00:15 oddp left 00:43 leont left
vrurg JVM build is broken again. :( 00:47
tellable6 2020-07-22T06:47:44Z #raku-dev <Xliff> vrurg For you? Available always. Email me at [email@hidden.address] if I'm not active on IRC
timotimo aw crap, the JVM doesn't have Uni 00:51
so we'll need a big if !moar with the older version, since i imagine js might also not have the Uni type? 00:54
i'm going to bed right now, perhaps someone else can get to it while i'm down
vrurg Hope someone would do it. I don't want to merge in a PR without successful CIs. 00:57
timotimo k maybe it'll only take a minute
patch has this pretty cool flag that makes #ifdef blah and #ifndef bl??h for you 01:03
actually that gives an unreadable mess as a result 01:05
so maybe i'll just concatenate both versions
01:07 nebuchadnezzar left
Geth_ rakudo: 43bd739ffc | (Timo Paulssen)++ | src/core.c/Rakudo/Internals/JSON.pm6
restore old from-json for backends without Uni (JVM)
01:10
timotimo let's hope azure will give the result quickly
vrurg timotimo: It's not that of an emergency! Take a rest! 01:11
timotimo ha
vrurg is anyway busy trying to clean up some of our roast mess.
Apparently, we have a bunch of passing tests outside of spectest.data. 01:12
timotimo i'll go to bed and since it takes me a good bit to fall asleep anyway, i can look at the ci results later
o/
vrurg o/ 01:14
Geth_ rakudo: vrurg++ created pull request #3818:
Do some clean up related to roast
01:39
roast/roast_654: 4a92caca23 | (Vadim Belman)++ | 3 files
Improve documentation

  - Extend CONTRIBUTING.md with information about problem-solving repo,
correct references, and added more sanity to test examples. Got rid of more `perl6` notions.
  - Cleaned up spectest.data header from all Rakudo-specific information.
Replaced RT references with GitHub.
Fixes #654
02:44
roast: vrurg++ created pull request #656:
Improve documentation
vrurg My time to give the eyes some break. 02:55
03:04 Xliff joined
[Coke] vrurg++ 03:12
04:10 Xliff left 04:37 lucasb left 04:52 nebuchadnezzar joined 07:00 samcv left 07:27 samcv joined
lizmat timotimo: maybe we should drop the updated JSON::Fast in the core... the idea is to remove it anyway? 07:28
08:01 leont joined
lizmat another thought: since the JVM doesn't have NFG, isn't it so that .ords on the JVM is basically .NFD ? 08:18
or would that be .NFC ? or undecided ?
09:01 oddp joined
lizmat m: class A:ver<1.2.3> { }; dd A.^ver # nice 10:42
camelia v1.2.3
lizmat m: class A { ... }; class A:ver<1.2.3> { }; dd A.^ver # huh ?
camelia Mu
lizmat looks like stubbing a class makes it lose meta-info
10:48 sena_kun joined 10:52 Xliff joined
jnthn lizmat: I think the info has to be on the stub 10:57
Since that's the point the meta-object starts existing 10:58
It's certainly lta that it's silently dropped later though
lizmat I thought that was only necessary for the REPR ?
anyways, why wouldn't ver / auth / api not be set if they weren't set on the stub yet ? 10:59
*would
jnthn Because they are arguments to .new on the meta-object, and that happens at the point it is stubbed
I think they're currently then immutable from that point on 11:00
Though I'm not totally sure
Hm, name isn't actaully
Anyway, it's fixable if so desired
(The repr case isn't though)
lizmat yeah, I'll make an issue
timotimo lizmat: what do you mean? 11:07
oh, now the JSON::Fast module will also not work on the jvm, huh 11:08
lizmat timotimo: yeah, :-(
timotimo d'oh
lizmat so maybe just revert it and let the Internals parser die a slow death
timotimo why is the plan to remove it? 11:09
lizmat and perhaps change it so that the core will use JSON::Fast if it is installed ?
why was it made Internals only ?
timotimo it was deemed to be for module-space, but it's also needed to load modules?
moritz because it's not part of the language? 11:29
lizmat afk for a few hours& 11:31
12:13 Altai-man_ joined 12:16 sena_kun left 13:23 Xliff left 13:37 [Tux] left 13:44 [Tux] joined 13:47 MasterDuke joined 14:00 maggotbrain left 14:01 maggotbrain joined 14:14 sena_kun joined 14:16 Altai-man_ left 14:22 Kaiepi left, Kaiepi joined 14:23 Kaiepi left 14:24 Kaiepi joined 14:25 Kaiepi left
tbrowder lizmat: got error installing Hash2Class. couldn't find bin/skeleton. think yr meta file needs changing or create a resources/bin in yr tree and copy bin/skeleton there, 14:27
Geth_ roast: 4a92caca23 | (Vadim Belman)++ | 3 files
Improve documentation

  - Extend CONTRIBUTING.md with information about problem-solving repo,
correct references, and added more sanity to test examples. Got rid of more `perl6` notions.
