00:07 pmurias left 01:14 Xliff left 01:28 lucasb left 02:15 ggoebel left 04:58 vrurg left 05:02 vrurg joined 06:05 jmerelo joined 07:06 jmerelo left 07:59 sena_kun joined 08:09 donaldh joined
El_Che AlexDaniel: I am uploading the container, it's huge to the zillion deps 08:34
I am off, butI'll leave my pc on :)
AlexDaniel: should work now: github.com/nxadm/raku-blin-docker (the dblin script) 08:42
AlexDaniel El_Che: “Images will be automatically built from the master branch with the "latest" tag” 08:43
El_Che: there's no need for that?
ahhhh sorry, misunderstood
nevermind :) 08:44
El_Che: any reason not to merge it into Blin repo itself? 08:45
09:17 ggoebel joined 09:56 Altai-man_ joined 09:59 sena_kun left 10:02 MasterDuke left
El_Che just to see ifbitbworks, not tobstep on any toes, easy upload tobdocker hub 10:28
if it works, can we merge it if you want to
mobile typing ;) 10:29
10:42 rba[m] left, timotimo[m] left, tyil[m] left, AlexDaniel` left 11:03 rba[m] joined 11:14 donaldh left 11:19 AlexDaniel` joined, tyil[m] joined, timotimo[m] joined, pmurias joined 11:23 squashable6 left 11:25 squashable6 joined 11:57 sena_kun joined 11:58 Altai-man_ left 12:03 ExtraCrispy joined 12:18 pmurias left 12:39 pmurias joined 13:12 lucasb joined 13:56 Altai-man_ joined 13:58 sena_kun left
tony-o there is a nightly built already on docker if thats what youre looking for El_Che 14:24
tellable6 2019-10-30T12:59:36Z #raku <tbrowder> tony-o thanks for the links
tony-o iirc its around 600M 14:25
15:26 vrurg left
El_Che (on mobile) It's for Blin and a zillion binary dependencies 15:28
tony-o ah 15:47
15:58 Altai-man_ left 16:06 Kaiepi left 16:12 Kaiepi joined 16:18 pmurias left 16:23 jmerelo joined 16:50 vrurg joined 16:52 ggoebel left
tbrowder :tony-o link to using unique constraint in Xoos, Xoos syntax fits my style of coding, so i look forward to playing with it 16:57
tellable6 2019-10-28T17:41:18Z #raku <tony-o> tbrowder an example can be found here github.com/tony-o/perl6-xoo/blob/m...und-keys.t
2019-10-30T16:40:56Z #raku <tony-o> tbrowder i assume you mean in the READMEs, no problem, thank you for writing those mods
tony-o .tbrowder happy to hear it, if you run into issues i'm happy to fix/help 16:59
discord6 <kelly> is nxadm.github.io/rakudo-pkg/ safe for installing raku 17:08
17:14 MasterDuke joined
jmerelo Raku container (github) action just published to the Github Actions marketplace: github.com/marketplace/actions/rak...iner-actio 17:22
tellable6 2019-10-29T19:45:15Z #raku-dev <rba> jmerelo: Not sure if I can now ask for the double.
jmerelo .tell rba no problem, we'll make do.
tellable6 jmerelo, I'll pass your message to rba
tyil github.com/perl6/nqp/blob/master/d..._guide.pod speaks about a `make release` step to create a tarball with all things in it, however, there's no makefile 17:27
tony-o jmerelo: was something missing from the rakudo nightly you wanted? 17:30
17:30 AlexDaniel left
jmerelo tony-o: sorry, you lost me here... 17:30
17:30 AlexDaniel joined, AlexDaniel left, AlexDaniel joined
tony-o cloud.docker.com/u/tonyodell/repos...do-nightly 17:31
vrurg tyil: what do you mean 'there is no makefile'?
