🦋 Welcome to the IRC channel of the core developers of the Raku Programming Language (raku.org #rakulang). This channel is logged for the purpose of history keeping about its development | evalbot usage: 'm: say 3;' or /msg camelia m: ... | Logs available at irclogs.raku.org/raku-dev/live.html | For MoarVM see #moarvm
Set by lizmat on 8 June 2022.
[Coke] and now they will live there forever. :( 00:22
ok. the percentage complete, as always, slows down as we get further. now at 80.44 00:23
Hopefully it'll be done in the morning.
01:39 MasterDuke joined 02:06 MasterDuke left
releasable6 Next release in ≈3 days and ≈7 hours. There are no known blockers. Please log your changes in the ChangeLog: github.com/rakudo/rakudo/wiki/ChangeLog-Draft 11:00
11:20 librasteve_ joined
[Coke] gist.github.com/coke/0a5f70703ba4f...fd8de8332d has the latest blin run output. To summarize, four module failures, all similar: 12:04
Physics::{Constants,Error,Measure,Navigation} - all die with Serialization Error: missing static code ref for closure '' (src/Perl6/Optimizer.nqp:3313) [Physics::Constants] at /blin/installed/Physics::Measure2.0.23/sourc 12:05
(obviously the installed path is different)
bisected to github.com/rakudo/rakudo/commit/16...b39723fe96 12:09
blin run finished in 12 hours, which seems on the low end for me. 12:10
librasteve_, lizmat: those are steve's modules and liz's commit, around making rationals remember where a /0 happened. 12:12
disbot2 <librasteve> hmmm - I wonder if that is because I use similar code on all these modules (eg use ../lib ... I do not do this any longer btw) or if there is some module loading circular dependency ... is there a way to install the latest rakudo locally (on Mac) - happy to help track this down 13:19
[Coke] rakubrew can do a build of HEAD 13:20
do you have xcode installed?
disbot2 <librasteve> =b
<librasteve> probably 13:21
[Coke] what does 'gcc --version' say?
(if installeddir ~~ Xcode, you can do a build, probably)
disbot2 <librasteve> reinstalling Xcode now
<librasteve> ... 13:22
[Coke] ... note that I did not do the blin run on a mac. 13:25
(it's some version of ubuntu)
disbot2 <librasteve> Apple clang version 21.0.0 (clang-2100.0.123.102) 13:37
<librasteve> what tag do I need for rakudo latest (for rakubrew build <tag> 13:42
<librasteve> )
<librasteve> since I added currency to Physics::Unit, the build step now needs web access 13:49
<librasteve> not sure if that could be a stopper?
lizmat confirmed problem with Physics::Constant 14:03
suggest we revert github.com/rakudo/rakudo/commit/16...b39723fe96
[Coke] works for me. 14:05
lizmat ok, double checking whether a revert fixes the issue :- 14:10
)
confirmed 14:16
Geth rakudo/main: 0fac56e299 | (Elizabeth Mattijsen)++ | 2 files
Revert "Make Rationals remember where they divided by 0 (#5659)"

This reverts commit 163acc853f8ced1cb1cdf51c8c4884b39723fe96.
Turns out this created some blin fallout:
   gist.github.com/coke/0a5f70703ba4f...fd8de8332d
... (8 more lines)		 14:18
[Coke] I don't treat those gists as permanent, btw. 14:26
disbot2 <melezhik.> Ok, my journey with segfault issue getting abit crazy 🤪 - github.com/rakudo/rakudo/issues/60...4128783056 18:24
timo having a way to guaranteed reproduce it is very valuable, even if it involves nuking a few folders between attempts 18:56
if i can get it reproduced on my machine, it should be much easier for me to actually find out what's wrong 18:57
can you tell me what distro you're using?
i can only see you're using the rpm packages but there's multiple distros that could apply to
disbot2 <melezhik.> Rocky Linux 9.7 19:06
timo i'm downloading the rocky linux 10 latest boot iso now so i can try to run it with qemu
disbot2 <melezhik.> Using nxadm.github.io/rakudo-pkg/ 19:07
timo btw this > echo '"task-run "tasks/task1"' > sparrowfile seems to give a sparrowfile with a syntax error (it starts with a double quote before task-run)
disbot2 <melezhik.> Yeah . But the test involves running qemu boot that produces some stdout 19:08
timo yeah
disbot2 <melezhik.> Maybe I can golf it abit further, I just any command that produce the stdout that would crash the Raku process reading it 19:09
<melezhik.> Yeah , seems a typo , let me update
timo ... 30% downloaded ... 19:10
disbot2 <melezhik.> Done
<melezhik.> What you are downloading ?
