01:09 oddp_ left
Geth_ rakudo: vrurg++ created pull request #3795:
Make core setting name mapping more consise
rakudo: vrurg++ created pull request #3796:
Fix for spectest not seeing Inline::Perl5
rakudo: 6e73aa52b1 | (Vadim Belman)++ | 2 files
Fix for spectest not seeing Inline::Perl5

Fixes #2815. The problem occurs when the module is installed in the system rakudo repository. Fix as suggested by @ugexe in
linkable6 RAKUDO#2815 [closed]: github.com/rakudo/rakudo/issues/2815 Inline::Perl5 integration tests are no longer run
Geth_ rakudo: 847dd8db43 | (Vadim Belman)++ (committed using GitHub Web editor) | 2 files
Merge pull request #3796 from vrurg/rakudo_2815

Fix for spectest not seeing Inline::Perl5
rakudo: ddebab5a69 | (Vadim Belman)++ | .gitmodules
Change perl6 to Raku in nqp-configure submodule URL
rakudo: 6e087e13ea | (Vadim Belman)++ | 4 files
Do core setting name mapping via compiler config

The main source of language revision data is tools/templates/RAKU_SPECS file which is parsed and processed by tools/lib/NQP/Config/Rakudo.pm. Therefore it is more reliable to automate the generation of NULL -> CORE mapping using build-time macros than to manually manage a hash somewhere in obscure ModuleLoader location. ... (6 more lines)
rakudo: 15fae6d4e1 | (Vadim Belman)++ (committed using GitHub Web editor) | 5 files
Merge pull request #3795 from vrurg/pre-gen-core-mapping

Make core setting name mapping more consise
releasable6 Next release in ≈1 day and ≈15 hours. There are no known blockers. Please log your changes in the ChangeLog: github.com/rakudo/rakudo/wiki/ChangeLog-Draft 03:00
03:00 sivoais_ left 03:01 sivoais joined 06:59 patrickb joined 07:17 JJMerelo joined 07:40 MasterDuke joined 07:51 sena_kun joined 08:24 vrurg left 08:25 vrurg joined
Geth_ rakudo: patrickbkr++ created pull request #3797:
Rakudo home env precedence
08:29 vrurg left
Geth_ nqp: patrickbkr++ created pull request #646:
NQP_HOME env var should override a static NQP home
08:36 oddp_ joined 09:12 Altai-man_ joined 09:14 sena_kun left
timotimo we didn't bump moar and nqp yet? 09:18
Altai-man_ timotimo, we did. 09:19
timotimo well, i assume we'll want this nqp pull request in there, as well
Altai-man_ I will cut release from github.com/rakudo/rakudo/commit/85...3807035b09 09:20
09:20 JJMerelo left
timotimo the last change to MOAR_REVISION that i can see is 2 weeks old tho 09:24
Altai-man_ Oh, nqp was not bumped, apparently. 09:26
timotimo don't we call it "bump moar" when we change nqp to require a newer moar, and "bump nqp" when we change rakudo to require a newer nqp? 09:27
anyway, if you're cutting from that earlier rakudo commit, the nqp::setthreadname stuff won't be used by Thread anyway
Altai-man_ You're right. 09:28
timotimo so only spesh, ioloop, and debugserver threads will have "a name"
(that is, if we do a moar bump as well, or cut the moar release from master rather than MOAR_REVISION) 09:32
lizmat Files=1308, Tests=113043, 218 wallclock secs (29.46 usr 8.33 sys + 3042.16 cusr 289.71 csys = 3369.66 CPU) 09:33
Altai-man_ The release will be cut from MOAR_REVISION. 09:34
timotimo we have a moar release volunteer yet? 09:35
Altai-man_ timotimo, I did it for last, dunno, three or something releases? :)
Oh, I see actual bugfix in those commits. Ok, screw that. 09:36
Altai-man_ bumps 09:37
timotimo if you mean the use-after-free, that bug was introduced just before that fix
Altai-man_ Sigh. 09:38
timotimo, I still see some spesh goodies, how nice are they?
timotimo slightly nice 09:39
i haven't done any significant benchmarks 09:40
Altai-man_ S17-procasync/kill.t fails single every time for me now. :( 09:45
09:48 vrurg joined 09:49 vrurg left, vrurg joined
patrickb The latest three commits (not part of the rakudo commit you intend to cut the release for) contain two fixes, one for argument handling on JVM on Linux and one for the `--profile` feature in relocatable builds. 09:52
Both are not particularly important, but want to make sure this doesn't pass unnoticed.
*NQP commits 09:53
Altai-man_: Can you send an invite for the Raku org to supernovus? His original invite (during the Perl6 -> Raku move) expired. 09:54
Geth_ nqp: 6bdf699219 | Altai-man++ | tools/templates/MOAR_REVISION
[MoarVM Bump] Brings 20 commits

