01:41 evalable6 left, linkable6 left 01:42 linkable6 joined 01:44 evalable6 joined 02:26 SmokeMachine left, SmokeMachine joined 02:53 camelCaser joined 02:54 ccamel left 06:37 sena_kun joined 06:47 frost-lab joined 07:04 leont left 09:08 Altai-man joined 09:11 sena_kun left 09:27 frost-lab left 09:45 domidumont joined
lizmat cog: I do not know of a way to prevent having to do :!c every single time 10:07
bisectable6: sub foo() { loop { LAST {say "LAST"}; say "IN"; last } }; foo 10:08
bisectable6 lizmat, Will bisect the whole range automagically because no endpoints were provided, hang tight
lizmat, Output on all releases: gist.github.com/bd0fea04b02f860f17...7ea001e40e
lizmat, Bisecting by exit code (old=2018.12 new=2016.01.1). Old exit code: 1
lizmat, bisect log: gist.github.com/8ad508434235c037e6...507960191b
lizmat, Bisecting by output (old=2015.12 new=2016.01.1) because on both starting points the exit code is 0
lizmat, Output on all releases and bisected commits: gist.github.com/d5a3b77b5f1ae9a3aa...625d4fe5aa
10:10 domidumont left 10:25 frost-lab joined 10:49 frost-lab left
nine releasable6: status 11:03
releasable6 nine, Next release in ≈7 hours. 1 blocker. 145 out of 146 commits logged
nine, Details: gist.github.com/639661186a2204da59...315b2f94e7
Geth rakudo: 4e645ee5da | (Elizabeth Mattijsen)++ | src/core.c/Any.pm6
Actually support running dd on native BOOT arrays

It actually didn't work for BOOT(Int|Num|Str)Array up till now.
lizmat Altai-man ^^ does not need to be in the release :-)
Altai-man lizmat, thanks for noting! The commit with typo fix will be the last included, I think. 12:22
lizmat Altai-man++
12:27 linkable6 left, evalable6 left, evalable6 joined 12:28 linkable6 joined 13:09 sena_kun joined 13:11 Altai-man left 13:17 leont joined
sena_kun Release time. 13:23
Geth rakudo/workaround_generics_slowdown_in_native_call: 026ab726db | (Stefan Seifert)++ | lib/NativeCall/Types.rakumod
NativeCall workaround for slowdown of calls to routines with generic arguments

If a routine's signature contains any generics (::T), we generate code to check at runtime, if the passed type is coercive. This is very expensive and costs some 70 % of performance in some benchmarks. For NativeCall there is a workaround though: the only generic parameter is
  ::?CLASS in the signature. Since the typed CArray roles are only known within
the CArray class, we can be sure that ::?CLASS is actually CArray or maybe a subclass. Thus we can make the type constraint more concrete. Brings whitenoise.p6 backup up from 45ish fps to > 180 fps
sena_kun I wonder what's with this comma in "The next release of Rakudo (#143), is tentatively scheduled for 2021-01-23", it feels so-o out of place somehow. 13:37
The script is 5 years old, though, and nobody bothered, it seems. 13:38
MasterDuke nine++, whitenoise.p6 is back to up ~240 fps on that branch 13:39
sena_kun Wow, those are awesome news!
Geth rakudo/release-2020.12: c3953d8d9b | Altai-man++ | 3 files
Update changelog + announcement

Deliberately not logged:
a375597 6843f80 604f64f 1b26983 8eb9db3 c63f078 04a5001 214df49 1cd8765 96ab6eb f7ab4f5 737560e 2e33736 1aff2ad b93d9b3 551b679 f2ad038 2a44729 e90ed75 22c4bb5 1ccfb09 3202466 4e21ada 3d8e942 734631b 260f015 a89d020 cce2fa2 436437b 1ae0478 ac84c18 79b55ac 5e791d7 1639c78 3808eaa 69f3e95 ce45754 3d46341
sena_kun It is the last chance to glance and tweak something in the announcement/changelog. (yes, the dates about 2021 are checked) 13:46
Don't tell me rakudo on master does not pass roast... 14:07
So it was a flapper in github.com/Raku/roast/blob/master/...s/native.t 14:10
14:17 linkable6 left, evalable6 left 14:19 linkable6 joined 14:20 evalable6 joined
sena_kun releasable6, status 14:27
releasable6 sena_kun, Next release in ≈4 hours. 1 blocker. 145 out of 147 commits logged
sena_kun, Details: gist.github.com/efebde3c56b199486e...7b9fc3b9d8
sena_kun lizmat, you around? 14:32
lizmat am for about 5 mins 14:40
sena_kun ^^
sena_kun lizmat, how it comes I have rakudo master and roast consistently fails 20 tests in S09-typed-arrays/native-int.rakudo.moar? 14:41
Is something not pushed to roast?
