00:02 lucasb left
MasterDuke m: sub a($a where 1|2|3|4|5|6|7|8|9|0) { $a }; my $b; my $c = 10.rand.Int * 0; my $s = now; $b = a($c) for ^100_000; say now - $s; say $b 00:34
camelia 1.3455308
0
MasterDuke locally that takes ~0.8s on stock rakudo. i now have a branch where it takes 0.06s 00:36
still needs a little cleanup, but when i'm done i think it should support mixing Types and literal ints, nums, and strs 00:37
but it'll have to wait until tomorrow to finish... 00:38
Xliff Ooh, nice! 00:39
Takes around ~0.9, here
MasterDuke: What did you do to speed it up?> 00:53
IIUC (if I understand correctly) you could then do something like "where 0|1|MySubsetOfInt" and it would still be faster than the current implementation? 00:54
01:05 Kaiepi left 01:06 Kaiepi joined 01:14 hungrydonkey joined 01:33 sena_kun joined 01:35 Altai-man_ left 01:44 hungrydonkey left 02:20 Kaiepi left, Kaiepi joined 02:21 squashable6 left 02:23 squashable6 joined
[Coke] bumps the Proc::Async issue at github.com/rakudo/rakudo/issues/3057 03:22
03:32 Altai-man_ joined 03:34 sena_kun left 04:54 hungrydonkey joined 05:33 sena_kun joined 05:35 Altai-man_ left 07:32 Altai-man_ joined 07:34 sena_kun left
lizmat Files=1302, Tests=111138, 226 wallclock secs (27.72 usr 8.10 sys + 2979.20 cusr 284.99 csys = 3300.01 CPU) 09:04
ok, I wonder what lost us so much in the spectest :-(
09:08 robertle left
tyil anyone around who may know better what gitlab.com/tyil/docker-perl6/-/job...65702#L402 is about? 09:14
seems to only break on alpine 09:15
09:27 ufobat joined 09:33 sena_kun joined 09:35 Altai-man_ left
nine tyil: maybe related to this? github.com/MoarVM/MoarVM/pull/1240 09:38
tyil perhaps 09:39
lizmat looks like the spectest slowdown is caused by 4e12365b197e5f765343 , the $?FILE fix 10:02
linkable6 (2020-02-14) github.com/rakudo/rakudo/commit/4e12365b19 Remove extraneous information from $?FILE
lizmat without 5629cdff581cf7b3b2a and 4e12365b197e5f765343 I get a better spectest result on a warm machine than during the night on a cold one 10:03
linkable6 (2020-02-15) github.com/rakudo/rakudo/commit/5629cdff58 Remove extra info from $?FILE from the end
sena_kun sees some blin fallout, investigating and preparing a ticket 10:15
Geth rakudo: Altai-man assigned to lizmat Issue [WIP] Blin 2020.02 github.com/rakudo/rakudo/issues/3494
2b7d5ffeb8 | (Elizabeth Mattijsen)++ | 2 files

This should fix some Blin fallout
10:30
sena_kun hopes blockers will be resolved by tne end of this week to release on Saturday 10:31
sena_kun dayjob &
11:10 pmurias joined
pmurias jnthn: what java package should we migrate to after the rename? 11:11
jnthn: does org.raku.nqp make sense (I'm refactoring the truffle backend to make it work under the newest graalvm and I'm thinking about building it up in the moved package) 11:13
jnthn pmurias: That's probably a decent choice, yes
11:32 Altai-man_ joined 11:34 sena_kun left 11:37 pmurias left 11:38 pmurias joined
lizmat .tell sena_kun will look at the other Blin failures much later today or tomorrow 11:41
afk for most of the rest of the day&
11:46 pmurias left
Altai-man_ lizmat, rest well! 11:50
12:35 travis-ci joined
travis-ci Rakudo build failed. Elizabeth Mattijsen 'Restore *exact* year stringification in Date/DateTime 12:35
travis-ci.org/rakudo/rakudo/builds/651937420 github.com/rakudo/rakudo/compare/1...7d5ffeb8c6
12:35 travis-ci left 12:52 patrickb joined
patrickb tyil: I'd highly suspect the dyncall bump (github.com/MoarVM/MoarVM/commit/65...42a950be). 12:53
12:56 patrickb left 13:02 pmurias joined 13:16 squashable6 left 13:17 squashable6 joined, squashable6 left 13:20 squashable6 joined 13:33 sena_kun joined 13:35 Altai-man_ left 13:36 MasterDuke left 13:55 Xliff left 13:56 lucasb joined, squashable6 left 13:58 squashable6 joined 14:24 dogbert17 joined
releasable6 Next release in ≈4 days and ≈3 hours. 5 blockers. Please log your changes in the ChangeLog: github.com/rakudo/rakudo/wiki/ChangeLog-Draft 15:00
15:32 Altai-man_ joined 15:34 sena_kun left 15:37 hungrydonkey left, hungrydonkey joined 16:07 pmurias left 16:30 hungrydonkey left, hungrydonkey joined 16:42 hungrydonkey left, hungrydonkey joined 16:43 hungrydonkey left, hungrydonkey joined 17:33 sena_kun joined 17:35 Altai-man_ left 17:37 domidumont joined 17:38 domidumont left 17:40 jmerelo joined
jmerelo Made another pull request so that, now that we have a .gitignore in the Github pull down, they call it after Raku github.com/github/gitignore/pull/3312 17:40
japhb Kaiepi: Why is Unix socket support going into IO::Socket::INET, and not for instance IO::Socket::UNIX? 17:58
18:03 hungrydonkey left 18:23 MasterDuke joined 18:43 squashable6 left 18:44 squashable6 joined 19:20 jmerelo left 19:21 Kaiepi left 19:27 Kaiepi joined 19:32 Altai-man_ joined 19:35 sena_kun left
Geth roast: c336b6eaa8 | (Christian Bartolomäus)++ | 3 files
Minor tweaks after renaming of perl6 to rakudo