[Coke] tyil ;the makefile was removed at some point.
er, the *build* doesn't use it anymore, I mean.
so it's possible other things that relate to it are no longer functional.
tony-o and: deathbyperl6.com/rakudo-nightly-faster-ci/
has instructions for travis & circle
tyil vrurg: well, just like that, there's no Makefile 17:32
[Coke]: is there any documentation on how to create a tarball for it now?
because I need one in order to build r*
jmerelo tony-o: cool. Still can't remember when I wanted that, though :-). Sorry. 17:33
[Coke] wait, in p6-doc?
there *is* a Makefile in perl6-doc
tyil what
[Coke] it's just not used to build the site anymore.
tyil no, nqp
[Coke] checks again
vrurg tyil: step3, perl Configure.pl creates the makefile
tyil why does it start with a make step if there's no makefile at that point 17:34
tony-o jmerelo: ah, i think i misunderstood what your container was for, i thought it was for testing raku modules in general but it's for testing your module specifically?
tyil I mean, thanks for the help, I really appreciate it, but documentation for releasing the parts of raku is quite bad
[Coke] is looking at github.com/perl6/doc/blob/master/Makefile 17:35
tyil I'm not sure why
I've said multiple times I'm looking at NQP 17:36
vrurg tyil: 'make realclean' is just to make sure everything is wiped out. If it fails – it's ok
jmerelo tony-o: ah, OK. No, there's this (relatively new) thing called GitHub actions. It needs to get its Dockerfiles published in some specific places (not Docker Hub).
tyil vrurg: something to update in the docs once I get this stuff working then
tony-o i don't mind extending my image to github actions if that's helpful to you
so we don't all have to maintain docker images of rakudo 17:37
vrurg tyil: the release guide is primarly for release manager. So, it expects people to be familiar with the process in general.
jmerelo tony-o: that's what I have just done; I've used the code for my perl6-test Docker container (based on an Alpine) and put it in a place where it can be used in GitHub actions.
tyil vrurg: but that leaves a massive hole in knowledge if there's no release manager :/
as we see with R* now
jmerelo tony-o: er. The image I've just published is already usable as a GitHub actions. (sign up for the beta), drop a .github/workflows/test.yml file that contains just that, and you've got your Raku doce tested every time you push by GitHub itself. 17:38
tony-o my concern is mainly that we just keep a consistent nightly available everywhere and i'm offering to maintain a nightly on github actions for testing with gh's ci
jmerelo tony-o: Ah, OK, great. 17:39
vrurg tyil: It's not for R* manager but for rakudo release manager. You, besides of R* there could other distros exists, we can't take care of everything. Hence the focus on the particular process.
tony-o are you interested in that or would you like to keep maintaining the gh ci image?
vrurg tyil: R* must have it's own guide for the purpose.
tyil yes, which expects a working NQP dist which I didn't have, so I have to understand at least some parts of the NQP release
jmerelo tony-o: yes, let's do that. Let me check, and I'll PR so that it can be used directly in GH
vrurg tyil: Though I agree that some details in the doc need updating. In particular, openjdk requirement has changed. 17:40
tyil which, it turns out, could use a little polishing, and then literally everyone can make a tarball when needed
[Coke] (docs vs. nqp) whoops. I saw docs in a different context, my bad, apologies.
vrurg tyil: If you provide a PR with doc updates – that'd be graat! :)
tyil perhaps once I get R* building
which can take a while at this speed 17:41
jmerelo tony-o: can you give me the GH URL? Docker hub is now 500'ing
tony-o rakudo nightly repo is here: github.com/tony-o/rakudo-nightly 17:42
vrurg tyil: We're in lack of resources. Not every aspect could be taken care in timely manner.
tyil hence I'm trying to bring some resources to release R*
preferably automated, ofcourse
jmerelo tony-o: I'm checking it out. It'll need a bit of work in the way of instructions. I'll work on that. Thanks! 17:44
vrurg tyil: regarding openjdk, I currently build with openjdk11. Not sure of 12 and 13 will do but should too.
