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.
MasterDuke ugh, i have some free time and kind of want to do some hacking, but am just too tired to think. i keep flipping back and forth looking at QRegex/Cursor.nqp and QAST/Regex.nqp and QRegex/NFA.nqp hoping for inspiration to strike 01:01
think it's time to go play some Moss or Disgae 5 instead... 01:02
timotimo disgaea 5! i loved that game ... then i moved to a new apartment and haven't seen the disc since :(
i don't know what Moss is, though
MasterDuke psvr game. sort of a puzzle/platformer 01:03
timotimo ah, psvr is kind of neat
i've been watching loads of linux.conf.au talks today and there was one about vr and vulkan on linux
the bloke has been working on psvr support for openhmd i think?
MasterDuke i really like the presentation, you're mostly stationary as the player, looking into set scenes, but certain secrets are only visible if you move you head around to look behind things and such 01:04
yeah, i've heard that's in the works
timotimo my dad got a psvr for christmas, i got to play some wipeout on it
MasterDuke and i think it was the original disgae that was the reason i bought a ps2
superhot is fantastic. i haven't played it on any other platform, but i can't imagine how it would be nearly as good not in vr 01:05
timotimo i never owned a ps2; the first time i played a disgaea it was Disgaea DS, which is a port of the first one (but it's not very good)
MasterDuke i think i may have accidentally skipped 4, but i've put a bunch of hours into the rest 01:06
timotimo 4 was the second one i played, after that came DD2
MasterDuke beat saber is a fun rhythm game that also only works in vr 01:07
timotimo i've heard of it, but i didn't actually look at it yet. i imagine it can be quite a workout :)
MasterDuke don't think it would ever get to the same level as ddr
timotimo do it completely in horse stance or what it's called 01:08
MasterDuke it really is mostly swinging your hands. some full body movement, but not a whole lot
timotimo ah, ok 01:09
MasterDuke well, off to go play something. maybe i'll be inspired tomorrow (or if anyone wants to write a detailed "type this here", "type that there" kind of instruction i'll follow that) 01:11
timotimo have fun!
03:46 vrurg left 03:48 vrurg joined 05:23 Ven`` joined 05:28 Ven`` left 06:11 robertle left 06:18 ufobat joined 07:51 tyil left, tyil joined
|Tux| Rakudo version 2018.12-304-geb8561ba1 - MoarVM version 2018.12-109-g80d86d590
csv-ip5xs0.754 - 0.758
csv-ip5xs-206.020 - 6.117
csv-parser21.675 - 21.923
csv-test-xs-200.433 - 0.435
test7.085 - 7.364
test-t1.853 - 1.925
test-t --race0.837 - 0.837
test-t-2031.128 - 31.287
test-t-20 --race9.934 - 10.091
 10:27
dogbert2_ lizmat, jnthn: you helped out last week with investigating and fixing a performance related regression in gist.github.com/dogbert17/7423640a...0295154ee9 10:55
Many thanks for that. There was however one thing which confused me. The current version does 29M scalar allocations while older versions does considerably less, i.e. 150k - 2M allocations depending on version 10:57
I found the offending commit, were the number of Scalar allocs starts to increase massively. It's eb3917c260bce1cb56e2ee40abebd12c4dbcd9b2 (Rakudo). 10:59
jnthn dogbert2_: Did you get HEAD MoarVM/NQP, which has profiling that doesn't block PEA opts? 11:11
With that, I saw the number of allocations/GC runs drop off massively :) 11:12
11:21 AlexDaniel joined 11:38 tyil left 11:40 tyil joined 11:44 reportable6 left 11:46 reportable6 joined
dogbert2_ is back after being kidnapped by a lunch party 12:12
jnthn: will try out head
12:18 dogbert2_ left 12:45 dogbert2_ joined
dogbert2_ jnthn: now the profiler output looks like this: 'The profiled code did 42 garbage collections. ... Scalar replacement eliminated 29171849 allocations (that's 89.39%).' 13:04
which is impressive. Thing is that these 29M allocs were not present at all as late as github.com/rakudo/rakudo/commit/6b...b1beaf4277 13:13
