»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg camelia perl6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by sorear on 25 June 2013.
00:13 carlin left
timotimo i'll write this weekly on tuesday, so that i'll have 6 days to cover rather than just 5. 00:18
00:22 rindolf joined 00:24 rurban joined 00:26 rurban1 joined 00:28 hoverboard joined 00:29 rurban left 00:56 Ben_Goldberg joined 00:59 BenGoldberg left, dmol left 01:09 rgbkrk joined 01:25 rgbkrk_ joined 01:28 Su-Shee_ joined, rgbkrk left 01:31 Su-Shee left 01:33 jpwillis_ joined 01:36 klapperl_ left, bjz joined 01:37 klapperl joined 01:45 jnap left 01:52 rgbkrk joined 01:53 rgbkrk_ left 01:57 chenryn left 01:58 chenryn joined 02:03 jnap joined
dalek rl6-roast-data: 0f823d1 | coke++ | bin/rakudo.jvm.sh:
Use eval server for jvm.
 02:05
[Coke] donaldh++ 02:06
# ^^
02:22 byte48 left 02:24 BenGoldberg_ joined 02:25 araujo left 02:26 araujo joined 02:27 Ben_Goldberg left 02:28 d^_^b left 02:38 xragnar_ joined, xragnar is now known as Guest75686, xragnar_ is now known as xragnar 02:41 Guest75686 left 02:42 SamuraiJack joined 02:55 dj_goku_ joined 02:58 dj_goku left 03:02 Alina-malina left 03:03 jnap left 03:08 rgbkrk left 03:09 rgbkrk joined 03:11 rurban1 left 03:15 dj_goku joined, dj_goku left, dj_goku joined, dj_goku_ left 03:18 rgbkrk left, rgbkrk joined 03:30 rgbkrk left, rgbkrk_ joined 03:31 Alina-malina joined 03:36 rgbkrk_ left 03:37 rgbkrk joined 03:39 BenGoldberg_ left 03:42 rurban joined 03:44 uniejo joined, uniejo left 03:47 rurban left 03:51 zengargo1le left 03:59 hoverboard is now known as moistcherry 04:02 kaare__ joined 04:06 rgbkrk left 04:20 SamuraiJack left 04:22 rindolf left 04:37 kurahaupo left 04:39 anaeem1 joined 04:41 moistcherry is now known as hoverboard 04:48 hoverboard left 04:51 kurahaupo joined 04:53 kaare__ left 05:10 kst joined 05:11 [Sno] joined, [Sno] left 05:20 kaare__ joined 05:22 thou left 05:24 kaare__ left, kaare__ joined 05:36 kaare__ is now known as kaare_ 05:42 jpwillis_ left 06:00 brrt joined 06:13 kaleem joined 06:21 SevenWolf joined 06:39 darutoko joined
ChoHag I cannot build rakudo on freebsd because dyncall has a special-snowflake build system (consisting of ~ 5900 lines, opposed to the 6200 lines of *.c and *.h) which breaks because gcc isn't installed. 06:46
And the special snowflake build system resists my attempts to wedge cc in whichever magic variables its interrogating. gcc always slips in somehow. 06:48
06:49 FROGGS joined
ChoHag Why, after >40 years, is it still considered difficult to build C? 06:50
nwc10 ChoHag: specifically, you're failing at the point where it (or you) is trying to build MoarVM? 06:54
ChoHag No. It's trying to build dyncall.
Despite it being installed.
nwc10 yes, but dyncall isn't in Rakudo
ChoHag Although the latter is after the former already failed once.
nwc10 Rakudo's Configure.pl is attempting to build MoarVM first? (which is where dyncall is) 06:55
ChoHag I downloaded rakudo and told it to build. It fell over when it got to dyncall.
That it built a subproject is immaterial.
nwc10 it is material in that I'm trying to work out your bug report.
ChoHag No, you're trying to assign blame.
What matters it whence dyncall came? 06:56
nwc10 No, I'm trying to work out which bit of the source tree to look at.
ChoHag The bit with dyncall in it.
nwc10 as you said - where does dnycall come from
ChoHag Oh look - did another magic special snowflake build system fall over and confuse developers and users alike?
No I'm not bitter this morning at all.
06:57 thou joined
nwc10 It did. It's dyncall's special build system, called from MoarVM, called from Rakudo. 06:57
ChoHag Rakudo downloaded a bunch of things, including (so it seems - signal is difficult to discern from the noise) nqp. 06:58
nwc10 yes, agree it's bust
no, I don't know how to fix it.
ChoHag Actually it seems to be getting it via parrot.
Certainly I'm seeing that in the noise more than moar.
nwc10 OK. Thanks. Although I'm afraid that I also don't know how to fix that.
ChoHag Well ultimately it's irrelevant.
dyncall has a special snowflake build system which melted. Rakudo, through some labrynth or other, depends on dyncall. The nature of the labrynth does not affect the brokenness of dyncall. 06:59
rakudo/moar/parrot/nqp/whatevertf needs to fix its dependency, or fix its dependent. 07:01
Or stop claiming to do everything all at once in what is effectively an alpha or beta product.
07:01 thou left
DrEeevil worksforme 07:02
ChoHag Yes thanks Joe Random Developer. Go back to your Apple.
