo/issue_2169: 6b57b73918 | (Vadim Belman)++ | 0 files
Merge branch 'issue_2169' of github.com:vrurg/rakudo into issue_2169
rakudo/Map-keeping-containers: 6a6532cc33 | (Vadim Belman)++ | 10 files
Merge branch 'master' of github.com/rakudo/rakudo into Map-keeping-containers
rakudo/will-build-typecheck: d83060b029 | (Vadim Belman)++ | src/core.c/core_epilogue.pm6
Add a couple of symbols to HLL state

  ... as required by create_BUILDPLAN typechecks
rakudo/will-build-typecheck: 35b425b13e | (Vadim Belman)++ | src/Perl6/Metamodel/BUILDPLAN.nqp
Get required symbols from HLL when no $*W available

Or `$*W` is available but can't provide the service to us. This would allow `create_BUILDPLAN` to typecheck defaults when a type is being created at run-time.
rakudo/will-build-typecheck: 69d8b41e64 | (Vadim Belman)++ | 20 files
Merge branch 'master' into will-build-typecheck
rakudo/problem-solving-112: 44 commits pushed by (Vadim Belman)++
review: github.com/rakudo/rakudo/compare/1...85fcde13b4
rakudo/rakudo_3132: 6276ba7e39 | (Vadim Belman)++ | tools/templates/moar/Makefile.in
Merge branch 'master' of github.com/rakudo/rakudo into rakudo_3132
rakudo/eval-with-unit: 9d79cbac49 | (Vadim Belman)++ | src/core/ForeignCode.pm6
Make EVAL thread-safe

Currently – by wrapping code compilation into a lock. This is due to
  'Decoder may not be used concurrently' error thrown by MoarVM more than
one compilation is being run in parallel.
rakudo/eval-with-unit: e827ae49bf | (Vadim Belman)++ | tools/lib/NQP/Config/Rakudo.pm
Merge branch 'master' of github.com/rakudo/rakudo into eval-with-unit
rakudo/eval-with-unit: 6000a0f5de | (Vadim Belman)++ | 2 files
Trying to get EVAL :unit to work.
rakudo/problem-solving-99-bad: 14 commits pushed by (Vadim Belman)++
review: github.com/rakudo/rakudo/compare/5...30524b807d
rakudo/rakudo_issue_3107_old: d46ecf15ed | (Vadim Belman)++ | 3 files
Merge branch 'master' of github.com/rakudo/rakudo into rakudo_issue_3107
rakudo/rakudo_issue_3107_old: 4feb77e806 | (Vadim Belman)++ | 3 files
Make every CORE compile with its respective language revision

This is a preparation work to assign language version to core classes. I.e. PseudoStash.^ver would report back either `6.c` or `6.e` depending from which CORE the class was imported.
rakudo/rakudo_2979: 4e27af6036 | (Vadim Belman)++ | 2 files
Do correct assignment to SetHash

Fix for rakudo/rakudo#1203
linkable6 RAKUDO#1203 [closed]: github.com/rakudo/rakudo/issues/1203 [data types] Would be nice to have an easy way to work with SetHashes and its kin
Geth rakudo/rakudo_3075_experimental: 6 commits pushed by (Vadim Belman)++
rakudo/build-revamp: a65279e9e4 | (Vadim Belman)++ | 3rdparty/nqp-configure
nqp-configure submodule update
rakudo/build-revamp: 1e908f56d4 | (Vadim Belman)++ | 2 files
Merge branch 'master' of github.com/rakudo/rakudo into build-revamp
rakudo/fix-DESTROY-order: 1a2dd0d9d2 | (Vadim Belman)++ | 5 files
Merge branch 'master' into fix-DESTROY-order
rakudo/ZhongnianTao-master: 196903c7ba | (Vadim Belman)++ | 2 files
Merge branch 'master' of github.com/ZhongnianTao/rakudo into ZhongnianTao-master
rakudo/problem-solving-74: 72e10e7970 | (Vadim Belman)++ | 4 files
Better support for EXPORT with Stash

Turn off containerization for Stash
Also added operator (<EXPR>):: which is a shortcut for Stash.new(<EXPR>)
Now during import values from Stash are used as-is without deconting.
rakudo/problem-solving-74: 97e022e754 | (Vadim Belman)++ | src/core/Rakudo/Internals.pm6
Merge branch 'master' of github.com/rakudo/rakudo into problem-solving-74
rakudo/problem-solving-74: 11f486777b | (Vadim Belman)++ | 3 files
Merge branch 'master' of github.com/rakudo/rakudo into problem-solving-74
vrurg One damn wrong completion... :( 22:35
22:49 cognominal joined
vrurg Sorry for the flood. All accidentally pushed branches are wiped out. 22:50
22:52 cognomin_ left
Geth roast: 11791284bb | Altai-man++ | S06-advanced/dispatching.t
Mark dispatcher tests as todo
roast: a6d0f04570 | (Vadim Belman)++ (committed using GitHub Web editor) | S06-advanced/dispatching.t
Merge pull request #638 from Raku/dispatcher-revert

Mark dispatcher tests with TODO