00:01 MasterDuke left 00:12 oddp left 00:15 sena_kun joined 00:17 Altai-man_ left, sena_kun left, sena_kun joined 01:28 yuplushi joined 02:04 yuplushi left 02:14 Altai-man_ joined 02:17 sena_kun left
Geth rakudo: vrurg++ created pull request #3829:
[WIP] Rework tools/update-passing-test-data.pl to support backends
02:48
roast/roast_657: bdb775f971 | (Vadim Belman)++ | S14-roles/composition.t
Diamonds were fixed with the last role overhaul

Time to unfudge.
02:54
roast: vrurg++ created pull request #658:
[WIP] Putting a few things in order
02:55
03:42 Kaiepi left 03:44 Kaiepi joined 03:45 Kaiepi left, Kaiepi joined 04:14 Kaiepi left 04:15 sena_kun joined 04:16 Altai-man_ left 04:18 Kaiepi joined 05:18 yuplushi joined 06:14 Altai-man_ joined 06:17 sena_kun left 06:49 leont joined 07:13 JJMerelo joined
JJMerelo m: say "a3_" ~~ /^^<alnum>+$$/ 07:44
camelia 「a3_」
alnum => 「a」
alnum => 「3」
alnum => 「_」
tellable6 2020-07-24T15:59:21Z #raku <[Coke]> jjmerelo stackoverflow.com/questions/519478...pl-plugins - can you accept ugexe's answer there?
JJMerelo .tell [Coke] I was going to, but it's apparently disappeared? 07:45
tellable6 JJMerelo, I'll pass your message to [Coke]
08:02 MasterDuke joined 08:15 sena_kun joined 08:16 Altai-man_ left 09:07 oddp joined 09:48 Geth_ joined 09:54 Geth_ left 09:55 raku-bridge left, Geth__ joined 10:00 Geth__ left 10:01 raku-bridge joined, Geth___ joined 10:14 Altai-man_ joined 10:17 sena_kun left, raku-bridge left 10:24 raku-bridge joined, raku-bridge left, raku-bridge joined 10:31 Geth___ left 10:32 raku-bridge left 10:41 Geth_ joined 10:47 raku-bridge joined, raku-bridge left, raku-bridge joined 11:33 patrickb joined 11:36 Geth left 11:40 Geth_ left, raku-bridge left
patrickb o/ 11:40
11:41 Geth joined, raku-bridge joined, raku-bridge left, raku-bridge joined
patrickb I'd like to merge github.com/rakudo/rakudo/pull/3810 soon-ish (far away from the next release). That PR is deprecating a commandline option. I've added a deprecation notice, but I think we never deprecated a commandline option before. Should I merge or wait for more feedback? (From whom?) 11:44
11:50 raku-bridge left
JJMerelo patrickb wait, how much? 11:51
patrickb JJMerelo: I don't understand. Do you mean how long I would wait?
JJMerelo patrickb I mean, you can wait for ever...
11:52 raku-bridge joined, raku-bridge left, raku-bridge joined
JJMerelo patrickb the middle way is probably have a small deprecation cycle of one month or until the next release and wait that much. 11:52
Altai-man_ patrickb, I'd say a deprecation cycle for a year is good enough, considering it gives a big WARNING on building with this option on. 11:53
patrickb That's why I asked. I hope to either provoke a "merge it now" or "I'll have quick look" response. 11:54
timotimo patrickb: we have a deprecation cycle going on for --profile and friends at the moment 11:56
i would assume it's fine to merge this. i wonder if Inline::Perl6 needs a tiny patch?
patrickb timotimo: How did you do that? Just add a 'note "Deprecated. Don't use."' ?
timotimo it's a little more complicated than that 11:57
patrickb Well, `--nqp-lib` doesn't have any functionality. The deprecation is solely about not complaining when the option is given. 11:58
timotimo right
11:58 JJMerelo left
patrickb I was mostly concerned whether the deprecation output could have unwanted sideeffects. 11:58
timotimo i need to rebuild my nqp and rakudo, currently heap snapshots die because they don't find nqp-home method on the compiler
Altai-man_ timotimo, github.com/rakudo/rakudo/issues/3824 <- 11:59
Also, we must add rakudo-level tests for, at least, zero exit code for profile / heap snapshots features. 12:00
timotimo this got into the release?
Altai-man_ Yes.
patrickb github.com/Raku/nqp/pull/647 <- that's the fix.
Altai-man_ So we're waiting for the fix to be approved-merged, then tests to be added, then point.
timotimo i'll go write a test or two
patrickb The merge is currently only waiting for CI.
Altai-man_ timotimo, please do, I will be really happy. We have various rakudo-only tests and it is a bit ouch we did not have _anything_ for the profiler, so this sneaked in. 12:01
timotimo have we considered using bindhllsym here?
patrickb timotimo: Are you asking about `nqp-home` and `rakudo-home`? 12:04
timotimo yeah
patrickb If so, they originally used bindhllsym. But the approach caused unwanted duplication and coupling of rakudo and nqp.
So using a method on the Compiler object was the best alternative idea we could come up with. 12:05
timotimo mhm
12:05 Geth left, raku-bridge left 12:06 Geth joined, raku-bridge joined
patrickb timotimo: github.com/rakudo/rakudo/pull/3793 12:06
timotimo it occurs to me that we'll also want a way to pass just a directory rather than a full path to a file
to get a combinatin of "rakudo/nqp generates the filename for us" and "we get to choose where the file ends up"
patrickb timotimo: I remember that it was you who gave the idea of using bindhllsym for this more than a year ago. :-) 12:08
timotimo m)
patrickb have to leave. $work
o/
12:08 patrickb left 12:09 Geth_ joined, raku-bridge1 joined, raku-bridge1 left, raku-bridge1 joined, raku-bridge1 left 12:10 Geth_ left 12:15 sena_kun joined 12:16 Altai-man_ left 14:00 raku-bridge left 14:02 raku-bridge joined, raku-bridge left, raku-bridge joined 14:14 Altai-man_ joined 14:16 sena_kun left
Geth roast/roast_657: 3ea66087e5 | (Vadim Belman)++ | S26-documentation/12-non-breaking-space.t
Rewrite test to avoid use of a file handle