MoarVM bump brought: github.com/MoarVM/MoarVM/compare/2...gd564695f6 d564695f6 fix the most obvious use-after-free in history 127803247 set debugserver's thread name 707f6ce70 Debugserver Protocol 1.2: add thread "name" field 1d52655e2 decode threadname with utf8-c8 ... (16 more lines)
rakudo: c38b5a366d | Altai-man++ | tools/templates/NQP_REVISION
[NQP Bump] 6bdf69921 [MoarVM Bump] Brings 20 c […]

NQP bump brought: github.com/perl6/nqp/compare/2020....g6bdf69921
Altai-man_ patrickb, done 09:57
patrickb Thank you!
Is the plan now to release from current HEAD? 09:58
Altai-man_ Yes.
timotimo am i allowed to put the commits on nqp master and rakudo master to make thread names a thing? 09:59
patrickb There is PR in Rakudo reverting / replacing a wrong commit in Rakudo: github.com/rakudo/rakudo/pull/3797
Altai-man_ timotimo, yes 10:00
timotimo cool
patrickb It's only waiting for review from vrurg as he wrote the original fix this PR replaces. Also not particularly bad to release the wrongish fix, but I dunno.
Geth_ nqp: 1cd9594ec5 | (Timo Paulssen)++ | src/vm/moar/QAST/QASTOperationsMAST.nqp
map nqp::setthreadname op to give current thread a name
Altai-man_ patrickb, we will wait, the release is tomorrow, now today anyway. 10:01
timotimo rakudo will want a bump for this tiny commit as well
Altai-man_ timotimo, will you do it or?
patrickb There is a corresponding NQP PR to the one I just mentioned. That will then also want a bump. 10:02
timotimo i don't have the script working that automatically generates a bump's commit message
Altai-man_ timotimo, ok, then let's wait for the patrickb case to be resolved. 10:03
timotimo sure
Altai-man_ The script is github.com/Raku/z and it is nothing but godlike IMO. 10:04
timotimo it wants full control of my clones 10:05
10:22 ShimmerFairy left
lizmat yeah, it didn't really match my setup either :-( 10:26
Altai-man_ lizmat, how so? 10:28
Asking out of curiosity.
lizmat I forgot the specifics
it didn't have additional value for me at the time to change my setup 10:31
Altai-man_: git pull, reconfig and again build failure for me :-(
Missing or wrong version of dependency '/Users/liz/Github/rakudo.moar/install/share/nqp/lib/MAST/Ops.nqp' (from 'gen/moar/BOOTSTRAP/v6c.nqp') 10:32
Altai-man_ lizmat, did it work after yesterday's cleanup or?
lizmat yesterday, I nuked my install dir 10:33
which means re-installing everything again :-(
realclean in nqp and rakudo did not fix this 10:34
lizmat nukes install again :-(
10:41 JJMerelo joined
timotimo huh that's weird 10:41
lizmat runs a spectest after installing zef and Inline::Perl5 again 10:42
timotimo: my changes to JSON::Fast made parsing my sample 5MB json sample file almost 2x as fast 10:45
timotimo oh wow 10:46
i was going to ask you for results :)
lizmat 3.262 -> 1.864
I'm not done yet
timotimo ask you for your results* 10:47
lizmat Altai-man_: make test and make spectest are clean for me
timotimo we shouldn't forget to merge your performance changes to rakudo's internal json impl 10:50
lizmat yeah, but *after* this release 10:51
timotimo sure
11:10 dogbert17 joined
dogbert17 hmm, is the build borked atm? 11:11
Stage start : 0.000 11:12
Missing or wrong version of dependency 'gen/moar/stage2/NQPHLL.nqp' (from 'gen/moar/BOOTSTRAP/v6c.nqp')
at <unknown>:1 (/home/dogbert/repos/rakudo/install/share/perl6/lib/Perl6/BOOTSTRAP/v6c.moarvm:<dependencies+deserialize>)
11:13 sena_kun joined
lizmat dogbert17: welcome to my world 11:13
dogbert17 :-)
I guess it will be fixed soonish 11:14
11:14 Altai-man_ left
lizmat dogbert17: afaik, nobody knows why yet 11:16
I only know it started happening after patrickb's merge yesterday
patrickb I have an idea what the probleme might be 11:19
github.com/Raku/nqp/commits/master <- Should those RakuAST commits be there on master?
I did an NQP bump in yesterdays merge. Those commits were part of the bump. 11:20
lizmat which ones? 11:21
dogbert17 32be685690ec87b perhaps 11:22
linkable6 (2020-07-13) github.com/Raku/nqp/commit/32be685690 Remove reliance in a $*W object in alt NFA gen
dogbert17 There are three commits from jnthn there. Question is if they made it into master by mistake, i.e. git happens :) 11:23
timotimo that looks harmless 11:27
yeah those can go in a release no-prob 11:28