the following commit, i.e. github.com/rakudo/rakudo/commit/eb...2c4dbcd9b2 jumps from 150k scalars to 54M scalars and 666 GC's, a few commits later it's down to 29M allocs and 376 GC's 13:15
13:20 lucasb joined
jnthn dogbert2_: Yeah, we relied on lazy vivification of lexicals avoiding those allocations before now. With the latest opts they are lowered to locals, where there's no laziness machinery, but PEA can determine they are unused and eliminate them. 13:29
It may be that we can determine it's unused statically and kill it off altogether, though :) 13:30
dogbert2_ sounds interesting, so there are still plenty of possible optimizations 13:39
lizmat notable6: weekly 13:42
notable6 lizmat, 6 notes: gist.github.com/c586cfb382637a00f6...08c291a0b5 13:43
lizmat is there a reason why MoarVM/nqp haven't been bumped? 14:03
15:07 [Coke] joined, [Coke] left, [Coke] joined 15:08 [Coke] left
Geth MoarVM must have its webhook fixed; no channels were sent by github, please check if # needs to be replaced by %23 15:10
nqp: b962a12cb9 | (Timo Paulssen)++ | src/vm/moar/HLL/Backend.nqp
Expose Replaced Allocations To SQL Profiler
 15:32
timotimo jnthn: if we want git notifications via geth to continue working, you'll have to edit the webhook to have %23 instead of # in the url
jnthn timotimo: Which repos? MoarVM? 15:37
timotimo yeah
and probably also the website one?
lizmat timotimo: good to bump ? 15:38
timotimo yes
lizmat oki
jnthn timotimo: hmm, I think the website one is pointing at the wrong url
Geth: help
Geth jnthn, Source at github.com/perl6/geth To add repo, add an 'application/json' webhook on GitHub pointing it to hack.p6c.org:8888/?chan=#perl6 and choose 'Send me everything' for events to send | use `ver URL to commit` to fetch version bump changes
timotimo ah that needs changed 15:39
jnthn Can the bot also suggest the right thing? :P
timotimo only the # to %23 thing
15:40 Geth left, Geth joined, ChanServ sets mode: +v Geth
timotimo Geth: help 15:40
Geth timotimo, Source at github.com/perl6/geth To add repo, add an 'application/json' webhook on GitHub pointing it to hack.p6c.org:8888/?chan=%23perl6,%23#perl6-dev and choose 'Send me everything' for events to send | use `ver URL to commit` to fetch version bump changes
lizmat and another Perl 6 Weekly hits the Net: p6weekly.wordpress.com/2019/02/18/...dy-thanks/ 15:43
vrurg jnthn: Can you help me with R#2698? I'm trying to work out a fix but it needs some kind of overview. Without a help I'm stuck and just wasting time.
synopsebot R#2698 [open]: github.com/rakudo/rakudo/issues/2698 Typecheck against a curried role dies
jnthn vrurg: Yeah, I'll try and catch up with tickets soon...meant to at the weekend, but was had other distractions. 15:44
vrurg jnthn: I know you're busy and sorry to bother. Just desperate now. :) 15:46
Actually, I'm looking for any opinion from anybody. In a way, the ticket is about language design matters. 15:47
And without fixing this one I can't work on another ticket. :(
Geth nqp: 9e539b64f6 | (Elizabeth Mattijsen)++ | tools/build/MOAR_REVISION
Bump MoarVM for the latest goodies
 15:49
¦ nqp: version bump brought these changes: github.com/MoarVM/MoarVM/compare/2...9-g80d86d5
lizmat notable6: weekly reset 15:56
notable6 lizmat, Moved existing notes to “weekly_2019-02-18T15:56:59Z”
Geth rakudo must have its webhook fixed; no channels were sent by github, please check if # needs to be replaced by %23 16:11
lizmat just bumped NQP_REVISION 16:12
so this is still a Github issue ?
timotimo yeah 16:14
if you set the url to be blah/bloop?meh=#hehe it'll just use blah/bloop?meh= and that's it
lizmat so why not make it bla/bloop?meh=%23hehe ? 16:17
timotimo yes 16:19
that's exactly what you have to do
16:25 jdv79 left 16:26 jdv79 joined
lizmat timotimo: but who needs to do that ? 16:26
16:28 ufobat left
timotimo whoever sees a "Settings" button near the top on the rakudo/rakudo repository 16:37
github.com/rakudo/rakudo/settings/hooks 16:38
jnthn I'm guessing there was some GitHub behavior change here?