timo dl.rockylinux.org/pub/rocky/10/iso...4-boot.iso
disbot2 <melezhik.> Ok 19:11
<melezhik.> Should be good
timo distro.qcow2 is just an empty disk volume? 19:18
disbot2 <melezhik.> Let me add qemu related steps 19:19
timo ah i also don't have a vmlinuz to go with the iso it looks like 19:20
disbot2 <melezhik.> I will update ticket , it easy 19:21
timo OK
disbot2 <melezhik.> Done 19:25
<melezhik.> github.com/rakudo/rakudo/issues/60...4124586227 19:26
<melezhik.> Basically you need to download iso , create qemu drive file ( distro.qcow2) and run qemu vm booting from this iso , that’s it 19:29
<melezhik.> In my example this is Rocky Linux 10 iso , but this probably could be any iso
<melezhik.> -machine parameter could be varying depending qemu installation you use 19:30
<melezhik.> That’s it
timo i don't have 40 gigs of storage free for this experiment :D but qcow2 starts off sparse I think?
disbot2 <melezhik.> It’s dynamic
<melezhik.> It does not mean it takes all 40 gb 19:31
<melezhik.> It’s limit
[Coke] timo - do you need a box to work on?
timo ah you extract these things out of the iso
disbot2 <melezhik.> But you can change to 10 gb dies not matter
timo [Coke]: i should be able to work with my machine
disbot2 <melezhik.> Yeah extract kernel and initrd out of iso yes , probably we don’t need it for the test , but let me check 19:32
timo i had to change the machine typo to pc-q35-9.2
disbot2 <melezhik.> Sounds good
timo right we should be able to boot the boot loader on the iso
ah --append for the ttyS0 only works with -kernel 19:34
makes sense that you can't change kernel boot options when booting via the isolinux or grub or whatever's on the iso
disbot2 <melezhik.> Yes . You have learnt too fast ))
<melezhik.> I am impressed ))) 19:35
timo haha, no i use qemu for other things too so I know some random things
i do get some output, it looks funky :D 19:36
disbot2 <melezhik.> The test is taken out of real Rocky Linux test scenario . Some things may seems not relevant but I know at least this combination can hit Rakudo in a hard way )))
<melezhik.> It usually seg faults soon after echo “go” , few seconds after boot started 19:37
timo i see the grub menu but i can't interact with it and it takes 60 seconds of waiting before it boots the entry on its own :D
disbot2 <melezhik.> Interesting 19:38
<melezhik.> Let me run as well
timo it gets a little messed up by the timestamp output added by sparrowdo
disbot2 <melezhik.> Which iso do you use in test ? The same as I wrote in the ticket ? 19:40
<melezhik.> Hrm - my test output differs 19:42
<melezhik.> gist.github.com/melezhik/5c9173241...2f1f971c9d 19:43
timo well i removed the -kernel and -append flags 19:44
disbot2 <melezhik.> This is what I get
timo i didn't extract the vmlinuz yet
disbot2 <melezhik.> Ah , ok maybe this matters , maybe not … but append “console=ttyS0” maybe essential , maybe not 19:45
<melezhik.> But I had it mostly for logs from kickstart anaconda, which we don’t have here anyway 19:46
<melezhik.> If you paste here your qemu command I can run myself as well
timo excellent, i see the kernel console output 19:54
19:54:25 :: [ OK ] Started cancel-multipath-wait-sda.…ting for multipath siblings of sda.
19:55:10 :: anaconda 40.22.3.33-1.el10.rocky.0.3 for Rocky Linux 10.1 started. 19:55
Please make a selection from the above ['c' to continue, 'q' to quit, 'r' to
well, now what? :)
disbot2 <melezhik.> Ok . I run your way now without kernel and initrd 19:56
timo (now i have to find out why it works here and not on your machine)
disbot2 <melezhik.> Can you paste here your qemu command ?
timo qemu-kvm -net nic -net user,hostfwd=tcp::10022-:22 -m 6024M -cpu max -smp 8 -cdrom Rocky-10-latest-x86_64-boot.iso -boot d -vnc none -kernel vmlinuz -initrd initrd.img -append "console=ttyS0" --drive file=distro.qcow2 -no-reboot -vnc none --nographic -machine pc-q35-9.2 19:57
but also i've been working off of a suse leap container, now that i know what distro you're using i can try that instead 19:58
disbot2 <melezhik.> Ok
<melezhik.> And can post here raku -v 20:00
<melezhik.> And which raku?