DrEeevil the problem is always that people diverge in unexpected ways, so assumptions fail, and then things go boom
ChoHag: hey now, be more constructive 07:03
if I were stuck on Apple stuff I wouldn't find my way to IRC
ChoHag I was, but the build system broke.
07:03 jsn_ joined, pdcawley left
FROGGS morning 07:03
DrEeevil so instead of trying the "incoherently yelling grandpa" routine give enough information so people can try to help 07:04
ChoHag And yes, people diverge, that's why gnu wrote the satanic work of art which the autotools abomination.
07:04 pdcawley joined
ChoHag I'm not here for help, I'm here for a rant. 07:04
DrEeevil points at WAF - autotools is quite saintly
ChoHag The solution is obvious - fix the build system (not gonna happen) or just install gcc.
DrEeevil scons is also a great one, or gyp if you want to go meta
ChoHag Since I turned to the dark side and went full sysadmin, I'll be installing gcc and SEPing it. 07:05
FROGGS ChoHag: have you already no-pasted the build errors you get? (I've not seen a link in the logs)
ChoHag Yes yet more developers decided autotools didn't solve the problem evilly enough and redid it.
Isn't it obvious? 07:06
gmake[2]: gcc: Command not found
DrEeevil they reimagined it, badly
ribasushi ChoHag: you are being disruptive for no reason. Please take a deep breath and let's restart he conversation
ChoHag That and some gmake spam is practically it.
FROGGS ChoHag: and you tried to build rakudo for what backends? 07:07
still, a paste of the build output would help
ribasushi including the build commands
FROGGS yeah
ribasushi as it's still unclear how/what you tried to build ;)
brrt seconded :-) 07:08
FROGGS hi brrt
brrt hi FROGGS
ribasushi ChoHag: make no mistake - we want to help you
FROGGS brrt: you feel more lucky today? :o)
07:08 zakharyas joined
ChoHag Then fix dyncall's abortive attempt to guess what compiler toolchain I have. 07:09
brrt ehm.... why? was my post yesterday as grumpy as i myself
ChoHag Its guessing code includes this gem:
brrt :-)
FROGGS ChoHag: in case I can reproduce your problem, I will
ChoHag guess_tool()
{
FROGGS hold on
ChoHag CONFIG_TOOL="gcc"
FROGGS please use a paste service
and also, I'd like to have the entire build output 07:10
not something you have guessed where the problem is
ChoHag I haven't guessed.
I tracked it down.
brrt please ChoHag, we understand your inpatience :-)
ChoHag That's where the (first) problem is.
Having overridden that, more came up.
brrt ok, but you understand that we can't help you if you won't show us build output?
ChoHag I'm not impatient. I'm tired.
FROGGS ChoHag: I cannot help you when I don't know the problem from the beginning to the end 07:11
ribasushi ChoHag: our help is conditioned on you not trying to diagnose your problem in the dark, and feeding us pieces of likely irrelevant codepaths - you need to show us everything, using a paste service
FROGGS ChoHag: well then: paste && sleep && apply patch
ChoHag FROGGS: I'm really not guessing. dyncall attempts to detect my C compiler and guesses gcc, which isn't installed. It then falls over.
How it gets there *is* entirely irrelevant.
brrt ok, awesome
FROGGS ChoHag: but which dyncall? 07:12
brrt if that is true, then we can probably repeat that, and file a bug report
ChoHag gmake[1]: Entering directory `/usr/home/mking/src/rakudo/nqp/3rdparty/dyncall'
FROGGS that seem to be MoarVM's
ChoHag See now if you'd asked that, I'd have though "Oh, there's more than one? Typical."
arnsholt No, that's the one used for Parrot, I'm pretty sure
brrt hmmm /me checking out nqp source
ribasushi brrt: except it isn't awesome - the gcc is likely a last resort fallback, and something else failed higher up, but we can't see it
FROGGS arnsholt: you are right, moar's would be in /moar 07:13
ChoHag Instead, I thought "how the fuck am I going to quickly and correctly paste a god-knows-how-many-line terminal transcript which is already long lost?"
FROGGS or /MoarVM
brrt what happens when you alias cc/clang to gcc
FROGGS meh
ribasushi brrt: this should *not* be necessary ever
ChoHag brrt: I have little doubt that it would work, but I'd sooner just install gcc to hide the problem than leave random symlinks lying around to system tools. 07:14
brrt isn't afraid of a hack or so
ChoHag You're young yet.
ribasushi brrt: with this approach ChoHag's problem will keep recurring and wasting a lot of time on all sides 07:15
brrt :-)
yes, which is the reason for my next question
what do you want rakudo for?
ChoHag Indeed. Hence why I tracked it down rather than merely avoiding it.
For shits. 07:16
brrt and giggles?
ChoHag The very same.
brrt what happens when you patch dyncall's build system? or is that inpractical?
ribasushi brrt: he is trying to report a bug (combative tone notwithstanding) 07:17
ChoHag dyncall's build system is just shy of 6000 lines of code.
arnsholt Yeah, it's a mess
ChoHag I tried wedging cc into the guess_tool function (and then fixing guess_as, which promptly broke), but gcc just came bacj. 07:18
FROGGS can't we update the clone of dyncall in the nqp repo? most parts of it are like three years old... 07:19
ChoHag That's about the point where I (silently, because my daughter is still sleeping) cursed ever developer who ever lived.