timotimo yeah
Geth rakudo: 51f035c4e2 | (Elizabeth Mattijsen)++ | tools/build/NQP_REVISION
Bump NQP to get the latest goodies
 16:39
rakudo: version bump brought these changes: github.com/perl6/nqp/compare/2018....1-g9e539b6
f56e0756f1 | (Elizabeth Mattijsen)++ | src/core/Lock/Async.pm6

As these are heavily used in server applications, it seemed worthwhile to get the 2.5% improvement. Benchmark based on basic react / whenever listen / whenever connection.Supply.lines -> $line block getting hammered by a single client for 100_000 times.
jnthn There we go
Fixed
lizmat jnthn++
timotimo: that link 404's for me, so I guess I don't have any rights to do that 16:41
timotimo that's probably it, yeah 16:44
jnthn vrurg: I've answered at least some things; I'll need to dig a bit more into the others. 17:26
MasterDuke jnthn, nine, lizmat, et al.: after the release, think github.com/perl6/nqp/pull/518 could get some eyes on? 17:38
jnthn MasterDuke: Yeah, I decided to start with reviewing the MoarVM one, which I did, and I think you've probably addressed all those bits now. :) 17:39
dinner, bbl 17:40
MasterDuke jnthn: yes, thanks. i suspect the nqp PR will be the most interesting
Geth evalbot must have its webhook fixed; no channels were sent by github, please check if # needs to be replaced by %23 17:46
vrurg jnthn: Thanks a lot! I'll be able to check it out later today. 17:48
AlexDaniel Geth: help 18:25
Geth AlexDaniel, Source at github.com/perl6/geth To add repo, add an 'application/json' webhook on GitHub pointing it to hack.p6c.org:8888/?chan=%23perl6,%23#perl6-dev and choose 'Send me everything' for events to send | use `ver URL to commit` to fetch version bump changes
evalbot: Xliff++ created pull request #10:
- Enables evalbot to process code sandwitched between a markdown code…		 18:27
AlexDaniel Geth: your help message is broken
Geth: you stoopid 18:28
hm it's actually different here github.com/perl6/geth/blob/fdc889d...eth.p6#L14 18:29
so I don't know what's going on
timotimo how is it broken, AlexDaniel? 19:16
AlexDaniel timotimo: “perl6” and “#perl6-dev” with # 19:17
timotimo no, you're broken :)
AlexDaniel and that message is not in the repo
timotimo oh
AlexDaniel I'm not broken, the thing it printed doesn't work, I tried
I removed the # and that worked
timotimo yes 19:18
19:18 Geth left, Geth joined, ChanServ sets mode: +v Geth
timotimo Geth: help 19:19
Geth timotimo, Source at github.com/perl6/geth To add repo, add an 'application/json' webhook on GitHub pointing it to hack.p6c.org:8888/?chan=%23perl6,%23perl6-dev and choose 'Send me everything' for events to send | use `ver URL to commit` to fetch version bump changes
timotimo now it's correct
AlexDaniel: gist.github.com/timo/a860058348128...18870932a8 19:27
AlexDaniel yeah that should be committed 19:32
samcv AlexDaniel, hoi 20:04
yoleaux 17 Feb 2019 17:17Z <tbrowder> samcv: do you have a p6 parser for any unicode xml data—specifically for CLDR?
AlexDaniel samcv: hello!
samcv hi. what timezone are you in btw
i'm working from home tomorrow so i will be responsive on irc during daytime amsterdam time and night 20:05
AlexDaniel +02:00 UTC with irregular sleeping pattern
samcv ah ok
well want to do the release tomorrow?
AlexDaniel ⏳ 1237 out of 1240 modules processed (left: Red PDF::Font::Loader ANTLR4::Grammar) 20:06
that's from Blin… I think it is bisecting Red again
yeah
samcv: I think tomorrow is a bit too early
but tomorrow we'll probably know where we stand exactly 20:07
so, looking at this: github.com/rakudo/rakudo/issues?q=...2%9A%A0%22
like, OK, maybe flappers we can somewhat ignore
Algorithm::LibSVM is perhaps a non-issue because the module was broken anyway… 20:08
#2635 I don't know…
it's a weird ticket :) 20:09
samcv okay that's fine :)
.tell tbrowder no i don't have any parser for the unicode xml. It would be nice to do. Which data did you want from CLDR? 20:10
yoleaux samcv: I'll pass your message to tbrowder.