jmerelo El_Che: this is failing app.shippable.com/github/perl6/doc.../1/console Is this temporary? 17:45
vrurg tyil: and good luck! looking forwartd for your success! 17:47
tyil same
jmerelo tyil++
lizmat hmmm looks like the search box on docs.raku.org generates improper links :-( 17:52
hmmm... looks like links like docs.raku.org/language/setbagmix create a 404
rba jmerelo ^^
docs.perl6.org/language/control#in...th_without 17:53
works and
gives a 404
jmerelo lizmat: hum 18:07
lizmat: state works, though docs-stage.raku.org/language/setbagmix 18:08
s/state/stage/ 18:09
lizmat I have no explanation, I guess the error log will tell more (maybe)
jmerelo weird 18:12
lizmat: the files are actually there. 18:16
lizmat yeah, feels like a server config issue
jmerelo .tell rba something weird is going on with docs.raku.org. The files are there, but they return 404
tellable6 jmerelo, I'll pass your message to rba
jmerelo this works: docs.raku.org/language/control.htm...th_without 18:17
It's a site configuration issue. I'll try and see if I can fix it. 18:18
Thanks for the heads up, lizmat
18:21 vrurg left
jmerelo lizmat: docs.rakulang.site/language/5to6-overview 18:28
lizmat: docs.perl6.org/language/control#in...th_without
lizmat that works for me
the second also 18:29
jmerelo docs.raku.org/language/control#ind...th_without
lizmat but that URL feels wrong to e
jmerelo nginx configuration
lizmat: sorry, which one?
lizmat the last one
jmerelo .tell rba it's fixed now.
tellable6 jmerelo, I'll pass your message to rba
jmerelo lizmat: the whole one, or the docs.raku.org part? 18:30
lizmat it repeats the "with_orwith_without" part ?
jmerelo lizmat: ah, that's another piece of cloth...
lizmat docs.perl6.org/language/control#wi...th_without # the perl6.prg version 18:31
discord6 <kelly> hi
jmerelo lizmat: it also works: docs.raku.org/language/control#wit...th_without
hi, +kelly! 18:32
lizmat if I click from the search results menu, I get docs.raku.org/language/control#ind...th_without
is there a reason for that ?
jmerelo lizmat: ... probably not? 18:33
18:33 sena_kun joined
lizmat :-) 18:33
jmerelo lizmat: that's indexed three times, as a matter of fact
lizmat ah? 18:34
rba jmerelo: I see what you changed. Thank you 18:36
tellable6 2019-10-30T17:22:57Z #raku-dev <jmerelo> rba no problem, we'll make do.
2019-10-30T18:16:29Z #raku-dev <jmerelo> rba something weird is going on with docs.raku.org. The files are there, but they return 404
discord6 <kelly> is raku as other scripting languages(perl5, python) for general use?
jmerelo rba: no problem. I remembered we did something like that back in August
kelly: yes.
lizmat: raised an issue: github.com/perl6/Documentable/issues/55 18:37
rba jmerelo: As I never had logon on the other tmp server, I couldn't know that.
lizmat kelly: yes 18:38
rba is afk for dinner. when pinged back in 30 mins...
jmerelo rba: no problem :-). Bon appetit
discord6 <kelly> Sorry, what i meant was *as fast as other scripting languages? 18:39
18:40 ggoebel joined
sena_kun kelly: from task to task it differs. some code can be faster than perl/python/ruby, some code in general is slower. 18:41
jmerelo kelly your mileage may vary, but it's pretty fast
discord6 <kelly> i'm mostly looking into cro for webservices 18:43
<kelly> however i have not stress tested the server
<kelly> if someone here has experience can tell abt performance related to web side of things 18:45
sena_kun it depends greatly on your setup, caching, database you use... 18:47
discord6 <kelly> but it will comparable to python/perl5 right? if the setup is more or less similar 18:49
nine kelly: that's really hard to answer in a general way. 19:02
19:14 ExtraCrispy left 19:59 sena_kun left
El_Che jmerelo: It seems I cleaned to much on my Fall cleaning, I will fix it now 20:23
jmerelo: shall I rename the repo to raku-doc? 20:25
jmerelo: or better: rakudo-pkg-for-doc? (dl.bintray.com/nxadm/rakudo-pkg-for-doc/...) 20:27
I'll put the old name back and rename it when it's easier for you
jmerelo: sorry for the breakage 20:28
jmerelo: building the doc packagesn ow 20:36
AlexDaniel: were you able to test the Blin image? Does it work as expected?