lizmat stashes what she was working on and builds 14:42
sena_kun So far this is a royal blocker for us. 14:43
lizmat there's quite a few todo's in there, but it doesn't fail for me 14:46
All tests successful. 14:47
Files=1, Tests=1833, 3 wallclock secs ( 0.15 usr 0.01 sys + 4.40 cusr 0.28 csys = 4.84 CPU)
sena_kun gist.github.com/Altai-man/f2d4e419...b20cdfd554 <- here is my log
Sigh, this is disturbing.
# Expected: Cannot delete from a natively typed array 14:48
# Got: Cannot delete from a native int array
lizmat and roast does not have an unpushed changes for me
sena_kun There was a commit to native-num making the change &
lizmat that error message *did* change in the past month and I did change that
sena_kun s/&/^/
But not for native-int.t
lizmat so maybe it's missing from errata ?
sena_kun github.com/Raku/roast/commits/mast...tive-int.t <- there are no commits?
lizmat, I am on master history for the file and last commit is dated 23 Oct, am I insane? 14:49
lizmat last one I see is b97dfa098fd2dd8
linkable6 (2020-12-17) github.com/Raku/roast/commit/b97dfa098f [JVM] Unfudge some now working tests
lizmat from 17 Dec
sena_kun See the URL I posted, github.com/Raku/roast/commits/mast...tive-int.t
lizmat dinner is served&
sena_kun investigates 14:50
lizmat will be back in 20 mins or so 14:51
sena_kun Hmm, master passes for me when doing `make stresstest`... 14:57
t/spec/S09-typed-arrays/native-int.rakudo.moar ................ Dubious, test returned 20 (wstat 5120, 0x1400) 14:58
Failed 20/1370 subtests
Ok, so I have no idea, but there is indeed a difference between master and errata. 15:05
The question is how we did not notice it in the last release.
What on Earth. 15:07
lizmat sena_kun: is there something I can do ? 15:12
sena_kun lizmat, can you try 6.c-errata roast test? 15:13
If it fails for you in the same place, maybe you'll be able to guess what it misses, because I am clearly not. 15:14
lizmat gimme a min :-)
sena_kun I restarted the process from scratch for now.
15:16 lucasb joined
lizmat running spectest now 15:18
sena_kun: can confirm breakage on native-num and native-int tests on 6-c.errata 15:20
lizmat looks deeper 15:21
sena_kun lizmat, native-num is fixed by a commit
one second...
lizmat, github.com/Raku/roast/commit/1a84e...61e4bd560b <- 15:22
but there is no such commit for native-int.t
lizmat well, I guess that needs to be applied to 6.c-errata then
sena_kun So the native-num will be fixed with a cherry-pick of the commit I posted, I just did not push it, but there is nothing to cherry-pick for native-int.t one.
lizmat hmmm... 15:23
perhaps 7495dca5755ac78 ? 15:24
linkable6 (2015-03-15) github.com/Raku/roast/commit/7495dca575 Split off native int|num specific tests
lizmat ah no
sena_kun Still the same after a full rebuild of everything. 15:26
lizmat, will you push cherry-picked 1a84eefa4734019923f7ec2db0639f61e4bd560b or should I? 15:27
linkable6 (2020-12-05) github.com/Raku/roast/commit/1a84eefa47 Fix up more exact error message
lizmat I'll push a single commit with both, ok ?
sena_kun lizmat, what is the second out of this "both"? 15:28
lizmat the num case ?
sena_kun Well, the first then?
Do you have a solution for native-int?
lizmat yeah, it's basically the same fix, with s/num/int/ 15:29
sena_kun Can it be broken solely when `use v6.c` is applied?