patrickb Does anyone actually know what the "Missing or wrong version of dependency" error is actually about?
timotimo something not being rebuilt when it should have been? 11:31
patrickb So changes to the build system would be potential causes. 11:32
dogbert17 last working rakudo commit seems to be 85320f46639a0 11:33
linkable6 (2020-07-12) github.com/rakudo/rakudo/commit/85320f4663 Revert "Add support for suggestions to X::CompUnit::UnsatisfiedDependency"
lizmat gist.github.com/lizmat/cf6f4178e6b...c2134edfe2 # patrickb
patrickb lizmat: That's in the NQP build, not Rakudo, right? 11:34
lizmat no, that's the rakudo build 11:35
MoarVM and NQP were already built
11:36 JJMerelo left 11:40 [TuxCM] left
patrickb lizmat: It breaks on nqp bumps, correct? So yesterday evening on my nqp bump and today on the one by Altai-Man? 11:41
lizmat yes
that would be a correct assessment, so far :-)
11:44 vrurg left 11:49 vrurg joined 11:52 [TuxCM] joined
patrickb lizmat: Do you build MoarVM, NQP and Rakudo separately or automatically via Rakudos Configure.pl? 11:53
11:54 vrurg left
lizmat automatically via Configure.pl 11:54
see my gist: gist.github.com/lizmat/cf6f4178e6b...c2134edfe2 :-)
patrickb lizmat: You did not clear the install/ folder prior to rebuilding nqp, correct? 12:05
lizmat indeed 12:06
nuking that makes the problem go away
12:08 wildtrees joined
patrickb The install/share/nqp/lib/MAST/Ops.nqp file that is claimed to have the wrong version is actually generated and installed by MoarVM. 12:11
jnthn It's compiled as part of the NQP build, however. 12:12
nine You should never have to nuke an install folder. I guess we're missing a dependency in a Makefile 12:14
patrickb I have to leave. Maybe I'll find some more time in the evening to have a look at this... 12:21
12:30 vrurg joined 13:12 Altai-man_ joined 13:15 sena_kun left
dogbert17 I got around the above problem by checking out and building 85320f46639a0, which worked, and then switching back to master and rebuild again, which also worked !?! 13:17
timotimo lizmat: i wonder if the cost of adding a tiny "nom-ws" version into parse-string (the fast-path version) to go in between cclass::word searches 13:21
is much higher than the win you could get for strings that only have space-separated words in them 13:22
i wonder if a split model of lexer, then parser, would be any win at all
it'd even be possible to run both in parallel, but the overhead may be completely unwarranted 13:23
perhaps "only if the json document is more than 32 kbyte big" or whatever 13:24
13:24 [TuxCM] left
timotimo i'm not very fond of just willy-nilly starting a promise in a library that's used in big as well as small programs 13:25
since if you just read a config at the start, then the rest of the program is fully single-threaded / synchronous, you'll have the supervisor running and twiddling its thumbs for the entire run time -> cpu and memory usage increased
13:37 [TuxCM] joined
timotimo (certainly not worth it with the implementation attempt i just came up with, d'oh) 13:38
lizmat timotimo: my approach atm is: 13:54
1. encountered a "
2. look for the closing unescaped "
3. convert string inbetween including the starter to Uni
4. drop the first element of Uni 13:55
5. generate the appropriate codepoints from the Uni array
6. convert to string and return
this should fix all of the decorated " stuff at the start of the string 13:56
without causing too much overhead
ah, Uni is NFD of course
jnthn Uni is unopinionated 13:58
timotimo Unipinion
14:07 wildtrees_ joined 14:10 wildtrees left 14:18 AlexDaniel left 14:21 vrurg left 14:24 vrurg joined
lizmat hmmm... looks like Pod::To::Markdown fails to install atm 14:42
which is annoying, as its one of the App::Mi6 dependencies :-(
ah, the joys of nuking ones install dir
[TuxCM] Unhandled exception: Missing or wrong version of dependency 'gen/moar/stage2/QAST.nqp' (from 'gen/moar/Pod.nqp') 14:45
at <unknown>:1 (/pro/3gl/CPAN/rakudo/install/share/perl6/lib/Perl6/Pod.moarvm:<dependencies+deserialize>)
lizmat Could not find Pod::To::HTML:auth<github:finanalyt> 14:46
lucs Should be 'finanalyst' I think. 15:02
lizmat yeah, already did a PR and an issue 15:03
lucs Ah, okay, just saying.
15:09 SmokeMachine left, SmokeMachine joined
Geth_ rakudo: 1a0be01fbc | (Patrick Böker)++ | 2 files
Revert "Merge pull request #3796 from vrurg/rakudo_2815"