FROGGS /o\
brrt thanks, we need the luck :-)
but you're quite right 07:20
gcc is refered many times in dyncall, directly
it seems that dyncall asssumes to only work with gcc 07:21
FROGGS ChoHag: can you please try this in another folder? git clone github.com/rakudo/rakudo.git && perl Configure.pl --gen-moar --backends=moar --make-install
arnsholt ChoHag: I'm a bit out of date one my fBSD knowledge, but the most recent versions have clang instead of gcc as the default compiler, no?
FROGGS ChoHag: this builds rakudo for MoarVM only, which has an up to date dyncall
ChoHag: (and building for MoarVM only is pretty much the fastest way to build a Perl 6 compiler these days) 07:22
07:22 hoverboard joined
ChoHag arnsholt: I believe so. 07:22
brrt wanted to install freebsd but couldn't boot from dvd :-s 07:23
arnsholt From what you're experiencing, that likely broke a pile of assumptions about fBSD in the config stuff
ChoHag FROGGS: I cheated a bit and cloned from ~/rakudo, because I only cloned that one last night.
ribasushi ChoHag: stepping back - do you care which backend you are going to build? perhaps building MV only would be a way forward for now, until the repo is updated?
arnsholt Moar likely has the same dyncall
ChoHag ribasushi: Not especially. I have linux boxes it works on.
Hence why this is all for the ranting, not the fixing.
ribasushi arnsholt : <FROGGS> ChoHag: this builds rakudo for MoarVM only, which has an up to date dyncall 07:24
arnsholt (We should probably set up our own fork of dyncall to make sure both backends get the same version, actually)
FROGGS ChoHag: sure, though the biggest repo is nqp here, both rakudo and moarvm are sort of tiny
ChoHag Yeah I see that...
arnsholt The easiest way to circumvent this would be to install just the JVM version
ChoHag Bandwidth not so great in rural England.
FROGGS arnsholt: moar's is 0.8, nqp's is 0.7
arnsholt Of course then you have to live with the JVM start-up time 07:25
FROGGS dang, he is a British
FROGGS hides
07:25 mr-foobar joined
arnsholt FROGGS: Really? In that case, we should probably bump the one used for Parrot and then set up a shared upstream fork 07:25
ChoHag Can upstream not be patched?
FROGGS ChoHag: you are on freebsd btw? what os/arch exactly? (then I can set up a VM, and debug it while you sleep or work) 07:26
arnsholt IIRC the actual dyncall people have not been responsive
FROGGS ChoHag: perhaps, yes
ChoHag FROGGS: 10-release (with updates) on kvm.
FROGGS arnsholt: yes, we should bump, but first I wanna know it works
ChoHag: thanks
brrt 0.8 says it should be clang-buildable on freebsd
ChoHag freebsd hasn't made it onto tin yet here. 07:27
This build may have got past the previous build's fallover point. I didn't stay to watch the first one.
brrt where is it now? if its building nqp-moar, it probably is (dyncall is integrated into moarvm) 07:28
ChoHag Oh it's done.
Just in time. School run imminent.
So yeah, moar works.
arnsholt Awesome
ChoHag Want the transcript? I should have turned on verbose.
arnsholt If it works, don't bother =) 07:29
Hopefully 0.8 can just be plonked in in place of 0.7
ChoHag Hmm.
I'll have to try this again when I got back.
FROGGS okay, I'll update nqp's dyncall, and test that parrot still does its thing with it
ChoHag Since starting the first (failed) build and this one, I've installed the system's dyncall. 07:30
arnsholt FROGGS: Cheers!
FROGGS ChoHag: thanks for the last ten minutes, these were productive :o)
arnsholt Remember to run the NativeCall test suite too, that's the one that really exercises it the best
brrt tl;dr: the vesion of dyncall bundled with nqp for the parrot backend is out of date
:-)
masak good morning, #perl6
ChoHag Is the distributed version _always_ used?
brrt morning masak
FROGGS ChoHag: I believe so, yes
brrt .... i think so ChoHag, i can't recall seeing anything checking for an installed version 07:31
brrt is off for an appointment, bye :-)
arnsholt Pretty sure the bundled one is hardcoded, yeah
FROGGS o/ brrt
07:31 brrt left
ChoHag OK must run. 07:31
07:33 dmol joined
masak ChoHag: wow, do you ever go on in the backlog. 07:35
ChoHag: I don't see much ranting on this channel. the most rant-y person is usually me, I think :)
arnsholt Yeah, that was a pretty epic rant. But at least it looks like we figured it out in the end =) 07:38
masak yeah, I'm caught up now. 07:40
ChoHag++ #perl6++ # for, despite everything, enduring each other all the way
arnsholt Le sigh. PayPal using "security" questions makes me sad 07:44
sergot morning o/ 07:47
FROGGS hi sergot
07:54 kivutar joined 07:56 pecastro joined
ChoHag masak: Well if everything wasn't so shit all the time, ranting wouldn't be necessary. 08:01
08:01 denis_boyun joined
ChoHag But dear god don't come at me with "worksforme". 08:01
If there was ever a way to annoy your local sysops... 08:02
DrEeevil the trick there is to figure out why you diverge from others
if you start ranting don't expect constructive responses 08:03
08:03 fhelmberger joined
ChoHag And what should you expect if you tell somebody with a problem that you don't have it? 08:04
Alice> Hey guys my thing broke. 08:05
Bob> Well mine didn't.