Follow up on github.com/rakudo/rakudo/commit/0bebe4e23a.
20:14
[Coke] sees $parrot_version in that commit and does a double take 20:17
20:42 AlexDaniel joined 20:43 AlexDaniel left, AlexDaniel joined
vrurg Hm, is there any reason for Attrbiute to have both $!rw and $!ro? 20:50
lizmat vrurg: I have no idea... 20:52
vrurg would have to grep through sources... 20:53
lizmat perhaps the naming of those attributes is subtly confusing
MasterDuke anybody interested in helping me debug my new attempt to also inline concrete things in any junctions for params? github.com/MasterDuke17/rakudo/tre...for_params 20:54
for this example, `sub a($a where 1|2|3|4|5|6|7|8|9|0|"abc") { $a }; say a("abc")`, i get `This type cannot unbox to a native integer: P6opaque, Str`, but `sub a($a where "abc"|1|2|3|4|5|6|7|8|9|0) { $a }; say a("abc")` succeeds 20:55
vrurg lizmat: They're just mutually exclusive. There're set_rw and set_readonly. Each setting corresponding attribute, but must also reset the other one, as I see it. 20:57
I would try getting rid of $!ro as this is the default and see we can get away with $!rw only. 20:58
20:58 pmurias joined
lizmat vrurg: please doublecheck with jnthn,I wouldn't be surprised if this is not a thing we can do 21:00
Geth rakudo: 0286904b45 | (Elizabeth Mattijsen)++ | src/core.c/Date.pm6
Fixed missed condition in move to native attribute

In the move to native attributes in Date, a check on daycount was missed: since nqp::isconcrete() is always true on a native variable, this was giving incorrect results when moving between dates. This is now fixed and unbreaks Date::WorkdayCalendar
21:15
tyil .tell patrickb is it something I can fix or does it need core dev attention? it seems to build fine on the other distros I build for 21:30
21:33 sena_kun joined 21:35 Altai-man_ left
[Coke] looking at Proc::Async::Pipe - the pipes are hanging on to open FDs. Looking at the native-descriptor attribute on the Pipes, they are all Promises. Anyone see where the actual native fd is? 21:47
[Coke] pokes at all the attributes. 21:50
sena_kun lizmat++ # blockers 22:04
sena_kun checks specs and sets up blin
Geth roast: 02680b6c93 | (Christian Bartolomäus)++ | test_summary
Revive --timing option

It looks like this has been broken for a long time.
22:26
rakudo: b74e5d7b39 | (Christian Bartolomäus)++ | .gitignore
Remove duplicate entry from .gitignore
22:28
23:06 pmurias left
vrurg .ask jnthn what is the point of having both $!ro and $!rw on Attribute? So far, I see that $!ro is simply prioritized over $!rw but does it really makes any sense? 23:12
23:32 Altai-man_ joined 23:34 pmurias joined 23:35 sena_kun left
Geth rakudo: vrurg++ created pull request #3496:
Set attribute's rw status when added to a container
23:40
roast/rakudo_3495: b12064dff5 | (Vadim Belman)++ | 2 files
Add tests for `is rw` trait on roles

In support of rakudo/rakudo#3496 and following rakudo/rakudo#3495
23:41
linkable6 RAKUDO#3496 [open]: github.com/rakudo/rakudo/pull/3496 Set attribute's rw status when added to a container
RAKUDO#3495 [open]: github.com/rakudo/rakudo/issues/3495 [attributes][objects][roles] `is rw` trait applied to roles does not work
Geth roast: vrurg++ created pull request #620:
Add tests for `is rw` trait on roles
23:42
Kaiepi japhb, because outside of how their address is created, they're used in the same way as ipv4/ipv6 sockets
jnthn vrurg: I'd forgotten ro even existed there, tbh... 23:43
Kaiepi with a solution to github.com/Raku/problem-solving/issues/111, the extra state related to specific types of addresses would be kept in their address, not in IO::Socket::IENT
s/IENT/INET/
IO::Socket::INET becomes somewhat of a misnomer though 23:45
jnthn vrurg: I think it was maybe meant to be about the scalar being ro, but I'm not really sure.
vrurg: Or maybe it was so if you put `is rw` on the class, you could opt out of it...
23:46 lucasb left
MasterDuke jnthn: do you happen to see anything obviously wrong in github.com/MasterDuke17/rakudo/com...54535548ac ? 23:51
jnthn MasterDuke: eq_i won't handle big integers 23:55
MasterDuke hm, yeah 23:56
jnthn And it'll bypass any lexical operator overloads too
MasterDuke don't think that's causing the problem i'm seeing though
i could change it to iseq_I, and isn't there some trick about checking whether an operator has been overloaded i could do? 23:58
this is what i'm getting right now `This type cannot unbox to a native integer: P6opaque, Str` for this example `sub a($a where 1|2|3|4|5|6|7|8|9|0|"abc") { $a }; say a("abc")` 23:59
putting the "abc" first works fine