timo v2026.02 from rakudo-pkg
disbot2 <melezhik.> Ok 20:01
<melezhik.> s6 —version
timo 0.0.87
disbot2 <melezhik.> Ok
timo ok, now i'm running qemu from a rocky linux 9.3 container and it's also going through the boot without apparent trouble 20:05
oh, but i'm not installing sparrow with --/precompile_install that was in your latest comment right? 20:06
it's running fine apparently 20:08
but it's supposed to crash very fast
next thing i'll have to try is installing rocky linux in a qemu and run all the things in there 20:09
disbot2 <melezhik.> How do you run qemu from container?
timo because with podman i still have the host's kernel which is different from what you have, and the kernel does have potentially some security settings and what-not
by ignoring that it says it can't find the "kvm" device and having it run slow :D
oh, this will be rocky 10, do you think that's a problem? is the other user you're in contact with also using rocky 9.3? 20:11
disbot2 <melezhik.> Yes the host system should be Rocky 9 20:12
<melezhik.> Or 10
timo if 10 for the host is fine, that's good. i'm running the installer right now 20:15
disbot2 <melezhik.> Ok
timo just to be sure, there's nothing additional near the segfault in the journal? 20:20
next is building my container on the rocky10 vm 20:29
> qemu-kvm: cannot set up guest memory 'pc.ram': Cannot allocate memory - is this why you were wondering how i can run qemu inside a container? 20:34
oh no i just tried to create the vm with 6 gigs of ram inside a vm that only has 2 or something 20:36
OK, still no crash. can you try with podman like I just did and see if it crashes for you on your machine? because if it works fine on your machine with my podman setup it probably doesn't have a hope of crashing here :) 20:37
[Coke] releasable6: next 20:38
releasable6 [Coke], Next release in ≈2 days and ≈22 hours. 1 blocker. 37 out of 46 commits logged
[Coke], Details: gist.github.com/522f09370299aca765...8e19a8a085
disbot2 <melezhik.> What do you mean by try with podman ?
[Coke] Took a first pass. librasteve_ - let's not include the rakuAST line in the weekly
let's maybe also not show the "Internal" section on the weekly 20:39
9 commits left to address in that gist if anyone wants to take a swing.
timo I created a simple containerfile that bases on rocky 9.3, installs rakudo-pkg, installs sparrow6, and then a corresponding "podman run" that puts the iso and vmlinuz and whatnot into the container and then runs sparrowdo there 20:41
disbot2 <melezhik.> How do it run qemu ? 20:42
<melezhik.> It’s getting late here I will be offline soon. 20:43
<melezhik.> One thoughts . Can you try the same test with Rakudo 2025.02 20:44
<melezhik.> Downloaded from Rakudo.org
timo yes, I'm doing that right now 20:45
Geth rakudo/coke/fix-6098: 92b688b828 | (Will Coleda)++ | src/core.c/Exception.rakumod
Fix grammar issue with exception message

Closes #6098		 20:46
timo Died because of the exception:
Error encoding UTF-8 string: could not encode codepoint 1577787664 (0x5E0B2110), codepoint out of bounds. Cann
disbot2 <melezhik.> The problem crash seems no longer is reproduced right now for Raku 2026.02 on my machine ( I don’t know why , because it did few hours ago ) but at least it reproduced for 2025.02
Geth rakudo: coke++ created pull request #6099:
Fix grammar issue with exception message
disbot2 <melezhik.> Funny thing is it sometimes that , sometimes segfault . Try to run few more times 20:47
timo ok i'll re-run. getting closer maybe?