I mean, for tests to be correct and for rakudo to behave differently under different pragmas.
lizmat those changes were in c
as far as I remember
ah, and they should be marked as todo I guess/ 15:30
sena_kun How could we release for a whole year with a test which was passing and now without any changes suddenly stopped?
Geth roast/6.c-errata: 2ef8cc4092 | (Elizabeth Mattijsen)++ | 2 files
Fix tests because of error message becoming more specific
lizmat sena_kun: I currently have no answer to your question 15:33
I only see that since 2020.11 *those* error messages have changed
(pretty sure there is no text for the native str case) 15:34
and I *did* the adapt the tests in roast, but forgot to do it in 6.c-errata
could that be an explanation ? 15:35
sena_kun lizmat, tests were adapted for num case only, not for int case, but on master both pass. I am not blaming or demanding right now, just very curious and trying to guess if we are in trouble.
I mean, yes, it is a common case something is changes on roast master and 6.c-errata/6.d-errata needs cherry-picking. I do it just fine, no sweat. 15:36
But here there was nothing to pick.
sena_kun digs into git history more
lizmat maybe vrurg's merge somehow ate a commit? 15:37
sena_kun bisectable6, uint.^name.say 15:39
bisectable6 sena_kun, Will bisect the whole range automagically because no endpoints were provided, hang tight
sena_kun, ¦6c (49 commits): «uint␤»
sena_kun, Nothing to bisect!
15:40 domidumont joined
lizmat sena_kun: I just noticed that those native int tests seem to have disappeared in roast on master 15:43
sena_kun It seems so. 15:44
Ok, I give up. 15:45
lizmat, sorry for all the commotion.
lizmat sena_kun: can you continue now? 15:47
sena_kun lizmat, yeah. I guess 6.d will want the same treatment, but if it's about a simple patch, I can do it myself if you want. 15:48
lizmat sure... then I will be going afk again&
sena_kun This is the first time in a year when things have diverged so much between branches. 15:49
lizmat, have fun.
lizmat thanks!
15:53 domidumont left
[Tux] Rakudo v2020.11-146-g604f64f46 (v6.d) on MoarVM 2020.11-89-g0323c7bce
csv-ip5xs0.843 - 0.883
csv-ip5xs-208.191 - 8.536
csv-parser27.981 - 28.488
csv-test-xs-200.390 - 0.412
test8.187 - 8.384
test-t1.930 - 2.058
test-t --race0.983 - 0.991
test-t-2032.697 - 33.998
test-t-20 --race10.133 - 10.360
16:06 Xliff joined
Geth nqp: 14bba20011 | Altai-man++ | tools/templates/MOAR_REVISION
[release] Bump MoarVM revision to 2020.12
nqp: 545e8c891b | Altai-man++ | VERSION
[release] Bump VERSION to 2020.12
rakudo/release-2020.12: ff53954fa5 | Altai-man++ | tools/templates/NQP_REVISION
[release] Bump NQP revision to 2020.12
rakudo/release-2020.12: 45b4e6464d | Altai-man++ | VERSION
[release] Bump VERSION to 2020.12
rakudo: Altai-man++ created pull request #4126:
Release 2020.12
rakudo/master: 4 commits pushed by Altai-man++ 16:13
sena_kun I guess this is it. 16:14
eeeeh 16:18
note twitter.com/koto_san_kana/status/1...0388462596
weekly twitter.com/koto_san_kana/status/1...0388462596 16:19
notable6, help
notable6 sena_kun, Like this: notable6: weekly rakudo is now 10x as fast # See wiki for more examples: github.com/Raku/whateverable/wiki/Notable
sena_kun notable6: weekly twitter.com/koto_san_kana/status/1...0388462596 16:20
notable6 sena_kun, Noted! (weekly)
AlexDaniel` Congrats for the release! Now is probably the best time, here is the announcement. I am stopping maintenance of the whateverable project, including hosting of the bots and builds. I know the installation instructions are a bit off, but I'm blocked from editing the wiki. It's a project I've put a lot of love into, and this love keeps pulling me back. So, for everyone's interest, that is it. I'll keep them up until 27th 16:40
just in case, and the data won't be wiped for some time after that. Let me know if there's anything else I can do for you.
sena_kun AlexDaniel`, hi. How much does it cost to run? 16:57
AlexDaniel` in terms of what?
sena_kun AlexDaniel`, in terms of $$$.