AlexDaniel El_Che: I haven't tested it yet
El_Che AlexDaniel: the prog does not take switches, like --help. It whould be handy to test it :) 20:37
Good you haven't tested it, I see and error in the image. Fixing it now. :) 20:38
20:42 lucasb left
El_Che AlexDaniel: is this normal behavious when running blin without params? gist.github.com/nxadm/1f034c65a156...70f36328cd 20:46
(if so, I'll push the new image) 20:47
20:48 pmurias joined
AlexDaniel El_Che: hmmmmmm no 21:00
El_Che: looks like a bug after updates in whateverable
jmerelo El_Che: I don't know.. If you do, please let me know. 21:01
21:23 jmerelo left 21:38 [Coke] left 21:45 [Coke] joined, [Coke] left, [Coke] joined 21:59 sena_kun joined
Geth_ nqp/truffle: 730812fff9 | (Paweł Murias)++ | src/vm/jvm/runtime/org/perl6/nqp/truffle/nodes/sixmodel/NQPAttributeGetNode.java
[truffle] Specialize attribute getting
nqp/truffle: e7964642bc | (Paweł Murias)++ | src/vm/jvm/runtime/org/perl6/nqp/truffle/nodes/sixmodel/NQPAttributeGetNode.java
[truffle] Remove unused attribute
nqp/truffle: 91ac2ace5d | (Paweł Murias)++ | 2 files
[truffle] Optimize nqp::bindattr
AlexDaniel El_Che: I fixed whateverable, it should now work 22:06
22:08 sena_kun left
pmurias MasterDuke: pushed some speed ups for the point benchmark 22:11
Geth_ nqp/truffle: 14fd2afe24 | (Paweł Murias)++ | src/vm/jvm/runtime/org/perl6/nqp/truffle/nodes/expression/NQPWValNode.java
[truffle] Remove some debugging WVal code we pay a hefty price for
pmurias MasterDuke: and the WVal code was seriously slowing our "benchmark" down
22:49 vrurg joined 22:52 Xliff joined
Xliff \o 23:06
m: use NativeCall; role R[::T] { method a { CArray[T].new }; }; class A is repr<CPointer> { }; R[A].a.say 23:07
camelia ===SORRY!===
Type check failed in binding to parameter '<anon>'; expected Any but got T (?)
Xliff m: use NativeCall; role R[::T] { method a { CArray[T].new }; }; class A is repr<CPointer> { }; R[uint32].a.say
camelia ===SORRY!===
Type check failed in binding to parameter '<anon>'; expected Any but got T (?)
Xliff ^^ Is something like this possible? 23:08
m: use NativeCall; role R[::T] { method a { CArray[Str].new }; }; class A is repr<CPointer> { }; R[uint32].a.say
camelia NativeCall::Types::CArray[Str].new
Xliff m: use NativeCall; role R[::T] { method a { CArray[::T].new }; }; class A is repr<CPointer> { }; R[uint32].a.say 23:12
camelia ===SORRY!===
Type check failed in binding to parameter '<anon>'; expected Any but got T (?)
Xliff m: use NativeCall; role R[::T] { method a { CArray[::T].new }; }; class A is repr<CPointer> { }; R[A].a.say 23:13
camelia ===SORRY!===
Type check failed in binding to parameter '<anon>'; expected Any but got T (?)
MasterDuke pmurias: nice! that gets truffle equal to regular jvm for 10000000 and beating it by 3s for 50000000 23:20
Geth_ rakudo: vrurg++ created pull request #3272:
Unify dynamic lookups performed by pseudo-packages and &DYNAMIC
Xliff MasterDuke: CArray doesn't support use in parameterized roles, does it? 23:21
MasterDuke Xliff: sadly i have no idea. i know almost nothing about nativecall 23:22
Xliff Ah.