Do everything on a in-memory object. After all, the file content is
  `slurp`ed into one anyway.
14:25
[Tux] Rakudo version 2020.07-41-g3843a1e89 - MoarVM version 2020.07-6-g1f182b48c
csv-ip5xs0.821 - 0.846
csv-ip5xs-208.119 - 8.268
csv-parser26.109 - 28.804
csv-test-xs-200.381 - 0.382
test7.836 - 8.181
test-t1.904 - 1.932
test-t --race0.828 - 0.837
test-t-2033.013 - 34.848
test-t-20 --race9.223 - 9.726
14:38
15:03 raku-bridge left 15:09 raku-bridge joined, raku-bridge left, raku-bridge joined
tbrowder vrurg: ping back...i've see the issue thread but i haven't figured out the problem. are the S26 tests missing from the spectest.data file? 15:19
.tell vrurg ^^^ 15:24
tellable6 tbrowder, I'll pass your message to vrurg
lizmat notable6: weekly 15:26
notable6 lizmat, 2 notes: 2020-07-22T16:43:06Z <Altai-man_>: twitter.com/koto_san_kana/status/1...1466088449 ; 2020-07-26T18:40:53Z <guifa2>: 5ab5traction5.bearblog.dev/for-lov...e-underdog
Geth rakudo: 9071653e12 | (Patrick Böker)++ (committed using GitHub Web editor) | 7 files
Deprecate nqp-lib (#3810)

  * Remove all uses of the `--nqp-lib` command line parameter
The parameter has currently no effect. It probably became effectless during the relocatability refactor a year ago.
  * Add a deprecation notice when `--nqp-lib` is used
15:34
linkable6 RAKUDO#3810 [closed]: github.com/rakudo/rakudo/pull/3810 Deprecate nqp-lib
nqp: c858780bc8 | (Patrick Böker)++ | 2 files
Register a plain HLL::Compiler object under the 'HLL' name

This allows the code in NQP to use functionality in the HLL Compiler. Before that wasn't possible as it wasn't known which name the subclassed Compiler object would have.
Use that generic name to access it from NQP code. In particular in the logic that writes the `--profile` output.
This fixes rakudo/rakudo#3824
15:35
linkable6 RAKUDO#3824 [closed]: github.com/rakudo/rakudo/issues/3824 Cannot profile code anymore
Geth nqp: b94f860d25 | (Patrick Böker)++ (committed using GitHub Web editor) | 2 files
Merge pull request #647 from patrickbkr/fix-profile

