00:18 Xliff joined 01:03 MasterDuke left 01:22 sena_kun joined 01:24 Altai-man_ left 01:53 tobs left 02:05 tobs joined 02:27 leont left 03:21 Altai-man_ joined 03:24 sena_kun left
Geth rakudo: 8f015aa0e2 | (Vadim Belman)++ | tools/templates/NQP_REVISION
[NQP Bump] Brings 8 commits

NQP bump brought: github.com/perl6/nqp/compare/2020....g0e341d45e 0e341d45e Merge pull request #601 from vrurg/rakudo_3499 87a811e58 Merge branch 'master' into rakudo_3499 83ffeec98 [MoarVM Bump] Brings 3 commits 3d69dc107 Merge pull request #602 from vrurg/fix-a-dep 2dad36fd4 Add nextdispatcherfor/takenextdispatcher to JVM backend a6d85cdce Merge branch 'master' into rakudo_3499 e143e024d Add a missing dependency aadea8867 Add nextdispatcherfor/takenextdispatcher ops
04:27 Xliff left 05:23 sena_kun joined 05:24 Altai-man_ left 06:03 lizmat left 06:38 Xliff joined 06:40 ZzZombo left 06:42 ZzZombo joined 06:48 ZzZombo left 06:50 ZzZombo joined 07:06 Xliff left 07:09 lizmat joined 07:21 Altai-man_ joined 07:24 sena_kun left 07:26 ZzZombo left 07:29 ZzZombo joined
lizmat pulled, built, test and spectest ok 07:35
07:36 lizmat left 07:54 ZzZombo left, ZzZombo joined 08:06 lizmat joined
nine This crops up quite frequently: push_o requires a concrete object (got a ThreadPoolScheduler::Queue type object instead) 08:49
08:52 lizmat left 09:01 lizmat joined 09:07 rypervenche left, CIAvash left, rba[m] left, Kaiepi left, finsternis left, unicodable6 left, greppable6 left, sourceable6 left, nativecallable6 left, shareable6 left, bloatable6 left, committable6 left, coverable6 left 09:22 sena_kun joined 09:24 Altai-man_ left 09:39 lizmat left 09:43 rypervenche joined, rba[m] joined, CIAvash joined, Kaiepi joined, finsternis joined, unicodable6 joined, greppable6 joined, nativecallable6 joined, sourceable6 joined, shareable6 joined, bloatable6 joined, committable6 joined, coverable6 joined 09:56 lizmat joined
nine The push_o thing actually seems to be relatively recent. I didn't do as many builds before as I'm doing now, but I don't remember running into this ever. 09:58
lizmat nine: do you have a gist reproducing that ? 10:00
nine Unfortunately not. It does crop at all over the place in some of the builds. I've seen it when building nqp and in several modules. 10:01
also in rakudo's install-core-dist.p6 10:03
Could be related to the I/O thread 10:04
Actually I'm not 100 % sure if I've seen it in nqp. That one seems to be failing most often in 111-spawnprocasync.t and 113-run-command.t with non-zero wait status: 6 10:07
11:21 Altai-man_ joined 11:24 sena_kun left 11:39 lizmat left 12:16 leont joined 12:20 lizmat joined
|Tux| Rakudo version 2020.02.1-99-g8f015aa0e - MoarVM version 2020.02.1-35-ga7fa6daad
csv-ip5xs0.699 - 0.707
csv-ip5xs-205.824 - 6.249
csv-parser24.320 - 24.479
csv-test-xs-200.360 - 0.364
test7.796 - 7.841
test-t1.799 - 1.829
test-t --race0.821 - 0.824
test-t-2029.812 - 30.573
test-t-20 --race8.838 - 9.107
12:50 ZzZombo_ joined 12:53 ZzZombo left, ZzZombo_ is now known as ZzZombo 13:22 sena_kun joined 13:24 Altai-man_ left
nine .seen pmurias 13:30
tellable6 nine, gist.github.com/5333b482be26404a5f...8fb25972ce
14:16 lucasb joined
vrurg lizmat: BTW, it turns out Backtrace is already lazy in a way. 14:22
15:21 Altai-man_ joined 15:24 sena_kun left
Geth rakudo/master: 11 commits pushed by (Vadim Belman)++
review: github.com/rakudo/rakudo/compare/8...e9b29429b6
roast: 92ee3413e6 | (Vadim Belman)++ | S06-advanced/dispatching.t
Add a regression test

This test is ok when dispatcher implementation utilizing
  `nextdispatcherfor/takenextdispatcher` ops is merged.
roast: 5a14ec2385 | (Vadim Belman)++ (committed using GitHub Web editor) | S06-advanced/dispatching.t
Merge pull request #627 from vrurg/rakudo_3499

