00:04 leont left 00:05 leont joined 00:22 leont left 00:25 sena_kun joined 00:27 Altai-man_ left 00:32 hungrydonkey left 01:04 Kaiepi left 01:05 MasterDuke left, Kaiepi joined 01:18 lucasb left 01:28 squashable6 left 01:31 squashable6 joined 01:34 Altai-man_ joined 01:36 sena_kun left
Geth rakudo: 8454f8dbff | (Vadim Belman)++ | 3 files
Re-revert "Big rewrite of dispatchers and wrapping"

This reverts commit a27005b374f8e55179e27e298b17b543e8124bf5.
It is not the cause of problems with Text::CSV.
rakudo: vrurg++ created pull request #3517:
Fix problems with nextcallee
rakudo: ac9c8ec802 | (Vadim Belman)++ | 2 files
Fix nextcallee breaking the dispatcher chain

Without moarvm/moarvm#1274 the only way to pass current dispatcher to a subsequent one is by wrapping next callee object into a stub which sets
  `$*NEXT-DISPATCHER`. Hopefully it's just a temporary workaround.
rakudo: 59a86c3a1c | (Vadim Belman)++ | src/Perl6/Metamodel/Dispatchers.nqp
Revert to using invocator

Instead of `::?CLASS`. For dynamically generated method objects ::?CLASS may contain not the class it's installed into.
rakudo: 0f223ac9d4 | (Vadim Belman)++ (committed using GitHub Web editor) | 2 files
Merge pull request #3517 from vrurg/rakudo_3499

Fix problems with nextcallee
roast: vrurg++ created pull request #626:
Added regression test for nextcallee
roast: 17f7e150d9 | (Vadim Belman)++ | S06-advanced/dispatching.t
Added regression test for nextcallee

It must not break the dispatcher chain.
roast: 23452a31f2 | (Vadim Belman)++ (committed using GitHub Web editor) | S06-advanced/dispatching.t
Merge pull request #626 from vrurg/rakudo_3499