Register a plain HLL::Compiler object under the 'default' name
lizmat notable6: weekly reset 15:37
notable6 lizmat, Moved existing notes to “weekly_2020-07-27T15:37:48Z”
15:38 MasterDuke left
vrurg tbrowder: I've added one more comment to the issue. 16:09
tellable6 2020-07-27T15:24:14Z #raku-dev <tbrowder> vrurg ^^^
16:15 sena_kun joined 16:16 Altai-man_ left
tbrowder vrurg: i understand now, looks good, i'm happy for you to merge as is. 16:26
16:36 Kaiepi left 16:38 Kaiepi joined 16:44 MasterDuke joined 16:45 Kaiepi left 16:49 Kaiepi joined
vrurg tbrowder: can you rewrite S26-documentation/12-non-breaking-space.t to make it less EVAL-encapsulated? 16:59
tbrowder: I mean the way it is implemented now it can serve as a bug-catcher, not certainly not a debug-helper. 17:00
*but certainly
tbrowder i would but i'm not sure how to do that unless i added it as a separate code generator behind the scenes inside the roast repo, but @ugexe doesn't like cluttering the tree. what do you suggest? 17:04
vrurg tbrowder: a compromise as in my last comment to github.com/Raku/roast/issues/657 17:06
tbrowder: a POD can be produced in an eval but then processed outside of it. 17:07
tbrowder hm, i can take a look. but i'm still confused about the "tests missing" part. are they or are they not currently in the master spectest.data being tested for a release? 17:21
vrurg tbrowder: they're not on spectest.data 17:34
17:35 |Tux| left
tbrowder hm, that surprises me, but i probably messed that up. 17:35
oh, well, moving on. i'll try to fix it all. thnx for heads up. 17:36
17:36 HarmtH_ left
Geth roast/roast_657: 64a70cf594 | (Vadim Belman)++ | 2 files
Add a canary test

The main tests are fudged with `skip` with which it is easy to miss when dispatching gets fixed. The canary test is fudged with `todo` and will result in a notification when it's about time to unfudge everything.
18:06
18:08 MasterDuke left 18:14 Altai-man_ joined 18:16 sena_kun left
Geth roast/roast_657: 002ed8b024 | (Vadim Belman)++ | 2 files
Add S16-unfiled/getpeername.t to spectest.data

The test is fudged as a `todo`
18:24
roast/roast_657: 1d7dc8591b | (Vadim Belman)++ | 3 files
Move the test into S32-str

And add it to spectest.data.
Two tests are fudged for JVM backend.
vrurg .ask patrickb github.com/Raku/roast/issues/657#i...-664565613 18:32
tellable6 vrurg, I'll pass your message to patrickb
18:37 lucasb joined
Geth nqp: f0cfdf2d87 | (Christian Bartolomäus)++ | 2 files
[JVM] Make osname available in nqp::backendconfig