This reverts commit 847dd8db436684b2e6b6eb29ba6fed4aad94340b, reversing changes made to 37d58315d68ab9ea529b0cf27632ff973c8014f4.
Proper fix follows.
rakudo: d56307201a | (Patrick Böker)++ | src/Perl6/Compiler.nqp
RAKUDO_HOME env var should override a static Rakudo home

So one can still influence the Rakudo home even with non-relocatable Rakudo builds.
This fixes #2815
linkable6 RAKUDO#2815 [closed]: github.com/rakudo/rakudo/issues/2815 Inline::Perl5 integration tests are no longer run
rakudo: 4a115c4070 | (Patrick Böker)++ (committed using GitHub Web editor) | 3 files
Merge pull request #3797 from patrickbkr/rakudo_home_env_precedence

Rakudo home env precedence
nqp: d70a1fa4bf | (Patrick Böker)++ | src/HLL/Compiler.nqp
NQP_HOME env var should override a static NQP home

So one can still influence the NQP home even with non-relocatable NQP builds.
nqp: 3e928af0be | (Patrick Böker)++ (committed using GitHub Web editor) | src/HLL/Compiler.nqp
Merge pull request #646 from patrickbkr/nqp_home_env_precedence

NQP_HOME env var should override a static NQP home
15:12 sena_kun joined 15:14 Altai-man_ left
lizmat patrickb: I guess we need another NQP bump ? 15:40
patrickb lizmat: Altai-Man planned to do one prior to the release, so we can just wait for that. 16:00
lizmat eh... don't we need a Blin run with it? 16:01
patrickb I guess so
Altai-Man: What's the plan wrt bumping nqp? 16:02
tellable6 patrickb, I'll pass your message to Altai-man_
sena_kun patrickb, was the fix of the fix you mentioned merged? 16:06
patrickb sena_kun: Yes. That's the commits you see above. From my side everything is ready for a bump. 16:07
sena_kun If so, I'll make a bump and it goes without saying run blin...
patrickb vrurg: Can you have a look at gist.github.com/lizmat/cf6f4178e6b...c2134edfe2 I'd value your opinion. 16:08
sena_kun++ 16:09
maybe I'll be back a bit later
16:09 patrickb left
Geth_ rakudo/rakuast: 9ed5daac0d | (Jonathan Worthington)++ | 7 files
Handle implicit $_ params in RakuAST