Alice> ...
moritz where's Eve? 08:06
ChoHag She and Adam shacked up ages ago.
Started the whole damn mess.
Alice had to step in, and now she broke it.
08:10 donaldh joined 08:12 virtualsue joined, denis_boyun left
DrEeevil so how did she break it, and why did no one stop her? ;) 08:14
donaldh Another PR to tune the eval server heap and permgen for better JVM spectest time: github.com/rakudo/rakudo/pull/278 08:15
ChoHag DrEeevil: That's the point when you just stop and say "Yes, dear." 08:16
08:16 spider-mario joined
masak anyway, glad to see so many p5 faces in here today. :) 08:16
dalek kudo/nom: 15002e3 | (Donald Hunter)++ | tools/build/create-jvm-runner.pl:
Tune the eval server max heap and permgen for best spectest time.
 08:19
kudo/nom: 7ca23fd | lizmat++ | tools/build/create-jvm-runner.pl:
Merge pull request #278 from donaldh/eval_server

Tune the eval server max heap and permgen for best spectest time.
kudo/nom: e4d97a5 | (Elizabeth Mattijsen)++ | t/spectest.data:
Add advent2011-day(03|04|05|07|10) tests
 08:23
08:25 jsn_ left 08:26 SamuraiJack joined 08:39 Su-Shee_ is now known as Su-Shee 08:41 dakkar joined
dalek p: 21fd47c | (Tobias Leich)++ | 3rdparty/dyncall/ (322 files):
bump 3rdparty/dyncall from v0.7 to v0.8

This is the same version that also MoarVM keeps in its repository. v0.8 should fix build problems for nqp@parrot on FreeBSD. ChoHag++		 08:45
08:45 thou joined
dalek kudo/nom: beb0525 | (Tobias Leich)++ | tools/build/NQP_REVISION:
bump nqp rev for dyncal@parrot fix
 08:46
ChoHag Is there anything like a consensus on testing modules? 08:47
FROGGS ChoHag: what do you mean exactly? 08:49
ChoHag I looked at Test.pm and it begins with the encouraging comment that "This is a temporary Test.pm to get us started"
What to put in t/*.t
use Test and ok 1 work, so the essentials are there. 08:50
08:50 thou left
FROGGS first, Test.pm is good enough, and there is no replacement in sight yet (and I don't need any) 08:50
ChoHag I see some test-related modules in the Big List Of Modules.
FROGGS and, yes, put your stuff in t/*.t, and run with prove for example 08:51
ChoHag So I am wondering what commentary and opinions there are on their various states of readiness.
FROGGS for some stuff it is good to test not only on a single backend though, which requires some patience
I think these modules are either for parsing TAP (what the module installer needs), or to run a smoker 08:52
ChoHag Well that'll just mean rerunning the test suite won't it?
FROGGS yes 08:53
that also means that you have to rebuild rakudo for other backends :o)
ChoHag So IOW just keep it simple - plain perl and Test.pm?
FROGGS yeah
ChoHag Well I see that's patched now, so I can probably pull and build.
FROGGS it just takes a while to build for all three 08:54
(because parrot and jvm are slower than moar)
09:07 brrt joined
donaldh Well, LOL, all the spectests that use is_run to spawn a subprocess fail on JVM because $*EXECUTABLE_NAME = perl6-eval-server 09:15
Two possible ways to fix this: 1. hard-code perl6.execname=perl6-j in the perl6-eval-server runner ; 2. check for perl6-eval-server and fix it in Test/Util.pm 09:18
I guess the question is: Should perl6-eval-server masquerade as perl6-j for all the spectests? 09:20
09:22 kivutar left
ChoHag Oh in all the ranting, I forgot that parrot has a similar problem, only I can fix it by specifying --link=c++ to parrot's Configure.pl 09:27
(via rakudo's --parrot-option option)
Because the option was right there, I made no attempt to diagnose why it was detecting g++ for the linker. 09:28
FROGGS j: say $*EXECUTABLE 09:36
camelia rakudo-jvm a7dd96: OUTPUT«Unhandled exception: java.lang.RuntimeException: Missing or wrong version of dependency 'gen/jvm/stage2/QRegex.nqp'␤ in (src/Perl6/Pod.nqp)␤ in (src/vm/jvm/ModuleLoader.nqp:78)␤ in load_module (src/vm/jvm/ModuleLoader.nqp:60)␤ in (src/Perl6/A…»
FROGGS hmpf
but that should do
ChoHag Bah. nqp fell over again. This time with a linker error against dyncall. 09:37
FROGGS ChoHag: it uses Perl 5's Config.pm to get defaults...