This key is available on the other backends, and there are a couple of special cases in Rakudo's code base because the operating system name had to be taken from nqp::jvmgetproperties(){'os.name'}.
18:52
Altai-man_ Any blockers for a point release? 20:07
timotimo do we want any goodies in? :) 20:09
let's put some very experimental fancy stuff in at the last second!
Altai-man_ timotimo, maybe profiler tests first? :P
timotimo ah, that landed today 20:10
Altai-man_ not sure if kidding or serious 20:11
timotimo the fix i mean
Altai-man_ Ah, yup. 20:12
timotimo we have "is_run" or something in our tests?
Altai-man_ Well, that's why I'm planning to quickly do a point and finally rest for a bit...
Yes.
timotimo use lib $?FILE.IO.parent(2).add("packages/Test-Helpers");
use Test::Util;
hurm
20:14 sena_kun joined
sena_kun timotimo, can you confirm the fix is also good with heap snapshots? 20:15
20:17 Altai-man_ left
timotimo well, heap snapshots don't have to find a file on disk to concat with the data 20:18
sena_kun timotimo, didn't you mention it to be broken? 20:19
timotimo calling out to rakudo from a test can be super annoying
sena_kun I had to left though, so maybe misunderstood.
timotimo like, when you have rakudo-gdb-m or so, it'll use that version there too and give you quite some trouble
sena_kun timotimo, colabti.org/irclogger/irclogger_lo...-07-27#l89 <- this seems like exactly the problem, no? 20:20
timotimo right, i didn't look closely enough
sena_kun I assume the profiler works with the fix, but not sure how to test if heap profiler is and don't want to do another point in case it isn't. 20:21
timotimo ooh we have is-run in t/ and is_run in t/spec/ 20:22
brilliant
sena_kun timotimo, well, t/spec is roast and we don't want to test profiler there for sure, so rakudo/t is good. 20:23
timotimo yeah
i was just worried because i saw no results for is_run outside of t/spec
and thought i'd have to come up with something myself, which could very well be brittle and wrong and whatever 20:24
.o( make-temp-dir puts a "perl6" in the past, and also "roast" even when it's not t/spec ) 20:30
Geth rakudo/release-2020.07.1: b09f26ec54 | Altai-man++ | docs/release_guide.pod
Add point and fix release names
20:33
rakudo/release-2020.07.1: 9724703e6c | Altai-man++ | 2 files
Changelog + release announcement
timotimo writing those tests now don't release without me lol 20:35
sena_kun timotimo, well, they won't get into this point anyway, I think, but will prevent some point releases in future.
timotimo ok 20:36
sena_kun And I don't have so much time today.
Geth ¦ problem-solving: vrurg assigned to jnthn Issue Do we need :desc named in Test routines? github.com/Raku/problem-solving/issues/215 20:38
lizmat and another Rakudo Weekly News hits the Net: rakudoweekly.blog/2020/07/27/2020-...t-on-time/ 20:50
timotimo lizmat: no mention of the point release? :D 20:53
sena_kun timotimo, it's a new tradition to skip a week. ;> 20:54
lizmat there was a point release ?
sena_kun lizmat, will be tonight.
timotimo :D
sena_kun lizmat, profile breakage is, apparently, fixed. 20:55
lizmat didn't make the deadline :-)
sena_kun Maybe heap snapshot one as well, though no idea if the issue timotimo has found was related or not.
lizmat I'm not seeing the profile breakage fixed :-( 20:56
sena_kun lizmat, you need to locally bump nqp. 20:57
To see this.
I did not do this because the bump will happen anyway with the release.
lizmat ok, then I'll wait :-)
sena_kun lizmat, can you please do a local bump and double-check? :) 20:58
lizmat I've been burned so many times with local bumps in the past, that I've become really reluctant to do that :-( 20:59
20:59 Xliff joined
sena_kun No problem, then wait for the release. :) 20:59
lizmat got a bit tired of needing to nuke my install dir 21:00
Geth rakudo: d7aa7e663c | (Timo Paulssen)++ | t/09-moar/01-profilers.t
add simple tests to make sure profilers can be invoked

  (and have an actual effect)
timotimo these are the tests
Geth rakudo/release-2020.07.1: 6d0eff0c6d | (Timo Paulssen)++ (committed by Altai-man) | t/09-moar/01-profilers.t
add simple tests to make sure profilers can be invoked