Though I have no idea how automated is it. 16:58
AlexDaniel` I'm not sure I understand the question. I am paying 32€/month for the server, but it's my server that also runs other stuff.
if you have another idle machine that can do the work then you'll be fine 16:59
there's also a 6lang.org domain that is used for serving the builds, obviously doesn't cost much, but you're probably much better off rewiring it through whateverable.raku.org or something 17:00
currently it is using 30GB of storage, that's not much and you probably have more than that available 17:01
sena_kun AlexDaniel`, maybe I don't know enough to make a nice one. I would like to keep whateverable alive, but not sure what resources (time, money, code) are needed to do that and what your "maintenance ends here" implies in practice.
AlexDaniel` ah I see
well, it implies the bots will be turned off, which also means shareable will be off, which also means Blin won't work 17:02
basically I want it off my server
this should've been done much earlier, but there was always an issue of finding a place to host it 17:03
sena_kun AlexDaniel`, that's understandable. I guess my way is to go through instructions (there are some) and host something myself, then re-wire URLs of bots, right-o?
AlexDaniel` and honestly there wasn't any reason to do that, I think I did pretty good job of doing that for the last 4 years or whatever 17:04
sena_kun: yes. There's a PR for a dockerfile, but it is so-so. If I was doing it today I'd write a good one, then also a docker-compose file to run every bot separately 17:05
sena_kun AlexDaniel`, you truly did, I cherish your passion and imagination to do what you have done over all those years.
AlexDaniel` thank you
sena_kun AlexDaniel`, roger. Thanks, I'll look into it ASAP, too beaten now after 3 releases in a row. 17:06
AlexDaniel` ideally, the build.p6 script should run in a container with a much older GLIBC version, that would solve some problems that people were having when trying to use the builds
I didn't know how to do any of that back in the day so I didn't
I'd also scrap the idea of using systemd services, just go fully with docker 17:07
I also have an uncommitted version that uses zstd instead of lrzip and has a separate bot instead of a build.p6 script
I didn't finish it because my flow was interrupted when I tried to get Cro to consume github webhook payloads… 17:08
17:08 Altai-man joined
AlexDaniel` but that's something you probably know how to do, so I can send you a patch with what I had, if you're interested. It's unfinished and needs work though 17:08
Altai-man "Most data files are available here: github.com/perl6/whateverable-data", but no such repo does not exist. : O 17:10
AlexDaniel` but the upcoming features were supposed to be: faster and more tightly packed long-term storage, webhooks for triggering builds
17:11 sena_kun left
AlexDaniel` Altai-man: forget the repo, it unfortunately didn't happen. There are just three files you need: notes tellable-tell tellable-seen 17:11
that's for notable and tellable, everything else is mostly stateless
well, reportable we can probably ignore. 17:12
as for the builds, as I said, if I was doing it today I'd let the server crunch through them again so that there's no GLIBC issue
Altai-man: try `ssh [email@hidden.address] it should work for you 17:14
Altai-man yup
I see the repo.
17:15 domidumont joined
AlexDaniel` there are no valuable unstaged changes, most of the data/ is generated automatically except for the important files I mentioned previously 17:15
please do copy config.json though 17:16
again, yes running them the same way is probably OK but docker will make you so much happier :) 17:18
Altai-man AlexDaniel`, can I copy it via curl pastie or it has sensitive data inside? "default-stdin" is full of garbage when doing plain cat here.
AlexDaniel` Altai-man: have you tried sshfs?
Altai-man not really. :)
AlexDaniel` Altai-man: sshfs [email@hidden.address] ~/sshfs/bisectable/ 17:19
need to make sshfs/bisectable locally, or whatever you want to call it
and it'll mount the whole thing
keep in mind though that you won't be able to copy all of the builds this way, because that'd take just too much time over sshfs. But configs and most other files work great!
and yes it does contain access tokens and stuff 17:20
now that you say it, there's this github account: gist.github.com/Whateverable/
but let's do it step by step, first get the dockerized stuff to run, then we'll figure out the accounts
MasterDuke you think it'd take too long over sshfs? i've done large downloads with sshfs before, i didn't think it was particularly slow 17:21
AlexDaniel` it's not so much about the size, it's the amount of files :)
Altai-man MasterDuke, no real reason to, I think.