:/
ChoHag /usr/bin/ld: ../../../../3rdparty/dyncall/dyncall/libdyncall_s.a(dyncall_api.o): relocation R_X86_64_32S against `a local symbol' can not be used when making a shared object; recompile with -fPIC 09:38
../../../../3rdparty/dyncall/dyncall/libdyncall_s.a: could not read symbols: Bad value
FROGGS ohh
ChoHag Yeah.
FROGGS it guesses PIC wrong
donaldh FROGGS: It tries to spawn perl6-eval-server when it really needs to run perl6-j 09:39
09:39 SevenWolf left
FROGGS hmmm 09:39
and $*EXECUTABLE says it is perl6-eval-server ?
donaldh I will need to double check that. 09:40
09:41 CAABDV6 joined
FROGGS okay, my vote is that we create a env var that influences $*EXECUTABLE 09:43
like RAKUDO_PERL6_EXE, which would just be perj6-j, and rakudo would then add the path (and .exe if needed) 09:44
donaldh FROGGS: you're right $*EXECUTABLE is correct.
FROGGS ohh, nice
donaldh we just need to change Test::Util to use $*EXECUTABLE instead of $*EXECUTABLE_NAME
masak +1 09:45
though it's weird that we have both of those variables.
they're a bit unfortunately named, no?
I mean, either of them in isolation would've been fine, but... 09:46
FROGGS masak: we are in transition from $*EXECUTABLE_NAME to $*EXECUTABLE
donaldh Ah, no wait. I had already hacked perl6-eval-server to hard-code perl6-j
FROGGS donaldh: btw, $*EXECUTABLE might be wrong on parrot before it is installed 09:47
ahh
donaldh let me check again
FROGGS yeah, I feared it is not that easy
09:47 Psyche^_ joined
FROGGS I think $*EXECUTABLE_NAME is $*EXECUTABLE.basename 09:47
so, both should be right or wrong
(except for the path)
donaldh Yep both are perl6-eval-server (full path and basename) 09:50
s/eval-server/j/ in EvalServer.java ?
Or fix it in the perl6-eval-server runner? 09:51
09:51 Psyche^ left
masak FROGGS: if one can be deduced from the other, isn't it dangerous to have them sitting around in two variables? what if one of them changes? 09:51
09:51 hoverboard left
donaldh EXECUTABLE_NAME will get deprecated at some point but is currently used. 09:52
FROGGS true, we need to fix parrot first
(which is not easy btw)
donaldh Any solution to this is a hack because perl6-eval-server is a hack :-) 09:54
It is easiest to fix in EvalServer.java because if we try and fix it in perl6-eval-server, we'd need to figure out how to do string substitution on both Windows and Unixy 09:56
10:09 brrt left
ChoHag After persuading the new dyncall to build with PIC - no easy task - rakudo built parrot through to the end. 10:15
FROGGS ChoHag: can you provide a diff/pull request? 10:22
ChoHag No. I didn't fix it. 10:23
I just got dyncall to build by hand.
However now I've confirmed that it does indeed fix it, I can get on that.
Whan rakudo finishes building.
FROGGS nice
ChoHag or not. SIGKILL during parse of /usr/local/bin/perl5.20.0 tools/build/gen-cat.pl parrot (lots of .pm files) > gen/parrot/CORE.setting 10:26
Why would it be KILLd? 10:27
FROGGS O.o
ChoHag What does the killing?
FROGGS I dunno
that is just Perl 5 fiddling with text files 10:28
ChoHag 1GB RAM not enough?
FROGGS should be enough for Perl 5
ChoHag Ohhhh yes. 5.
Weird.
No wait. It says this:
FROGGS 1GB is a problem for rakudo@parrot and rakudo@jvm though
ChoHag ./perl6-p --setting=NULL --ll-exception --optimize=3 --target=pir --stagestats --output=src/gen/p-CORE.setting.pir gen/parrot/CORE.setting 10:29
So it's 6. Sanity restored.
FROGGS IIRC r@p takes about 650MB for a hallo world
k
MoarVM is your friend then
ChoHag Meh. It's a VM.
FROGGS (limit resources was one of its goals btw)
limited* 10:30
everything is a VM nowadays
tadzik OS is a VM :)
ChoHag Yeah. I realised that wasn't exactly clear.
10:30 donaldh left
ChoHag Each pid is a VM - its own little virtual unix. 10:30
Oh no I'm going to get started on about docker if this one goes on. 10:31
I meant I'm building rakudo inside kvm and can easily increase the available ram. 10:32
10:33 donaldh joined, thou joined 10:34 brrt joined
ChoHag aaand it passed. 10:34
Rakudo builds in full on parrot and moar on an updated stock freebsd 10-release. 10:35
(amd64/kvm/amd64)
Needs >1GB RAM
FROGGS \o/ 10:36
10:38 thou left
ChoHag Just need to persuade dyncall to build with PIC 10:39
Also I say amd64, it's actually intel.
dalek rlito: 4d01fd7 | (Flavio S. Glock)++ | src5/lib/Perlito5/Javascript2/Emitter.pm:
Perlito5 - js - create an additional scope for variables declared in 'for' list
 10:40
ChoHag Right. I spent so long getting that installed I can't remember why I wanted it. 10:41
10:41 SamuraiJack left
ChoHag However while you're here, and in lieu of a proper report on the matter, the debugger (moar and parrot) doesn't work. 10:41
10:42 SamuraiJack joined
ChoHag As soon as it hits a statement: 10:42
[bash]mking@egwene$ perl6-debug-p
> say 1
Start argument to substr out of range. Is: -1, should be in 0..0
The ensuing trace differs between moar and parrot, which I presume is expected. 10:43
jnthn ChoHag: Yes, somebody reported that as broken on latest yesterday. Nobody got to it yet. I'll get to it tomorrow if I'm not beaten to it.
jnthn suspects some change elsewhere mighta done it. 10:44
ChoHag Dear god no!