disbot2 <melezhik.> Sorry ? 20:49
<melezhik.> Ahhh . Yes. But I expected you can reproduce it for 2026.02 two
<melezhik.> Two -> too 20:50
timo it also messes up my terminal and the container i built doesn't have stty in it to make it good again :D
or killall apparently
there's the segfault with v2025.02. 20:52
disbot2 <melezhik.> At least that
<melezhik.> Now to make for 2026.02 ha ha 😆
timo if i can track it down in the older version it'd be a very good start already 20:53
patrickb \o/
disbot2 <melezhik.> It’s abit bizzar test case, right? ))) 20:55
timo eh, it's okay 20:56
disbot2 <melezhik.> In the end I would like to see it can be golf down to something lighter ))
timo yes of course 20:58
disbot2 <melezhik.> Even though we can reproduce that for 2025.02 those two things keeps bothering me : 1) two users said today they still have crashes by running pretty much similar tests after they upgraded to 2026.02, one for Rakubrew , another for Rakudo-pkg, but those failures are random , not for every time 2) why can’t I reproduce the issue for Rakudo 2026.02 on my machine which has the same OS ? And why I was able to do it at least in limited “first 21:01
run only after reinstall” fashion and can’t now ? 21:02
<melezhik.> So questions , questions
timo maybe the precompilation files ended up where you didn't expect and you didn't clear them successfully? 21:04
you can set RAKUDO_MODULE_DEBUG=1 in the environment to get debugging output about that
> MoarVM oops: MVM_nfg_get_synthetic_info call requested a synthetic codepoint that does not exist. 21:06
now that's much more interesting and possibly a good hint
aha. fantastic. i have replaced qemu-kvm with cat 21:07
disbot2 <melezhik.> And still get segfault , right ? 21:09
timo looks like mostly the internal error for now 21:11
disbot2 <melezhik.> Yep sometimes it has this error as well , choose the one you like - 1) utf8 code point 2) get_syntetic 3) or segfault
<melezhik.> But segfault occurs more often out of these three errors
<melezhik.> Yes this was my initial line of thoughts , however if I clear everything and reinstall everything it should not happen, this is why I asked those two users not just upgrade Rakudo but also rm -rf ~/.raku before that … and reinstall Sparrow and friends 21:13
21:14 djinni` left
disbot2 <melezhik.> And btw don’t know if thus helps but one of the users just said to me “BTW -- the last two runs I've done have completed with no problem” 21:15
<melezhik.> Or maybe there is more to clear not just ~/.raku and Rakudo sometimes hit old cache or something? 21:16
<melezhik.> Second user report as well “Since upgrading to 2026.02 my tests are working so I'm good for the moment. :-)” 21:19
<melezhik.> Funny cause just after upgrade he complained on crashes as well . Is it really sparkling issue? ))) 21:20
timo sparkling? 21:22
disbot2 <melezhik.> I mean flaky 21:23
<melezhik.> random
timo ah
disbot2 <melezhik.> Sorry for my English , anyways 21:24
21:25 [Coke] left
timo no problem 21:25
thanks for taking so much time to go through this with me
disbot2 <melezhik.> Yeah . Likewise - thanks for all help and readiness to get deep 21:29
21:36 djinni` joined
disbot2 <melezhik.> And btw you don’t need sparriwdo to run tests just that - “raku -MSparrow6::DSL sparrowfile” 21:40
timo that would be a bit easier 21:43
disbot2 <melezhik.> Or just that - raku -e 'use Sparrow6::DSL; task-run "tasks/task1"' 21:45
<melezhik.> I have just checked it crashes too
<melezhik.> So we at least eliminate sparrowdo interference
<melezhik.> Only Sparrow 21:46
timo if you have xxd locally, try replacing the qemu invocation with this gist.github.com/timo/c8cba57454e55...775f50b69e 21:47
i can't seem to get the crash reproduced inside a podman container so i'm installing rakudo-pkg and Sparrow6 in my virtual machine directly 22:10
I do believe we did fix something that looks like the invalid synthetic issue in the near past, so maybe that's the only issue i can reproduce, which is not helpful since that would already be fixed with a newer version 22:33
even if i track that down there wouldn't be anything to do
so my current assumption is that there are two distinct problems, one of which was fixed in newer rakudo releases 22:35
unless you also saw the "invalid synthetic" error with 2026.02 from rakudo-pkg?
Mar 25 23:38:02 localhost.localdomain kernel: 4: AffinityWork[179466]: segfault at 4180000000a ip 00007f270d60ca12 sp 00007f27091565b0 error 4 in libmoar.so[35ca12,7f270d2b0000+7a8000] likely on CPU 1 (core 0, socket 1) 22:38
!!!!!!
of course i didn't even get a core dump, and my core dumps are all truncated as well without a clear reason why 22:46
really really really wish there were debug symbols in rakudo-pkg 23:07
Geth rakudo/main: 14eabf1ec4 | (Will Coleda)++ (committed using GitHub Web editor) | src/core.c/Exception.rakumod
Fix grammar issue with exception message (#6099)

Closes #6098		 23:09
timo let's see if the devbuild action gives me something i can use 23:13
huh. devbuild doesn't seem to put the build results anywhere 23:32
23:50 librasteve_ left

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!