Handing this in the AST means that those doing synthetic construction may simply create a RakuAST::Block as the body of `when`/`if`/`for`
  (and further cases) and have the right thing happen with regard to
implicit declaration of a $_ parameter. This fixes up handling of $_ in a `{...}` block too. ... (5 more lines)
jnthn Phew, that'll unblock various things. 16:30
sena_kun jnthn++ 16:31
Geth_ rakudo/rakuast: f7d09f275a | (Jonathan Worthington)++ | 4 files
RakuAST for the `given` statement
jnthn Like that, which became trivial
timotimo way cool 16:53
soon we'll have the entirety
jnthn FSDO "soon" ;) 16:57
17:12 Altai-man_ joined 17:14 sena_kun left
jnthn home time o/ 17:16
17:46 HarmtH left
timotimo lizmat: if we toss out the bounds check in nom-ws, it falls below the inline threshold 17:49
18:04 leont left
timotimo i think we'll want to rename "at end of string" to "at end of input", since strings are also an element of a valid json document, and it could be confused for "at the end of one of the strings inside the file" 18:16
18:40 HarmtH joined 19:13 sena_kun joined 19:14 Altai-man_ left
sena_kun releasable6, status 19:19
releasable6 sena_kun, Next release in ≈23 hours. There are no known blockers. Changelog for this release was not started yet
sena_kun, Details: gist.github.com/0e7967a130cfb52b12...d681d8293f
Geth_ rakudo: 8543f855b3 | Altai-man++ | tools/templates/NQP_REVISION
[NQP Bump] Brings 3 commits