Added regression test for nextcallee
02:58 leont joined 03:35 sena_kun joined 03:37 Altai-man_ left
vrurg .tell lizmat Just have tested Text::CSV installation again on the HEAD. It works. That include my dispatchers. The git log says that method inlining is reverted. Hope this info helps. 03:46
tellable6 vrurg, I'll pass your message to lizmat
03:52 leont left 05:34 Altai-man_ joined 05:36 sena_kun left 05:50 jmerelo joined 06:53 jmerelo left 07:02 cognominal joined 07:05 cognomin_ left 07:35 sena_kun joined 07:36 Altai-man_ left
rba hankache: Still getting the same error build windows rakudo star: "Cannot chmod 755 perl6:No such file or directory at C:/Perl64/lib/ExtUtils/Command.pm line 284." see dev.azure.com/infra0037/raku/_buil...amp;l=1109 Any idea? 07:58
tellable6 rba, I'll pass your message to hankache
rba is start using Azure DevOps to automate windows builds. Anyone who like to join, please ping me. It's public viewable at dev.azure.com/infra0037/raku 08:01
08:07 domidumont joined
lizmat Files=1304, Tests=111208, 210 wallclock secs (28.52 usr 8.26 sys + 2943.95 cusr 280.90 csys = 3261.63 CPU) 08:23
tellable6 2020-02-27T03:46:27Z #raku-dev <vrurg> lizmat Just have tested Text::CSV installation again on the HEAD. It works. That include my dispatchers. The git log says that method inlining is reverted. Hope this info helps.
08:42 domidumont left
lizmat ok, I'm now 100%ish sure that the MoarVM bump is responsible for not being able to install any more modules 08:42
1. nuke install
2. Configure with bump and build
3. re-install zef
4. try to install Text::CSV: gist.github.com/lizmat/9a8d5929b19...6e2a17a000
sena_kun jnthn, perhaps jnthn will be able to investigate it today?
oops, lizmat ^ 08:46
lizmat jnthn: for reference, make test and make spectest are *clean* after a bump 08:58
lizmat reverts the bump locally, nukes install again and builds again 08:59
ok, looks like it is an interaction with work I did yesterday and the MoarVM bump 09:24
Geth rakudo: dedb7444bc | (Elizabeth Mattijsen)++ | 2 files
Revert the work I've done yesterday
09:34 Altai-man_ joined 09:37 sena_kun left
lizmat with a bumped system now, I get spectest errors that only occur with "make", not when running the test directly with raku 09:52
jnthn make test / make spectest were clean for me, fwiw, and had less problems under spesh stressing than before 09:55
lizmat it's very confusing... but at least make spectest / test are clean for me now *before* the bump 09:56
jnthn I've just pulled Rakudo HEAD so will run a test with that 09:57
But you've sent me a stack trace that has precomp store at the top, and my pull has a 220 line diff in that, so... :)
lizmat but installing with zef fails :-( 09:59
ok, going to revert more of my work of the past days
jnthn runs a `zef install Text::CSV` without the latest Rakudo pull 10:00
it fails because I typed Test::CSV :P 10:03
lizmat he...
it doesn't have to be Text::CSV, P5localtime will also do :-)
or unprint :-)
jnthn unprint is a cute name for read :P 10:04
Or it prints backspaces, or? :)
lizmat it bypasses $*OUT
jnthn OK, Rakudo HEAD was make test/spectest clean for me; installing it now 10:06
Text::CSV installed fine before that
Trying unprint
Also worked out 10:10
Only thing I didn't do is install a new zef... 10:11
Geth rakudo: cbd75f7230 | (Elizabeth Mattijsen)++ | 3 files
Revert more of my work of this week
lizmat it's now stable for me after this commit
now trying with a bump again
10:13 domidumont joined
lizmat my workflow: nuke install, build, cd zef, nuke .precomp in zef, raku -I. bin/zef install . , zef install P5tie 10:14
if that works, make test and make spectest
with the bump, installing zef itself works, but installing a module with it doesn't 10:18
reverting the bump locally again, to make 200% sure
ok, verified, for me at least: HEAD without the bump is stable, passes test / spectest and allows zef to install modules 10:29
bump MoarVM and zef cannot install modules 10:30
afk for a few hours&
jnthn OK. I won't have any time for it this side of GPW. 10:32
And if I can't repro it, then there's not much I can do. I've too much to do today to nuke my install. 10:33
11:04 titsuki joined
|Tux| Rakudo version 2020.02-68-gcbd75f723 - MoarVM version 2020.02-25-g69951892a
csv-ip5xs0.717 - 0.720
csv-ip5xs-205.855 - 5.895
csv-parser23.536 - 24.241
csv-test-xs-200.365 - 0.406
test7.633 - 7.841
test-t1.800 - 1.806
test-t --race0.804 - 0.816
test-t-2030.973 - 31.340
test-t-20 --race8.748 - 8.785
11:28 titsuki left 11:34 leont joined 11:35 sena_kun joined 11:37 Altai-man_ left
lizmat jnthn: looks like it was an artefact of my own setup... 11:47
jnthn lizmat: Hmm...interesting...details?
lizmat I used to have a ~/bin dir early in my path with symlinks to zef/ raku etc 11:48
i've now change that to have install/bin and install/share/perl6/site/bin in my path 11:49
and not have ~/bin in my path
no idea how this was different functionally from your setup 11:50
but since you couldn't repro and vrurg also couldn't repro
I figured the problem was with either MacOS or me
and figure /me would be more likely :) 11:51
running final spectest now, and then I will bump MoarVM
rakudo rather :-)
Geth rakudo: 4f63764c82 | (Elizabeth Mattijsen)++ | tools/templates/NQP_REVISION
Bump NQP to get latest MoarVM optimizations
jnthn Phew!
lizmat now testing reverting my own reverts 11:53
jnthn: BTW, any comments about modules.raku.org/dist/ValueType ? 11:54
Geth rakudo: 81ac37f88b | (Elizabeth Mattijsen)++ | 3 files
Revert "Revert more of my work of this week"

