Perl 6 language and compiler development | Logs at colabti.org/irclogger/irclogger_log/perl6-dev | For toolchain/installation stuff see #perl6-toolchain | For MoarVM see #moarvm
Set by Zoffix on 27 July 2018.
01:00 ufobat_ joined 01:03 ufobat left 03:05 gerry83 joined 03:07 gerry83 left 05:37 robertle left
Geth ¦ problem-solving: AlexDaniel assigned to jnthn Issue What's up with hyper/race semantics? github.com/perl6/problem-solving/issues/77 07:35
¦ problem-solving: AlexDaniel self-assigned Make Rakudo read from META6 modules with :ver<>, :auth<> and :api<> github.com/perl6/problem-solving/issues/76 07:43
07:54 robertle joined 08:34 go|dfish left
AlexDaniel c: all say 42 08:37
committable6 AlexDaniel, ¦all (61 commits): «42␤» 08:38
AlexDaniel heh… it has to run it 61 times
10:21 ilogger2 joined 10:25 |Tux| joined
Geth rakudo/p6o-setup: 4 commits pushed by (Jonathan Worthington)++ 10:49
jnthn (Just a rebase) 10:50
12:21 greppable6 joined
Geth nqp/nested_stagestats: 4fb76beeca | (Timo Paulssen)++ | src/HLL/Compiler.nqp
output "stage parse" later (for nested output), also flush stderr every line
 12:41
12:57 lucasb joined 13:29 ggoebel joined 13:40 [Coke] joined, [Coke] left, [Coke] joined
jdv79 what happened to the irc weblog? 13:48
lizmat it moved ? 13:50
[Coke] colabti.org/irclogger/irclogger_log/perl6-dev can't be reached 13:51
(from the channel title)
vrurg Confirm, colabti is down.
lizmat confirmed here as well
[Coke] tries to do a rakudobrew build and it dies. 13:56
===SORRY!===
No active backends found. Please, report this bug!
13:56 pamplemousse joined
vrurg [Coke]: I don't remember if rakudobrew reports the command line it is using for Configure. Does it? 13:58
jdv79 anyone know who runs collabti? 14:00
[Coke] blowing away ~/.rakudobrew and trying again. 14:06
vrurg jdv79: from whois: [email@hidden.address] 14:07
that's as much as I can find.
jdv79 does that email actually work? 14:10
haha
[Coke] that did it, building now 14:11
jdv79 i emailed...
vrurg jdv79: It's what gandi provides if privacy mode is on by the client
14:14 pamplemousse left 14:16 pamplemousse joined
jdv79 seems back up 14:36
[Coke] hurm. installed rakudobrew and it was fine, now in another shell, perl6 not found (there is rakudobrew init code in my .bashrc) 14:50
|Tux| Rakudo version 2019.07-91-gdcf8c16d4 - MoarVM version 2019.07-35-g2911ec7e4
csv-ip5xs0.686 - 0.735
csv-ip5xs-205.184 - 5.334
csv-parser21.781 - 22.602
csv-test-xs-200.426 - 0.435
test6.877 - 7.131
test-t1.686 - 1.699
test-t --race0.783 - 0.792
test-t-2029.342 - 30.074
test-t-20 --race8.970 - 9.160
 15:03
15:07 pamplemousse left
[Coke] gives up on rakudobrew. 15:15
[Coke] apologizes for blabbing in -dev instead of #perl6 15:21
samcv AlexDaniel, is there a release branch for nqp 15:25
15:26 pamplemousse joined
samcv ah i found it on github. not sure why my local git wasn't fetching it 15:30
16:07 go|dfish joined
vrurg What may cause thred scheduler to lock up to the state where even Promise.in() isn't fired? 16:50
jnthn Exhausting the pool?