  - Cleaned up spectest.data header from all Rakudo-specific information.
Replaced RT references with GitHub.
Fixes #654
linkable6 ROAST#654 [closed]: github.com/Raku/roast/issues/654 spectest.data in roast needs some updates
roast: 38a907e027 | (Vadim Belman)++ (committed using GitHub Web editor) | 3 files
Merge pull request #656 from Raku/roast_654

Improve documentation
¦ rakudo/master: 5 commits pushed by (Vadim Belman)++ 14:28
¦ rakudo/master: 1e20dfcad5 | Fix scripts to be compatible with the current build system 14:30
14:31 Geth_ left 14:32 Geth joined 14:37 vrurg left 14:38 vrurg joined
Geth roast: 347f029402 | MasterDuke17++ (committed using GitHub Web editor) | spectest.data
Some minor editing cleanup to comments
14:45
roast: efd6269bb7 | MasterDuke17++ (committed using GitHub Web editor) | CONTRIBUTING.md
Some minor editing cleanup
14:53
lizmat tbrowder: interesting... 14:56
15:02 eaterof joined
lizmat .tell tbrowder A fixed version of Hash2Class is on its way to CPAN, thanks for the headsup! 15:02
tellable6 lizmat, I'll pass your message to tbrowder
tbrowder yr wlcm, see name clash discussion on #raku, pls 15:03
15:03 eater left
[Tux] For twitter, how do I get more digits on «perl6 -e 'say τ'» 15:08
m: say τ
camelia 6.283185307179586
[Tux] will use "raku" instead of "perl6" on twitter 15:09
15:11 MasterDuke left
[Tux] Rakudo version 2020.07-41-g3843a1e89 - MoarVM version 2020.07-3-gb56e62f18
csv-ip5xs0.788 - 0.799
csv-ip5xs-207.971 - 8.237
csv-parser25.064 - 26.640
csv-test-xs-200.392 - 0.397
test7.711 - 7.839
test-t2.100 - 2.147
test-t --race0.835 - 1.019
test-t-2033.637 - 34.055
test-t-20 --race9.106 - 9.503
15:11
[Coke] the constant is defined as: my constant tau = 6.28318_53071_79586_476e0;
... I can't get it to output the 476 at the end. 15:13
m: sprintf("%0.20f", tau).say
camelia 6.28318530717958600000
lizmat [Tux]: 2.100 - 2.147 that feels rather high ? 15:17
[Tux] me too, but there was no teams session. 15:19
twitter.com/m_ou_se/status/1285991222088208385 - feel free to add 15:20
I have to go now, sorry
[Coke] tux++ 15:21
timotimo [Tux]: ideally, #rakulang if you tag it 15:40
[Coke] I replied with #rakulang 15:47
not sure if that'll help or not. :)
15:51 travis-ci joined
travis-ci Rakudo build errored. Vadim Belman 'Merge pull request #3818 from vrurg/roast_654 15:51
travis-ci.org/rakudo/rakudo/builds/711116707 github.com/rakudo/rakudo/compare/4...020c1ca4fd
15:51 travis-ci left
lizmat m: sub a() { }; &a.set_why("foo"); dd &a.WHY # shouldn't this just work ? 15:59
camelia No such method 'set_docee' for invocant of type 'Str'
in block <unit> at <tmp> line 1
[Coke] why do we have a "set_why" that's all lowercase with an underscore? 16:03
lizmat been that way for at least 5 years, according to git blame 16:04
[Coke] (same with set_docee), guessing those are old. :) 16:05
looks like it's expecting a particular type of POD object. we should probably at least declare that set_why takes something that implements the role/class it's expecting. 16:09
(and maybe add a Str target that creates that object as if it were a small pod block)
16:13 Altai-man_ joined
lizmat R#3819 16:14
linkable6 R#3819 [open]: github.com/rakudo/rakudo/issues/3819 Classes with a stub silently ignore class adverb decorations
lizmat R#3820
linkable6 R#3820 [open]: github.com/rakudo/rakudo/issues/3820 Cannot use simple string to set the WHY on a Block
16:16 sena_kun left
[Coke] should we have one for updating method names with underscores? 16:20
should r#3820 go through problem-solving? (new method?) 16:26
linkable6 R#3820 [open]: github.com/rakudo/rakudo/issues/3820 Cannot use simple string to set the WHY on a Block
18:15 sena_kun joined 18:16 Altai-man_ left 18:51 Kaiepi joined
gfldex make spectests fails consistantly with gist.github.com/gfldex/c473eae3eb0...69072abc19 19:12
sena_kun gfldex, odd, CI is green. Can you please share some details about your system? 19:14
lizmat maybe ETOOMANYFILEHANDLES ? 19:15