Add a regression test
16:23 lizmat left 17:22 sena_kun joined 17:24 Altai-man_ left 17:33 leont left 17:47 Xliff joined
Xliff How can you get the value out of a parameter object? 17:47
m: sub a ('one', 'two', 'three') { say &?ROUTINE.signature.params.map( *.values ).join(', ') }; a('one', 'two', 'three'); 17:48
camelia Parameter<60800104>, Parameter<60800240>, Parameter<60800376>
Xliff I want to get <one two three>
m: sub a ('one', 'two', 'three') { say &?ROUTINE.signature.params.map( *.type.^name ).join(', ') }; a('one', 'two', 'three');
camelia Str, Str, Str
Xliff m: sub a ('one', 'two', 'three') { say &?ROUTINE.signature.params.map( *.name.^name ).join(', ') }; a('one', 'two', 'three');
camelia Nil, Nil, Nil
17:49 leont joined
sena_kun Xliff, if you know it's type, you can .gist and coerce 17:49
s/it's/its/ 17:50
Xliff m: sub a ('one', 'two', 'three') { say &?ROUTINE.signature.params.map( *.gist.Str ).join(', ') }; a('one', 'two', 'three');
camelia "one", "two", "three"
Xliff \o/
sena_kun++! 17:51
m: sub a ('one', 'two', 'three') { say &?ROUTINE.signature.params.map( *.gist ).join(', ') }; a('one', 'two', 'three');
camelia "one", "two", "three"
sena_kun Xliff, I am not sure if this approach is LTA, maybe it is, but at least it's obvious. :)
Xliff Actually, it's not. 17:52
A Parameter.value would have been obvious.
But this'll do.
17:53 MasterDuke joined
sena_kun well, maybe not "obvious", which is not the right word, but I mean "it seems like a bruteforce hack and it works". 17:54
as I imagine coercing is not the efficient/fastest thing in the world
Xliff You are correct, sir! 17:55
19:10 MasterDuke left 19:21 Altai-man_ joined 19:24 sena_kun left 20:24 MasterDuke joined 20:37 Xliff left 21:14 TreyHarris left 21:23 sena_kun joined 21:24 Altai-man_ left
MasterDuke m: constant N = 1_000_000; my int @a[N]; @a[$_] = $_ for ^N; say @a[N-1]; say now - INIT now 21:51
camelia 999999
MasterDuke ^^^ takes ~3.2s on my machine, but i have a terrible hack of a patch that drops it to ~0.16 21:52
unfortunately the rakudo build dies at the 'Preparing installation' stage with `STable conflict detected during deserialization. (Probable attempt to load a mutated module or modules that cannot be loaded together).` 21:53
the hack is to add `self.publish_method_cache($obj)` github.com/rakudo/rakudo/blob/mast...ch.nqp#L12 and github.com/rakudo/rakudo/blob/mast...ch.nqp#L19 and `self.publish_type_cache($obj)` 22:12
github.com/rakudo/rakudo/blob/mast...ng.nqp#L20 and github.com/rakudo/rakudo/blob/mast...ng.nqp#L25
before there were 1m calls to find_method and 2m calls to type_check (taking almost 3s in total), after about 20 apiece (taking no time) 22:15
setting RAKUDO_MODULE_DEBUG during the 'Preparing installation' stage gives: ` 1 RMD: Precompiling /home/dan/Source/perl6/install/share/perl6/core/sources/947BDAB9F96E0E5FCCB383124F923A6BF6F8D76B failed: 1` 22:18
`8660F65A7B3492675BB3B2058DB30E411A4C4E54core#sources/8660F65A7B3492675BB3B2058DB30E411A4C4E5489970C41D2528746FA4420FEC534FFFB40D3DB64CompUnit::DependencySpecification.new(short-name => "NativeCall::Types", from => "Perl6", version-matcher => Bool::True, auth-matcher => Bool::True, api-matcher => Bool::True)`
but that may not be all that relevant anyway, since there should be some better way of updating those caches 22:21
22:30 lizmat joined
MasterDuke ah, wrapping the `nqp::setmethcache, nqp::setmethcacheauth` and `nqp::settypecache` in `nqp::scwbdisable/nqp::scwbenable` let the rakudo install finish 22:31
heh, but neither test or spectest are happy 22:37
failure in t/04-nativecall/05-arrays.t and 16 spectest files 22:38
22:56 lucasb left 23:08 MasterDuke left 23:22 Altai-man_ joined 23:25 sena_kun left
Geth rakudo: MasterDuke17++ created pull request #3533:
Force publish type cache if found in the MRO