Bloody developers. Always changing things.
FROGGS yeah :/
jnthn Well, that's how development happens...
FROGGS I tell them if I got one
jnthn: psst!
we are just the suffering first line :o) 10:45
ChoHag Users on one side, devs on the other, manglers over all. It's a wonder I stay sane. 10:46
FROGGS *g*
10:48 rom1504 joined
FROGGS jnthn: it is about Version.new 10:48
jnthn FROGGS: "it"? 10:49
FROGGS the substr error
jnthn Oh
FROGGS m: say $*DISTRO.version 10:50
camelia rakudo-moar beb052: OUTPUT«v3.2.0.37.generic␤»
FROGGS m: say $*PERL.version
camelia rakudo-moar beb052: OUTPUT«vunknown␤»
FROGGS O.o
jnthn FROGGS: Now that we actually have perl6-debug in the Rakudo repo, we can include the proper version information in it, rather than the fake stuff (see hll-config in the source) it used when it was in another repo. 10:51
masak haha, "vunknown" :P
ChoHag masak: You can't laugh. ufo has bugs too. 10:53
<name> should be .name, and $path can't be .split because it's not a string.
I forget the line numbers.
FROGGS I remember that too 10:54
lizmat good *, #perl6! 11:04
FROGGS hi lizmat
lizmat seems the last change breaks building on parrot 11:05
gist.github.com/lizmat/420648c5184ba6389bc3
ChoHag What happened to the -X functions?
11:06 integral left
FROGGS -.- 11:06
ahh, it is just our configure line
11:06 integral joined
FROGGS we can steal from MoarVM there too 11:07
ChoHag Don't forget to make it build with PIC where it needs it.
11:12 fhelmberger left 11:13 fhelmberger joined 11:17 fhelmberger left 11:19 anaeem1 left, anaeem1_ joined 11:24 fhelmberger joined
colomon 163 modules passed the smoke test overnight. think that's the highest I've ever seen. 11:33
11:42 BeigeCougar joined 11:45 skids left
dalek kudo/nom: d5fdd24 | (Elizabeth Mattijsen)++ | src/core/Distro.pm:
Some $*DISTRO.release related tweaks
 11:47
11:47 BeigeCougar_ joined
lizmat wow, with these changes, I finally get the jvm spectest to an end: 11:47
Files=887, Tests=31414, 1256 wallclock secs ( 5.35 usr 2.92 sys + 17.99 cusr 4.33 csys = 30.59 CPU) 11:48
but what a weird amount of CPU used :-)
also: this was with TEST_JOBS=1 11:49
afk for a bit 11:50
11:50 carlin joined
masak ChoHag: thanks for the bug reports. hint: I'm more likely to get to them if they show up in github.com/masak/ufo/issues 11:51
ChoHag Yeah I have a fork and patch on the agenda. 11:54
masak najs. 11:57
11:57 BeigeCougar_ left
dalek rlito: f09fe6d | (Flavio S. Glock)++ | / (2 files):
Perlito5 - update bootstrap code
 12:00
12:00 immortal joined
dalek ast: 63d2915 | (Elizabeth Mattijsen)++ | S02-magicals/VM.t:
Make sure test count is ok on JVM
 12:01
ast: c6d7463 | (Elizabeth Mattijsen)++ | S02-magicals/VM.t:
Some more fudging for JVM
 12:06
lizmat really afk now 12:07
12:21 brrt left 12:22 thou joined 12:23 jnap joined 12:24 dmol left 12:25 rurban joined 12:26 thou left
[Coke] begins to wade through grumpy backscroll, stops, realizes that he should get a coffee first. 12:26
colomon o/ 12:29
specs seem to say that something like --counts-only is a valid command line argument for MAIN, but when I actually try it I get Usage: diff-report.pl [--counts-only=<Any>] <before> <after> 12:34
can work around by just making MAIN a multi, but this is less elegant than I expect from p6. :) 12:35
12:35 SamuraiJack left
donaldh try Bool 12:35
carlin panda's t/installer.t segfaults on moar, works with spesh disabled 12:37
colomon oh, no, the problem was that I was putting the --counts-only flag at the end of the command line. everything seems to work as expected if I put it at the beginning.