(e.g. all threads are busy, or blocked in a non-cooperative way) 16:51
vrurg Ok, so it might be my problem after all.. Thanks, will be tracking it down. 16:52
ugexe that should be easily testable with RAKUDO_MAX_THREADS 17:00
vrurg ugexe: I created a new schedule instance with max_threads set. Not fixable, threads are being leaked somewhere. 17:01
*scheduler
jnthn RAKUDO_SCHEDULER_DEBUG=1 in the environment may also help 17:05
vrurg jnthn: Not much helpful. The pool gets exhausted - that's what I see. But I can't track down what consumes and doesn't release afterwards. Perhaps RAKUDO_SCHEDULER_DEBUG_STATUS would be more helpful on this. 17:11
jnthn Maybe, or just run it under the debugger and hit pause and see what's on the stack :) 17:12
vrurg BTW, perhaps this kind of situation must be a warning. Or a callback on pool scheduler. Somehow the end user must know what happens.
jnthn: I had problems with debugger. But the time has come to do something about it. 17:13
ugexe you can get deadlock without threads too fyi -- github.com/rakudo/rakudo/issues/3080 17:14
loop { run("xxx") } for instance
jnthn vrurg: Grab latest versions of things, including MoarVM HEAD, where I fixed some issues with heavily threaded programs using it (also some profiler and coverage crashes in threaded programs had the same root cause) 17:17
vrurg jnthn: Good news and good idea. Will try. 17:18
jnthn I guess if one exhausts the thread pool and the heuristic deadlock detection hits in and can't create another thread, we could consider a warning.
vrurg jnthn: does it include a fix for R#2898? 17:19
synopsebot R#2898 [open]: github.com/rakudo/rakudo/issues/2898 DESTROY is not called in a multithreaded environment.
jnthn No
That's an unrelated/different problem
vrurg Yep, just somehow related. ;) Ok, doesn't matter.
jnthn And in some senses a bit of a difficult one :)
Well, if we consider every problem that involves threads related, then yes, it is. ;) 17:20
I guess in general we can't know if the program is making poor use of threads but will complete, or if it's doomed, though I figure the deadlock heuristic probably captures the latter 17:21
And if it triggers on a program that does work, there's probably a better way to have written that
We can try it, anyway
vrurg ugexe: thanks for the link, BTW 17:22
jnthn dinner & 17:29
vrurg App::MoarVM::Debug seemingly not compatible with the HEAD. Still expects a bash script. 17:30
timotimo oh no, does it do that in the tests or something? 17:52
oh, it doesn't have tests 17:53
vrurg timotimo: it was about the README on github. ;)
I can't install it anyway because: ===> Fetching [FAIL]: App::MoarVM::Debug:ver<0.0.1> from [email@hidden.address] 17:54
timotimo can you clone manually from that url? do i have to change it to be an https url or something? 17:55
vrurg timotimo: https is more reliable, I think. I tried cloning, but then I get: Aborting due to fetch failure: Data::MessagePack:ver<0.1.2> (use --force-fetch to override) 17:57
timotimo oh, huh?
that's worse, that's not under my control
vrurg I would install them one-by-one when done trying intelliJ with the latest comma plugin. 17:58
Or won't need to if the plugin works. ;) 17:59
timotimo :) 18:01
vrurg Arrgh... comma CT almost works. But I can't create a new project! (facepalm) :( 18:03
timotimo is there an error? can you "report it to the edument team"? 18:09
vrurg timotimo: nothing at all. It was ignoring any close/new/open project . 18:12
vrurg wiped out all settings and starting from scratch.
18:13 pamplemousse left
timotimo o_O 18:13
vrurg timotimo: do you know where to get url of the standard plugin marketplace? 18:14
timotimo you'd like to add plugins to the comma standalon community edition? 18:15
vrurg timotimo: I'd like to get vim mode. ;) 18:19
timotimo it's certainly the most requested plugin 18:20
currently there is no support, but i forgot why
vrurg timotimo: I had it installed with a previous version. And it worked in 2019.07 until I deleted all settings. Now I have blank link of plugins. 18:23
It doesn't see marketplace, I guess. 18:24
timotimo yeah, sorry, i dunno :( 18:25
when you use comma as a plugin, it should definitely work with other plugins, but surely the plugin version of comma has some other limitations somewhere? 18:29
vrurg Comma suffers from the same problem I'm trying to debug in my code: java.util.concurrent.RejectedExecutionException: Thread limit exceeded replacing blocked worker :D 18:35
.ask AlexDaniel can you run blin against github.com/rakudo/rakudo/pull/3081? 18:58
yoleaux vrurg: I'll pass your message to AlexDaniel.