This reverts commit cbd75f7230562cb32605309055efe5f221d5839e.
After a local reorg, this now passes all tests and lets zef keep doing its work.
jnthn lizmat: Interseting. 12:13
lizmat hmmm... reverting my next revert shows the problem again 12:19
===SORRY!=== Error while compiling /Users/liz/Github/rakudo.moar/site#sources/092D7DA0185CE0F1584F42452CD1E526B4B9DE17 (Zef::CLI)
New type Perl6::Metamodel::ClassHOW for NQPRoutine is not a mixin type
so I guess some of my commits borked stuff
12:23 titsuki joined
nine lizmat: but please don't push any more reverts or fixes until we know they are gonna stick 12:24
lizmat nine: this is what I'm doing now: for each change, a complete recompile with a nuked install dir and a fresh zef 12:25
rba May I ask to get a commit bit on the rakudo github org? [Coke] can confirm that I have signed the CLA. I play around with build automation and like to give some love to the star release. 12:32
lizmat jnthn: ^^ 12:37
moritz: ^^ ?
jnthn rba: github username? 12:38
oh hah, it's just the same as here :) 12:39
rba jnthn: rba
jnthn Invitation sent
rba jnthn: Thank you. 12:44
jnthn: Does this give access to the star repo too? (Haven’t tried yet.) 13:10
13:15 hankache joined
hankache .seen rba 13:15
tellable6 hankache, I saw rba 2020-02-27T13:10:05Z in #raku-dev: <rba> jnthn: Does this give access to the star repo too? (Haven’t tried yet.)
Geth roast/RAKULIB: 114738cd50 | (Tom Browder)++ | 2 files
separate into separate test files to ease removal
hankache .tell rba github.com/hankache/rakudo-star-win this is what I managed to patch together to build star on windows 13:18
tellable6 hankache, I'll pass your message to rba
hankache .tell rba it needs a lot of work, but for the moment it does the job
tellable6 hankache, I'll pass your message to rba
rba hankache: got it 13:21
jnthn rba: Yes, I put you in that group also 13:22
Geth roast/RAKULIB: 01ea55c721 | (Tom Browder)++ | spectest.data
add the separate RAKULIB test file
13:29 hankache left 13:35 Altai-man_ joined 13:36 sena_kun left
lizmat na, afaik, it's just the import process looking for a EXPORT associative in the GLOBALish of the thing being imported 13:37
found the breaking commit, and it's not one that I expected: 747a35ba0b92cb3937a4 13:49
jnthn Yeah, 747s tend to be find, it's the 737s that are broken... 13:52
lizmat double double checks 13:54
Geth rakudo/master: 4 commits pushed by (Elizabeth Mattijsen)++ 13:58
vrurg hopes this will fix broken precomp dependencies too... 14:17
lizmat vrurg: please let me know if it doesn't 14:19
vrurg lizmat: sure. I had to wipe out .precomp after every change yesterday. 14:20
lizmat :-( sorry 14:21
I guess our tests of precomping are very insufficient :-( 14:22
vrurg lizmat: nothing to be sorry about. It's HEAD, things are breaking there! :D 14:25
BTW, my own fix of re-dispatching seemingly broke my own project. Though it most likely means I did it all wrong in the project. :) 14:26
lizmat: I couldn't stand but test it again. Unfortunately, deps are still broken. Changing a file causes: Missing or wrong version of dependency '/Users/vrurg/src/Raku/raku-Vikna/lib/Vikna/Widget.rakumod (Vikna::Widget)' 14:29
But I'd try nuking the whole install first, considering your story.
lizmat yeah, that's what I'm doing now to test every commit related to precomp 14:33
15:03 lucasb joined
vrurg lizmat: bad news: the deps are still broken. The likely cause is diamond dependecy. Changing a file works fine for a module where deps are linear vs. breaking one where two direct deps require a third one: A -> B -> C (ok) vs. D -> B -> C & D -> E -> C (fails) 15:07
15:11 lizmat left 15:35 sena_kun joined 15:36 Altai-man_ left
Geth roast/RAKULIB: d0fcd070ab | (Tom Browder)++ | 4 files
add missing test for RAKUDOLIB; change module file suffixes to 'rakumod'; update tests require
15:53 lizmat joined
lizmat so I'm looking at the code in CompUnit::PrecompilationId 16:19
and I wonder, why does that need caching at all? (except for the new-from-string case) 16:20
to not have to do the sanity check ?
or to have value type semantics? In the latter case, it would need a proper WHICH 16:21
16:25 hankache joined
lizmat .ask vrurg how is the precomp going for you ? 16:27
tellable6 lizmat, I'll pass your message to vrurg
16:29 hankache left
Geth rakudo: b712789e72 | (Elizabeth Mattijsen)++ | src/core.c/CompUnit/PrecompilationUnit.pm6
De-hllize PrecompilationUnitId cache
16:46 lizmat left 16:47 lizmat joined 16:49 hankache joined, lizmat left
Geth roast/RAKULIB: 87ecddc62c | (Tom Browder)++ | 3 files
Ensure test files are unique to allow parallel testing
16:55 titsuki left
japhb Just did a rebuild-all (rakudo from scratch + modules) a few minutes ago, and it died trying to install a module that is normally not a problem: Test::Mock 17:00
Now attempting against 2020.02, to make sure it's not HEAD instability 17:01
Geth rakudo: eb19a3d2e5 | (Patrick Böker)++ | .circleci/config.yml
Revert "Temporarily disable CircleCI building"