MasterDuke pipe through tar or rsync 17:22
AlexDaniel` yes, but that's no longer sshfs is it? :) that's what I'm saying
maybe today it's not that bad, reportable used to store everything in files but then I turned it off at some point
MasterDuke eh, still easier to use "local" commands 17:23
Altai-man AlexDaniel`, I copied a bunch of text files plus three files from data/ (though they will be outdated on replacement day, I'll try again, though). Going afk now, sorry. 17:24
AlexDaniel` it's cool, let me know if you need more time or whatever
we just have to time in between releases so that Blin is unaffected
17:24 b2gills left
Altai-man AlexDaniel`, what server specs do you recommend to rent? 17:28
AlexDaniel` SSD for storage will make a big difference I thin
k 17:29
Altai-man Digitalocean offers only SSD storages.
AlexDaniel` great! How much RAM can you get?
Altai-man www.digitalocean.com/pricing/calculator/
I can get whatever is needed. :)
AlexDaniel` 32G of RAM would be nice. Rakudo is definitely not forgiving in terms of RAM 17:30
Altai-man Hmm, those are not 30 euro. :S 17:31
AlexDaniel` yeah forget digitalocean
Altai-man AlexDaniel`, another option?
AlexDaniel` E3-SSD-1-32 from here: www.soyoustart.com/ie/essential-servers/ 17:32
Altai-man AlexDaniel`, thank you! 17:33
AlexDaniel` this is what I have from them: E3-SSD-1-16 Server - E3-1225v2 - 16GB - SoftRaid 3x120GB SSD
although I'm pretty confident that they didn't have 1225v2 and put 1245v2 into my server 17:34
yay free upgrade
they bumped the price of mine to 32€/month even though it was 30€/month for a long time
if dockerized properly, you'll be able to use the same machine for all kinds of other stuff for Raku 17:40
say crunching roast on PRs or whatever you want
most of the time whateverable is not doing much, although that is of course different if you're creating all the builds from scratch. Expect it to finish in a month or so, but IIRC it's easy to make it build the most recent commits first (after the tags), this way it'll be functional right away and historical stuff will appear over time 17:41
17:45 b2gills joined
AlexDaniel` having dockerized bots actually makes me pretty excited 17:48
you should be able to lift the private message limit they currently have
nine Altai-man: I still got plenty of free resources on my server 18:07
18:27 leont left
cog lizmat, thx about :!c as a default 18:33
18:36 domidumont left
Geth rakudo: 357a7f300d | (Elizabeth Mattijsen)++ (committed using GitHub Web editor) | docs/announce/2020.12.md
Update 2020.12.md
bartolin Altai-man lizmat: regarding the tests in S09-typed-arrays/native-int.t: If I'm not mistaken the specific error message (that has changed recently) has been removed back in 2017 by Zoffix: github.com/Raku/roast/commit/20fe657466 18:50
It still was in 6.c-errata, so that needed a fix. 18:52
the only strange part was the log from gist.github.com/Altai-man/f2d4e419...b20cdfd554
but since there are 1833 tests nowadays in S09-typed-arrays/native-int.t it looks like a different version of the file was run. Maybe there were unstaged changes and the version from 6.c-errata was run? 18:55
Geth ¦ problem-solving: shishini assigned to JJ Issue Add a forum to support the Raku community and provide a searchable repository of Q&A github.com/Raku/problem-solving/issues/251 18:56
bartolin maybe you just changed to master and re-run with the preprocessed version generated from the 6.c-errata branch (native-int.rakudo.moar)? 18:57
however, I think we are fine now
AlexDaniel` Altai-man, nine : ideally, shouldn't rba be involved? They were doing a great job at managing community resources, last time I checked 19:22
also [Coke] did suggest that there can be funding for whateverable-related machines. I refused it in the past but it doesn't mean that you should :) 19:23
Geth ¦ problem-solving: JJ assigned to rba Issue Add a forum to support the Raku community and provide a searchable repository of Q&A github.com/Raku/problem-solving/issues/251 19:55
21:09 sena_kun joined 21:10 Altai-man left 22:24 sena_kun left 23:06 Xliff left 23:07 Xliff joined 23:16 lucasb left 23:36 Xliff left