(reading more of the spec)++
woolfy wants to compliment all regular #perl6-people for ultimate proof of friendliness and helpfullness. I love you all with great huge love (I know, should be all-caps). 12:38
So #perl6++
colomon carlin: if you've got the time, figuring out what moar commit broke that would be handy. jnthn asked me if I could look into it, but I'm busy with $work. (Yes, coding scripts in p6 for work. :)
[Coke] is now caffienated… just in time for $dayjob. 12:39
happy UGT, everyone.
colomon carlin: I did trace it as far rakudo's commit 408ebb9738d468cb996a8b70ea9eae4629a5b1a7
carlin: so the problem moar commit is between 2180903 and g7a36ff2 12:40
carlin colomon: thanks, I'll see if I can find it 12:41
12:44 [Sno] joined
dalek ast: a599638 | (Elizabeth Mattijsen)++ | S02-magicals/pid.t:
Unfudge passing test on jvm
 12:53
lizmat seems all the tests doing "is_run" are now broken on rakudo.jvm 12:54
[Coke] lizmat++
O_o?
lizmat: presumably because of going through the eval server.
12:54 anaeem1_ left
lizmat perhaps: errors such as: 12:55
# got err: "Usage: org.perl6.nqp.tools.EvalServer [-bind-stdin] -cookie <name-for-access-token-file> -app <main-class>\n"
dalek kudo-star-daily: 7462a88 | coke++ | log/ (13 files):
today (automated commit)
[Coke] yup, that's new. 12:58
are you seeing these locally or in the daily roast output, OOC?
masak lizmat: there's backlog about that around 11:30 UTC+2 12:59
[Coke] rakudo.parrot is the only clean R*
donaldh lizmat: [Coke]: see backtrace
13:00 guru joined
[Coke] (oh wait, can't be daily roast, ecause I haven't posted one yet) 13:00
donaldh here forwards irclog.perlgeek.de/perl6/2014-06-10#i_8847148
13:00 guru is now known as Guest82028, xinming left 13:01 Guest82028 is now known as ajr_
donaldh simplest hack (I have a patch ready) is to modify perl6.execname within EvalServer.java 13:01
[Coke] lizmat: S02-magicals/pid.rakudo.parrot 1 todo PASSED: assigning to $*PID does not die
13:01 xinming joined
lizmat donaldh: waiting for your patch yhen 13:02
[Coke] integration/advent2009-day23.rakudo.parrot......... 1 skipped: lists aren't properly lazy in Rakudo yet [ I am dubious! ]
donaldh lizmat: do you see two failures in filestat.t ? Or are they maybe OSX specific?
lizmat [Coke]: that's fine, but rakudo.parrot doesn't even build for me atm 13:03
[Coke] only mentioned it because I think you just unfudged that for jvm.
(pulling that off yesterday's roast run)
lizmat donaldh: t/spec/S16-filehandles/filestat.t passes fine for me on OSX
(and jvm) 13:04
dalek p: 8a47b77 | (Donald Hunter)++ | src/vm/jvm/runtime/org/perl6/nqp/tools/EvalServer.java:
Work around perl6.execname being set to perl6-eval-server
donaldh hmmmm 13:05
donaldh thinks we can unfudge the IO-Socket-INET tests on JVM now that we have permgen and working is_run 13:06
lizmat donaldh: maybe we also need an NQP version bump?
donaldh lizmat: yes, that too :) 13:07
bbl 13:08
13:08 donaldh left
lizmat ok, what was the magic git incantation again to get the tag to be used for NQP_REVISION 13:15
lizmat promises to remember this time 13:16
moritz 'git tag' in nqp/
lizmat that gives me a list of tags ?
13:16 donaldh joined
lizmat LizyPro-2:nqp liz$ git tag 13:17
2011.05
2011.06
2011.09
etc.
jnthn git describe ?
lizmat that's the one
moritz erm, yes, sorry
lizmat jnthn++
moritz confused
jnthn git describe --match 2* # makes sure you get an appropriate tag
dalek kudo/nom: 54796d5 | (Elizabeth Mattijsen)++ | tools/build/NQP_REVISION:
Bump NQP revision to get donaldh++ jvm patches
 13:19
lizmat jnthn: just git describe gives the same, but that may not always be the case? 13:20
jnthn lizmat: There's a risk of getting the relesae's code name
Since both are the "most recent" tag
lizmat ok 13:21
13:21 pdcawley left 13:22 pdcawley joined, treehug88 joined
lizmat meanwhile, parakudo's build is still broken ... :-( 13:23
13:26 kaare_ left
dalek ast: 128464c | (Elizabeth Mattijsen)++ | S02-magicals/pid.t:
Unfudge for parrot
 13:30
lizmat [Coke]++
13:31 skids joined
lizmat is_run seems to be ok again now on jvm, donaldh++ 13:32
13:32 hummeleB1 joined 13:33 rurban left
donaldh just slow 13:34
13:35 rurban joined
lizmat don't cares about slow atm :-) 13:35
13:36 rurban left
lizmat
.oO( doesn't care about grammar either )
 13:36
donaldh needs to implement watch_path on JVM 13:38
13:39 ggoebel111116 joined
dalek kudo/nom: a9d9412 | (Elizabeth Mattijsen)++ | docs/ChangeLog:
Mention $*USER and $*GROUP
 13:41
13:43 SamuraiJack joined
dalek ast: f0e0cee | (Donald Hunter)++ | S32-io/IO-Socket-INET.t:
Unfudge INET tests on JVM.