timotimo oh, hm, theoretically we could check for ETOOMANYFILES or whatever everywhere we can see it and give a more helpful error message or warning or whatever directly from moarvm 19:16
gfldex Linux dexhome 5.4.0-4-amd64 #1 SMP Debian 5.4.19-1 (2020-02-13) x86_64 GNU/Linux
model name : AMD Ryzen Threadripper 1900X 8-Core Processor 19:17
I will try a different machine with a clean rakudo pull in a bit. 19:18
19:26 sena_kun left 19:27 sena_kun joined
bartolin a quick question: do you see any problem with using NQP's gen-jvm-properties.pl to add more info to nqp::backendconfig for the JVM backend? For instance: 19:40
r: use nqp; for <name osname osvers> { say nqp::backendconfig(){$_} }
gfldex same result on a different box 19:41
camelia moar
linux
4.12.14-lp151.28.52-default
(Any)
(Any)
(Any)
bartolin some things are available in nqp::jvmgetproperties, so there would be some redundancy 19:53
j: use nqp; say nqp::jvmgetproperties()<os.name> 19:54
camelia Linux
ugexe j: say $*KERNEL.name 19:59
camelia Error while reading '/home/camelia/p6eval-token': Permission denied at /home/camelia/rakudo-j-inst/bin/eval-client.pl line 10.
ugexe (for jvm that shells out to uname)
m: say $*KERNEL.version 20:01
camelia v1.SMP.Fri.Dec.6.13.50.27.UTC.2019.8.f.4.a.495
ugexe m: say $*KERNEL.release
camelia 4.12.14-lp151.28.36-default
ugexe although if you are just trying to track those items at build time then adding other properties sounds reasonable 20:02
bartolin I thought it would be possible to remove some special cases in rakudo (like this one: github.com/rakudo/rakudo/blob/3843...M.pm6#L110 ) 20:08
*be possible and simplifies things 20:10
ugexe i wonder why it acts one way for :U and another for :D... that seems off 20:11
bartolin didn't even spot that (it was just one result from grep) 20:13
20:13 Altai-man_ joined 20:16 sena_kun left
bartolin it was introduced with github.com/rakudo/rakudo/commit/e79d749864 (the commit message talks about the difference) 20:16
ugexe yeah i saw it was intentional. however it still seems off to do that unless its some hack for the build to work 20:17
bartolin anyway, thanks for looking. I think, I'll prepare a PR to add a few items to nqp::backendconfig. 20:21
20:37 Altai-man_ left 20:38 sena_kun joined
Kaiepi i'm working on porting my moarvm changes for problem-solving #111 to the jvm now 20:40
once i'm certain the api i have atm will work there as well, the solution should be pr-ready
tbrowder ugexe: does zef warn about bin files overwriting an existing file? 20:58
if not, what do you recommend doing about the problem of dup bin files being installed by different host modules? 21:00
ugexe zef doesn't overwrite anything 21:12
as for dupe bin files being installed by different modules... how would you do it for programs outside of raku? because PATH isn't owned by raku 21:16
remember all files are sha1-d, including the original bin scripts. CURI installs a shim for a given bin name (so two dists with a bin/foo there is only one `foo` shim), and that shim loads whatever bin script sha1 file has the highest version 21:18
even then, thats just at a single CURI and you can have many CURIs. so even though can shadow each other 21:19
and as mentioned earlier -- all sorts of stuff outside of raku-land can shadow it
As for the unofficial don't dont this answer -- the shims add in parsing for `--auth` and `--ver` so you can theoretically get a specific version or auth instead of highest version. HOWEVER that needs to go away because it means any raku program cant have --ver or --auth in their bin scripts 21:21
github.com/rakudo/rakudo/blob/3843...m6#L54-L56 21:22
but honestly we need to get rid of that and use a ENV variable or something 21:23
or perhaps just call them --rakudo-bin-version --rakudo-bin-auth etc (to free up stuff users will actually call their variables like `:$ver`) 21:27
22:03 squashable6 left 22:05 squashable6 joined 22:06 Xliff joined 22:13 Altai-man_ joined 22:16 sena_kun left 22:45 leont left
tbrowder ugexe: thanks. 22:56