AlexDaniel greppable6: kill
greppable6 AlexDaniel, 565 lines, 81 modules: gist.github.com/122a43ea349370973c...8db4ffe18b
AlexDaniel greppable6: \.kill
greppable6 AlexDaniel, 139 lines, 36 modules: gist.github.com/0025fd5f68fb497823...26df2f01e8
pmurias MasterDuke: what is your exact benchmark? 23:27
MasterDuke pmurias: time ./nqp-j nqp-truffle.nqp -e 'class Point { has $!x; has $!y; method new(:$x, :$y) { my $obj := nqp::create(self); nqp::bindattr($obj, Point, q|$!x|, $x); nqp::bindattr($obj, Point, q|$!y|, $y); $obj; }; method x() { $!x }; method y() { $!y } }; my $total := 0; my int $i := -1; while ++$i < 50000000 { my $p := Point.new(x => 2, y => 3); 23:28
$total := $total + $p.x + $p.y; }; say($total)'
pmurias MasterDuke: I'm using this: paste.debian.net/1112140
MasterDuke yep, looks the same
23:29 ggoebel left
Geth_ nqp/truffle: b8b625c7d9 | (Paweł Murias)++ | src/vm/jvm/runtime/org/perl6/nqp/truffle/nodes/expression/NQPWValNode.java
[truffle] Optimize WVal with specialization
23:31 MasterDuke left
pmurias MasterDuke: with that commit we should be beating the old backkend on 10_000_000 23:32
tellable6 pmurias, I'll pass your message to MasterDuke
AlexDaniel greppable6: ^\s*await
greppable6 AlexDaniel, 658 lines, 136 modules: gist.github.com/67beac3ef7f845a784...a110763016
23:35 MasterDuke joined
MasterDuke pmurias: awesome. and yeah, consistently ~0.5s faster for me after that last commit 23:36
tellable6 2019-10-30T23:32:05Z #raku-dev <pmurias> MasterDuke: with that commit we should be beating the old backkend on 10_000_000
MasterDuke wish i could say i understood exactly what was going on in those optimizations you just did, i'll have to study the commits a bit more
jnthn I figure the WVal one is just saying that once we have resolved it once, then it's a constant. 23:37
And so it can effectively treat it like a Java `final` and do the related Nice Things. 23:38
MasterDuke yeah. i understand some of the intent. but how it's achieved... 23:39
pmurias MasterDuke: I would suggest reading the truffle slides from the readme 23:42
jnthn: yes, that what it does 23:43
lafo.ssw.uni-linz.ac.at/pub/papers...ruffle.pdf 23:44
Xliff "How it should be" --> "Write in Java" 23:46
pmurias: So are you using truffle concepts for nqp/moar? 23:50
Or the JVM backend?
jnthn pmurias: ooc, how does the JVM and (not fair 'cus it's still in progress) truffle compare to the MoarVM result? 23:51
MasterDuke jnthn: for 50000000, moarvm is 2.4s, truffle is 9.5s, and jvm is 14s 23:52
jnthn moarvm master, I guess? 23:55
In theory (totally untested) more-pea would also help this one
Geth_ roast: 464b05b1f1 | (Aleks-Daniel Jakimenko-Aleksejev)++ | S17-procasync/kill.t
Fix test by ignoring the failed Proc

The test is trying to check if rapid-killing a Proc doesn't hang, but while doing so it awaits the result in sink context. Of course, a killed Proc should complain loudly, but it didn't do that in the past because of a bug in Rakudo.
Resolves rakudo/rakudo#3271.
MasterDuke moarvm and jvm HEAD of master, truffle head of truffle branch (which is way behind master, but don't know that any of the missing commits would do much for performance anyway)
Geth_ roast/6.d-errata: 4330d8bf22 | (Aleks-Daniel Jakimenko-Aleksejev)++ | S17-procasync/kill.t
Fix test by ignoring the failed Proc

The test is trying to check if rapid-killing a Proc doesn't hang, but while doing so it awaits the result in sink context. Of course, a killed Proc should complain loudly, but it didn't do that in the past because of a bug in Rakudo.
Resolves rakudo/rakudo#3271.
AlexDaniel jnthn: ↑ thoughts? 23:57
jnthn: it's in 6.d-errata too, fyi
jnthn AlexDaniel: Just found that ticket while trying to catch up on the neverending email inbox... :)
AlexDaniel: I think I agree with nine's call
I suspect it might fix as much as it breaks in the wild. :)
AlexDaniel jnthn: I'm not entirely convinced that this won't break code, but I couldn't come up with a real snippet that'd break because of the change 23:58
jnthn *nod*
AlexDaniel the change doesn't matter if you're using react/whenever
jnthn I guess blin has already had its say?
MasterDuke jnthn: i just tried to use the more-pea branch, but it was too old for nqp and there was a merge conflict when i tried to rebase it head of master
AlexDaniel jnthn: for that change yes, but I'm still waiting for results on HEAD :(
jnthn MasterDuke: OK. I really hope to have time to get back to that next week. 23:59
AlexDaniel so to actually have your code broken, you'd need to await a Proc::Async in sink context, while also .kill-ing it yourself
jnthn Couldn't something else have killed it?