The problem has been analyzed, so we enable CircleCI again.
This reverts commit adc61a8f57f38196483dfb5e20ea2e94f1a6551e. This reverts commit efba963910f3f556cc68e172f2e5689487b81386.
sena_kun u: on 17:06
unicodable6 sena_kun, U+0021 EXCLAMATION MARK [Po] (!)
sena_kun )!"
unicodable6 sena_kun, U+0022 QUOTATION MARK [Po] (")
sena_kun, Cowardly refusing to gist more than 5000 lines
sena_kun u: "
unicodable6 sena_kun, U+0022 QUOTATION MARK [Po] (")
sena_kun hmm
nine lizmat: the PrecompilationId cache is just to avoid the costly validity check. 17:14
tellable6 nine, I'll pass your message to lizmat
17:34 Altai-man_ joined
vrurg lizmat: Oh, you didn't see this message from me: bad news: the deps are still broken. The likely cause is diamond dependecy. Changing a file works fine for a module where deps are linear vs. breaking one where two direct deps require a third one: A -> B -> C (ok) vs. D -> B -> C & D -> E -> C (fails) 17:36
tellable6 vrurg, I'll pass your message to lizmat
17:36 domidumont left 17:37 sena_kun left 17:45 MasterDuke joined 18:07 domidumont joined
Geth ¦ problem-solving: vrurg assigned to jnthn Issue RAKUDOLIB destiny? github.com/Raku/problem-solving/issues/167 18:15
18:34 hankache left
japhb BTW, rebuild-all at 2020.02 did *not* fail on Test::Mock. It did however fail at a later module that seems to have trouble on this particular machine. 18:56
19:32 domidumont left 19:35 sena_kun joined 19:37 Altai-man_ left, lizmat joined
lizmat japhb: could you create a recipe for the breakage 19:43
tellable6 2020-02-27T17:14:19Z #raku-dev <nine> lizmat: the PrecompilationId cache is just to avoid the costly validity check.
hey lizmat, you have a message: gist.github.com/b0b9d2b5b7720e5eff...ca575cbcb2
20:16 hankache joined 20:25 lizmat left 20:50 dumarchie joined
dumarchie My fix for infix:<eqv> between Iterables did not make it into the new release :( 20:51
Is there a way to export a custom infix:<eqv> that will be picked up by Test's is-deeply ? 20:52
japhb lizmat: I'm a bit slammed with $dayjob right now, but I'll see what I can do about narrowing it down. 21:03
tellable6 japhb, I'll pass your message to lizmat
21:11 dumarchie left 21:34 Altai-man_ joined 21:37 sena_kun left
Geth roast/RAKULIB: 8c5405dae8 | (Tom Browder)++ (committed using GitHub Web editor) | spectest.data
remove test for env var RAKUDOLIB

It is Rakudo specific and not part of the Raku specification, thus it is removed from roast.
roast/RAKULIB: 6545551cd2 | (Tom Browder)++ (committed using GitHub Web editor) | S11-modules/rakudolib.t
Delete RAKUDOLIB test file

The RAKUDOLIB env var is Rakudo-specific and thus is not part of the Raku language specification or its roast repository.
22:33 lizmat joined
lizmat is back from NR.pm meeting 22:35
tellable6 2020-02-27T21:03:14Z #raku-dev <japhb> lizmat: I'm a bit slammed with $dayjob right now, but I'll see what I can do about narrowing it down.
lizmat .ask vrurg does the problem still exist on HEAD ? 22:43
tellable6 lizmat, I'll pass your message to vrurg
vrurg lizmat: I'm on another bug in dispatchers. Will test when done. 22:48
lizmat oki!
vrurg thanks!
lizmat no, thank you! :-)
22:51 hankache left
vrurg lizmat: I just've notice there were no more commits. So, the problem is still there. 23:03
lizmat ok 23:04
vrurg The last commit from you is b712789e721ba01f43f7f8174c2ccbe40d4d8f2b
linkable6 (2020-02-27) github.com/rakudo/rakudo/commit/b712789e72 De-hllize PrecompilationUnitId cache
vrurg linkable6: thanks :)
linkable6 vrurg, Anytime!
lizmat just as a sanity test, could you check if the problem also exists when at 66b78ccbefecd6cb2ff41 ?
linkable6 (2020-02-23) github.com/rakudo/rakudo/commit/66b78ccbef Merge pull request #3510 from tinmarino/master
vrurg loves these smart guys!
lizmat: rebuilding... 23:06
lizmat: looks like it's not there. I get an error, but it's runtime and unrelated. 23:11
lizmat could you produce a gist to re-create the problem that you see on HEAD
I'll take that on tomorrow morning... first getting some sleep
vrurg lizmat: I'd try but can't promise. 23:12
lizmat whatever you can provide me to help me understand how I borked things
vrurg Sure. I just want to hunt down my bug and have not much time left today. 23:13
lizmat understood...
good hunting!
sleep& 23:14
vrurg Hope I got the cause right, then it shouldn't be hard to golf down.
lizmat: 'night!
23:16 lucasb left
vrurg .tell lizmat Here is the gist: gist.github.com/vrurg/7f190ded041b...57576dbc4e ; interesting that I can only reproduce it with classes. 23:29
tellable6 vrurg, I'll pass your message to lizmat
23:35 sena_kun joined 23:37 Altai-man_ left