  (and have an actual effect)
21:09
rakudo/release-2020.07.1: 37b8d5bc9d | Altai-man++ | docs/release_guide.pod
Add point and fix release names
rakudo/release-2020.07.1: d11c27f8c0 | Altai-man++ | 2 files
Changelog + release announcement
sena_kun Just a rebase to get tests in.
sena_kun timotimo, "t/09-moar/01-profilers.t ........................................ Dubious, test returned 2 (wstat 512, 0x200)" 21:28
timotimo great! 21:29
why?
sena_kun Eeh. 21:30
It passes fine when I run it using built rakudo... Weird... 21:31
Failed tests: 2-3
timotimo plenty other tests use is-run successfully, right? 21:33
sena_kun Yeah.
timotimo grmbl 21:34
how do i get "make blah.t" to output more than just the ok or not ok lines 21:35
sena_kun Yes, that's the question.
The problem it, the test passes just fine if I run it directly, without make.
Trying again.
timotimo --no-merge on the harness 21:36
sena_kun Did not help, hmm. 21:40
===( 1;0 0/6 1/12 0/? 0/? 0/? 0/? )========================# Failed test 'profiler html output file exists on disk'
timotimo yeah, the file does not actually exist 21:41
sena_kun But we test for its existence?
>ok $htmlpath.IO.f, "profiler html output file exists on disk";
The $htmlpath for me is "/tmp/perl6_roast_01_profilers_t_line1_0_41108281079771091595886049/profile.html"
timotimo # Writing profiler output to /tmp/perl6_roast_01_profilers_t_line1_0_97792504619008721595886087/profile.html
# Could not locate profiler/template.html; should have been at /home/timo/perl6/share/nqp/lib/profiler/template.html; outputting sql data instead
# Writing profiler output to /tmp/perl6_roast_01_profilers_t_line1_0_97792504619008721595886087/profile.sql
there's an install/ missing in the middle 21:42
/home/timo/perl6/install/share/nqp/lib/profiler/template.html exists
sena_kun So who is wrong?
timotimo the test is successfully detecting that stuff isn't working 21:43
sena_kun But if I do `install/bin/raku --profile -e 'say "foo"'` it says `Writing profiler output to profile-1595886239.0356207.html` and the file is here and working. 21:44
Let me do a bump... 21:45
timotimo nqp::static-nqp-home=/home/timo/perl6/install/share/nqp
nqp::source-digest=6c4c5320411a7c2bfcf5d1224e20d255671e3bc7
sena_kun Oh, hm, it's not so easy.
timotimo nqp::prefix=/home/timo/perl6/install
nqp::nqp-home=/home/timo/perl6/install/share/nqp
nqp::libdir=/home/timo/perl6/install/share/nqp/lib
these look fine
Raku::prefix=/home/timo/perl6/install 21:46
Raku::static-nqp-home=
Raku::static-rakudo-home=/home/timo/perl6/install/share/perl6
i wonder why the static-nqp-home there is empty, but probably just not using it? and maybe that's fine? or maybe that's wrong? 21:47
sena_kun Right, it creates .sql instead of .html. 21:49
Ok. 21:50
timotimo execname is /home/timo/perl6/rakudo/rakudo-m
sena_kun This sounds like a legitimate bug...
timotimo i believe the code we have cuts off two slashes from the end, then just plops /share/perl6 at the end
that's in Compiler.nqp in method rakudo-home inside rakudo 21:51
ok i can reproduce it without "make" now 21:52
./rakudo-m -I lib t/09-moar/01-profilers.t inside my rakudo build folder 21:53
sena_kun timotimo, are you sure you are using rakudo with bumped nqp?
timotimo This is nqp version 2020.07-3-gf0cfdf2d8 built on MoarVM version 2020.07-3-gb56e62f18 21:54
sena_kun Yeah. 21:55
timotimo, thanks for writing a test, this is a life saver, huh.
Ok, now sleep. 21:56
timotimo ok, so the code in rakudo will probably go via HLL::Compiler to moar's HLL/Backend, which uses getcomp("default")'s .nqp-home to find the right path
good night sena_kun
# execname: /home/timo/perl6/rakudo/rakudo-m 21:57
# rakudo-home: /home/timo/perl6/rakudo/gen/build_rakudo_home
# def-nqp-home: /home/timo/perl6/share/nqp
these two last ones are both wrong for my installation
22:02 travis-ci joined
travis-ci Rakudo build failed. Timo Paulssen 'add simple tests to make sure profilers can be invoked 22:02
travis-ci.org/rakudo/rakudo/builds/712344358 github.com/rakudo/rakudo/compare/9...aa7e663c07
22:02 travis-ci left 22:14 Altai-man_ joined
Geth rakudo: 51ea87a3e7 | Altai-man++ | t/09-moar/01-profilers.t
Skip dying tests to unbreak master builds

While github.com/rakudo/rakudo/issues/3824 is not fixed yet, keep the working commits to master green.
22:15
Altai-man_ Hope this helps.
22:15 leont left 22:17 sena_kun left 23:16 travis-ci joined
travis-ci Rakudo build passed. Altai-man 'Skip dying tests to unbreak master builds 23:16
travis-ci.org/rakudo/rakudo/builds/712364395 github.com/rakudo/rakudo/compare/d...ea87a3e763
23:16 travis-ci left 23:47 finsternis joined 23:48 oddp left