20:16 patrickb joined
Geth whateverable must have its webhook fixed; no channels were sent by github, please check if # needs to be replaced by %23 20:23
tbrowder samcv: thanks. guifa and i are looking particularly at date names for my Date::Names modulei found a p5 parser on github/patch for patch's cpan CLDR modules. i'm using his mkcldr.pl prog as a base for my p6 version, with changes to extract just pieces i need. 20:29
yoleaux 20:10Z <samcv> tbrowder: no i don't have any parser for the unicode xml. It would be nice to do. Which data did you want from CLDR?
samcv tbrowder, oh nice. is it online yet if i want to take a look? 21:07
yours i mean
tbrowder look at github/tbrowder/Unicode*/bin/mkcldr.p6. just started, p5 version is in a bin subdir. 21:11
22:03 patrickb left, dct joined
AlexDaniel updated list: github.com/rakudo/rakudo/issues?q=...2%9A%A0%22 22:10
jnthn AlexDaniel: I'm not convinced there's anything wrong Rakudo wise in 2635. I mean, it's not "the compiler"'s fault if the program it's running requests a ton of memory. 22:17
AlexDaniel yeah, but what is requesting memory?
jnthn (compilation is long over by that point, so I guess compiler is being used in a handwavey way)
AlexDaniel: Apparently, Cro::WebSocket's parser if you throw HTTP/2 at it.
AlexDaniel oh well… 22:18
I'm OK with blaming it on Cro::WebSocket and removing the label :)
jnthn I don't understand why we're seeing github.com/rakudo/rakudo/issues/2704 now 22:19
(And not before)
I can look in to that one, perhaps tomorrow
Yeah, should get time tomorrow. Too tired today. 22:20
AlexDaniel jnthn: I also don't quite understand how I missed it… 22:23
it's here github.com/perl6/ecosystem-unbitrot/issues/553
22:23 Ven`` joined
AlexDaniel so it has always been there… 22:23
my mistake
jnthn ah, ok 22:27
AlexDaniel well, even though it is the first module in this list, it is still hard to see… gist.github.com/AlexDaniel/b681cae...3a29f8232c 22:29
wait uh that's the wrong list
MasterDuke i'm finding it difficult to use nqp::radix_I in NQP. in NQPMatchRole to be precise 22:30
AlexDaniel updated
the list is correct now
maybe I should still do the meta ticket thingie where I list all of the modules… 22:31
MasterDuke nqp: say(nqp::radix_I(10, "12000000000000000000000000000000000000", 0, 0, nqp::knowhow().new_type(:repr("P6bigint")))[0]) 22:33
camelia Cannot unbox 124 bit wide bigint into native integer
at gen/moar/stage2/NQPCORE.setting:958 (/home/camelia/rakudo-m-inst-1/share/nqp/lib/NQPCORE.setting.moarvm:join)
from gen/moar/stage2/NQPCORE.setting:946 (/home/camelia/rakudo-m-inst-1/share/nqp/lib/…
MasterDuke m: use nqp; say(nqp::radix_I(10, "12000000000000000000000000000000000000", 0, 0, nqp::knowhow().new_type(:repr("P6bigint")))[0]) 22:37
camelia Cannot find method 'gist': no method cache and no .^find_method
in block <unit> at <tmp> line 1
AlexDaniel jnthn: also github.com/rakudo/rakudo/issues/2706 22:40
and yeah now there's a meta ticket :) github.com/rakudo/rakudo/issues/2705 22:41
heh that graph is pretty cool gist.github.com/AlexDaniel/803489b...erview-svg 22:42
(right click → View Image) 22:43
22:55 Ven`` left, Ven`` joined
jnthn AlexDaniel: Urgh, that's triggered by a fix for another module :/ 23:29
It's a totally weird failure mode though
I can't imagine how that change would casue that failure...
Ah well, tomorrow 23:30
23:34 dct left 23:39 lucasb left 23:44 Ven`` 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!