NQP bump brought: github.com/perl6/nqp/compare/2020....g3e928af0b 3e928af0b Merge pull request #646 from patrickbkr/nqp_home_env_precedence 1cd9594ec map nqp::setthreadname op to give current thread a name d70a1fa4b NQP_HOME env var should override a static NQP home
19:35 oddp_ left
lizmat after this bump, rakudo does not build for me anymore again 19:59
what is the current remedy apart from nuking install ?
sena_kun Yes, I noticed that too. Something is _very_ busted. 20:00
This is probably a release blocker.
lizmat yeah :-(
so there is no remedy known atm? 20:01
sena_kun lizmat, can you please fill an issue? Not sure if I know more than you do, while I'm writing changelog.
lizmat will do 20:02
sena_kun Thanks.
lizmat hmmm.. is Geth_ awol? Or github not delivering? 20:06
Geth_ ¦ rakudo: lizmat assigned to patrickbkr Issue Build after an NQP bump, is broken github.com/rakudo/rakudo/issues/3798
sena_kun Geth just announced my bump, no?
lizmat it didn't announce the ticket, just the assigning ?
anyways, ticket made ^^ 20:07
sena_kun nice
lizmat meanwhile /me nukes install again :-(
vrurg lizmat: totally weird kind of problem. I plan to work on it today. 20:30
vrurg doubts it's about a dependency.
lizmat afaik I reverted all of my work on module loader 20:31
sena_kun vrurg, good luck!
lizmat vrurg: maybe you should double check my commits didn't do it
20:32 oddp joined
vrurg lizmat: I most afraid it's something about moar itself. 20:33
Geth_ ¦ rakudo: vrurg self-assigned Build after an NQP bump, is broken github.com/rakudo/rakudo/issues/3798 20:35
sena_kun releasable6, status
releasable6 sena_kun, Next release in ≈22 hours. 1 blocker. 83 out of 85 commits logged
sena_kun, Details: gist.github.com/526013d73e391ec300...e4ba937395
sena_kun Ok, so changelog is ready... 20:36
vrurg It's so nice to get a block in less than a day to the next release schedule...
sena_kun I think it was introduced last night (for my timezone, hm), so it is not like we had it for weeks. And we can always push the release a bit later, it is a matter of quality at this stage anyway. 20:37
AlexDaniel` at least it's not a day after the release
sena_kun Yesh. So much better. 20:38
.tell patrickb hi! Can you please review the changelog at github.com/rakudo/rakudo/wiki/ChangeLog-Draft and maybe fix description of your recent fixes? Thanks.
tellable6 sena_kun, I'll pass your message to patrickb
AlexDaniel` but maybe it's still an option to cut the release from before the breaking changes? 20:39
20:40 [TuxCM] left
sena_kun If the time comes to it - yes, no problem at all. Will see how the debug goes. 20:40
sena_kun goes afk 20:42
20:43 squashable6 left, squashable6 joined 20:53 [TuxCM] joined 20:55 [TuxCM] left
dogbert17 ha, the build broke again ? 21:00
fortunately the workaround I used before still seems to work 21:01
vrurg AlexDaniel`: there is no certainty about what is the breaking change. :( 21:02
vrurg is testing jvm backend for the problem.
dogbert17: nuke 'em all works like always. 21:03
dogbert17 I didn't have to do that fortunately
vrurg dogbert17: what have you done then? 21:04
dogbert17 I checked out an older rakudo commit, 6e087e13ea021260, configured an built it, then I checked out master and built it again and it worked 21:05
strange I know
git checkout 6e087e13ea02 21:08
linkable6 (2020-07-16) github.com/rakudo/rakudo/commit/6e087e13ea Do core setting name mapping via compiler config
dogbert17 perl Configure.pl --gen-moar --gen-nqp --backends=moar
make install
git checkout master
perl Configure.pl --gen-moar --gen-nqp --backends=moar
make install
21:12 Altai-man_ joined 21:14 sena_kun left
vrurg dogbert17: thanks! It might help in solving the mistery. 21:14
21:15 MasterDuke left 21:34 patrickb joined
patrickb vrurg: Please see my latest comment on the NQP bump bug. I think I found an important clue. 21:38
tellable6 2020-07-17T20:38:32Z #raku-dev <sena_kun> patrickb hi! Can you please review the changelog at github.com/rakudo/rakudo/wiki/ChangeLog-Draft and maybe fix description of your recent fixes? Thanks.
patrickb sena_kun: The changelog looks good. I'm sorry I didn't manage to do it myself in time this time. 22:29
tellable6 patrickb, I'll pass your message to sena_kun
Altai-man_ patrickb, oh, that's part of my duties anyway, so don't sweat over it. What matters is to explain what was done correctly, as I recall some fixes of --profile with relocatable builds, but did not found traces of it. Or I misunderstood them. 22:36
tellable6 2020-07-17T16:02:15Z #raku-dev <patrickb> Altai-Man: What's the plan wrt bumping nqp?
patrickb Altai-man: There is a fix for --profile. But it's in NQP. And as far as I know the Changelog only lists Rakudo commits. 22:37
I think that doesn't make much sense, but that's how it was handled in the past, right? 22:38
Altai-man_ patrickb, oh, I see. 22:40
patrickb WRT our NQP bump problem: I think I have found the cause, but am unsure how to best proceed. I hope vrurg can weight in with his opinion. 22:41
I'll have to go to bed soon. I'm unsure how much time I'll have tomorrow. :-/
vrurg patrickb: anywhere to look at your findings?
patrickb vrurg: See the bugreport
Altai-man_ vrurg, ticket description?
vrurg patrickb: reading it right now. Ok. 22:42
22:50 wildtrees_ left
patrickb I need to leave. I'll have a look tomorrow. 22:51
22:52 patrickb left 22:53 wildtrees_ joined
timotimo lizmat: github.com/timo/json_fast/tree/nom-ws-smaller - heres the change that moves error checks from nom-ws to other pllaces 22:54
22:56 wildtrees_ left 22:57 wildtrees_ joined
releasable6 Next release in ≈19 hours. 1 blocker. Please log your changes in the ChangeLog: github.com/rakudo/rakudo/wiki/ChangeLog-Draft 23:00
23:10 wildtrees_ left 23:13 sena_kun joined, wildtrees joined 23:14 Altai-man_ left 23:18 sena_kun left