19:07 chloekek joined
vrurg timotimo: moar-remote finally works for me. How can I find out file and line number for frames which are looking like: /Users/vrurg/src/Perl6/Perl6-As... 85 chars, 1 lines, key: observandum /Users/vrurg/src/Perl6/BO-Trading/li... 131 chars, 1 lines, key: tairn 19:18
timotimo i'm not sure what you're pasting there 19:20
but usually to get the right paths i'll start the moar process that is being debugged with the env var MDS_PROTOCOL set; in the future the debugserver will have some way to report what exactly is available 19:21
vrurg timotimo: the paste is from a stacktrace. I cannot see where a frame originates from. 19:26
AlexDaniel . 20:50
yoleaux 18:58Z <vrurg> AlexDaniel: can you run blin against github.com/rakudo/rakudo/pull/3081?
AlexDaniel vrurg: not yet. Currently it only works for commits on master
I felt like fixing that yesterday but didn't get to it :) 20:51
21:28 chloekek left
samcv 2019.07.1 does not pass against roast 21:52
AlexDaniel, i could release but want to confirm that's okay 21:55
lizmat samcv: what fails ?
AlexDaniel samcv: try agains rakudo-release-2019.07 branch in roast
t
samcv results: gist.githubusercontent.com/samcv/7...tfile1.txt 21:56
AlexDaniel yeah that's on master
lizmat AlexDaniel: why wasn't that merged yet ?
samcv any way to get a list of all branches on a remote?
AlexDaniel lizmat: it's 0 commits ahead, there's nothing to merge 21:57
timotimo "git branch -a" should show everything
oh, on *a* remote
samcv timotimo, thanks!
timotimo "git remote show origin" will do that
lizmat AlexDaniel: then why should samcv test against a branch ? 21:58
samcv i've known how to fetch if if if i know the name. but otherwise no
AlexDaniel lizmat: because every past release only passes roast at the time of the release. After the release people add more tests that can't pass on older rakudo versions 21:59
samcv well i was on roast master and it didn't pass. unless i'm wrong
AlexDaniel well, “at the time of the release” is conveniently marked as rakudo-release-2019.07 branch :) 22:00
basically, master is a moving target, you can't make a release that will work for tests that don't exist yet 22:01
the only issue here is that we never marked which roast commit worked with which release 22:02
samcv AlexDaniel, good point. not sure if we need a roast branch too. even if just a matter of formality 22:05
it could make testing previous releases easier 22:06
AlexDaniel this is the list that I have: gist.github.com/AlexDaniel/b0b85da...05c3509939 22:11
you can get roast commits for other releases by looking at the date, I think 22:13
± a few commits
I have these because they were released from branches, so I had to know which commit I'm testing against
22:14 BeastieBot joined 22:16 BeastieBot left, BeastieBot joined, BeastieBot left 22:17 BeastieBot joined
AlexDaniel samcv: does it pass now? 22:37
vrurg timotimo: github.com/edumentab/p6-app-moarvm-debug/pull/7 – it's easier to implement on my own then to explain what's wrong. ;) 23:00
timotimo cool 23:01
vrurg it's more like a quick hack, but it all looks like a one big hack to me. ;) 23:02
timotimo it was quickly done, yeah :(
no overarching plan to how the commands should look
the help page is like multiple screens full, which is also bad
vrurg ah, whatever. There're bigger problems than a hacky debugger. 23:03
timotimo honestly, now that i'm doing a bit more with the gtk bindings ... 23:04
no promises!
did you see the gtk/cairo livecoding toy?
vrurg Hm... I heard nothing interesting here...
timotimo media.discordapp.net/attachments/5...height=714 23:05
vrurg No, I don't play with GUIs usually.
timotimo it's probably a bit easier to get tab completion or something like it to work in a gui than it is in the terminal
vrurg Something like this I saw recently at an Apple Store. ;)
timotimo because seriously F having to deal with all the terminal oddities
vrurg Besides, that's where I love GUIs: when all your debug info is sitting right beside the source and the output. No need to jump back and forth between windows. 23:06
timotimo yeah 23:07
.o( no need to jump back and froth over windows )
vrurg One more thing: github.com/edumentab/p6-app-moarvm-debug/pull/8 23:09
vrurg is back to catching gone away threads... 23:10
timotimo thank you for your contributions
i hope the changed url makes it easier to install
vrurg np. I hope for that too. 23:15
timotimo though making cpan releases would also be good 23:16
23:23 evalable6 joined
timotimo funnest fact: the application with the screenshot there crashed soon after and didn't save my changes so i'll have to OCR the code or something 23:28
and i was too lazy to do that so far, so the project it's attached to has kind of fallen asleep

Powered by Cro and the Raku® Programming Language.

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

Thank you!