 13:44
13:46 molaf joined 13:47 rurban joined
dalek rlito: ee86608 | (Flavio S. Glock)++ | / (3 files):
Perlito5 - js - fix autovivification in string scalar dereference # ${'VAR'}
 13:47
13:49 ggoebel111116 left, zengargoyle joined
moritz huh, yesterday m-spectest was a mess, today (on a different machine) it's all clean 13:52
13:55 ggoebel111116 joined, sqirrel joined, sqirrel left, BeigeCougar left, sqirrel joined
moritz still lots of fails on the other machine 13:57
colomon differences between the machines? 13:58
moritz failing machine: ubuntu 14.04 13:59
OK machine: Mint
13:59 erdic left
moritz OK machine has twice the RAM (8 vs. 4 GB) 13:59
FAIL machine has gcc 4.8.2, OK machien has gcc 4.7.2 14:00
colomon fires up a moar build on his ubuntu box 14:01
moritz tries with gcc 4.7 on the FAIL machine 14:03
FROGGS lizmat: does that help? gist.github.com/FROGGS/0d55f25d9932fc3eba1e
that is an nqp patch btw
lizmat FROGGS: will try
FROGGS thank you
14:04 kivutar joined 14:08 thou joined
lizmat seems to build with these changes, at least on OSX 14:08
14:09 sqirrel left
FROGGS nice 14:09
works for me also
14:10 bluescreen10 joined
dalek p: 69c7822 | (Tobias Leich)++ | Configure.pl:
fix building dyncall on osx

Dyncall v0.8 does not support flags like --target-x64 anymore.		 14:11
moritz hopes that doesn't break darwin 32 bit
dalek kudo/nom: 0a4adcc | (Tobias Leich)++ | tools/build/NQP_REVISION:
bump nqp rev for dyncall configure fix
 14:12
FROGGS moritz: then MoarVM would have been broken
lizmat JVM: Files=887, Tests=31413, 2243 wallclock secs ( 5.17 usr 2.85 sys + 18.28 cusr 4.28 csys = 30.58 CPU) 14:15
only 2 files with test failures
errands&
colomon moritz: ubuntu 12.10 with gcc-4.7.2 latest moar spectests clean 14:24
panda still fails 14:26
14:31 kaare_ joined 14:38 dmol joined 14:40 SamuraiJack left 14:42 bjz left 14:43 kaleem left
moritz still lots of FAIL on failhost and gcc-4.7.3 14:44
FROGGS moritz: what are the fails about? 14:45
moritz FROGGS: lots of diverse tests (S32-temporal, integration/advent*, S32-num, S32-io) dying with "No subtests run" 14:46
FROGGS hmmmm 14:47
moritz Unhandled exception: Cannot find method 'Stringy': no method cache and no .^find_method at <unknown>:1 (././CORE.setting.moarvm:print_exception:4294967295)
[Coke] looks like 7ca23fd killed the jvm spectest run for me. 14:49
can we revert that and go back to the first setting?
dalek rl6-roast-data: 2db3a33 | coke++ | / (5 files):
today (automated commit)
14:50 immortal left
[Coke] sorry, I didn't actually check that that particular commit was the problem. but it worked at 15002e3 14:50
FROGGS moritz: do you have more of that backtrace?
[Coke] looks like S02 passed 3053 tests, and then after that, everything aborted. 14:51
first failure was S02-magicals/VM.rakudo.jvm aborted 1 test(s)
rakudo.moar still clean!! 14:52
rakudo.parrot has 7 failures that need fudging and RTs.
FROGGS sergot: that should be interesting for you -----^ :o) 14:54
colomon didn't lizmat++ say this morning that parrakudo wouldn't build?
moritz FROGGS: perlpunks.de/paste/show/53971c55.7bff.d3 14:55
colomon "meanwhile, parakudo's build is still broken ... "
FROGGS colomon: I broke and fixed it today
colomon FROGGS++
14:56 brrt joined
FROGGS moritz: that is not very much :/ 14:56
something NULLish sneaks into the to be created exception
not really NULLish, but at least Muish
dalek rlito: d629c6a | (Flavio S. Glock)++ | / (3 files):
Perlito5 - js - fix autovivification in string dereference # push @{"VAR"}, $v
 14:59
15:03 fhelmberger left
sergot FROGGS: thanks :) 15:04
FROGGS sergot: you got that wrong... that means work for you :o) 15:05
15:05 mr-fooba_ joined
FROGGS but it is good to not start on the release day, instead having an overview a few days before is better 15:06
bbiab
rurban_ The new parrot 6.5.0 will probably have a new vtable method optimization for writers. (our gsoc project). many unneeded or duplicate write barriers can be omitted now, and a nested method call from the method into a _orig and then out to do a GC write barrier is now gone. rakudo-p tested fine. I'll look out for such WB opportunities with nqp and perl6 PMC's 15:07
[Coke] colomon: parrakudo worked for me as of...
colomon [Coke]: FROGGS says he fixed lizmat's bug already today.
[Coke] beb0525
rurban_ This branch: github.com/parrot/parrot/commits/r...ig2-gh1069 15:08
[Coke] rurban++ thanks.
If you make it a pull request, we can combine it with a bump to the latest parrot version required.
15:08 mr-foobar left
carlin looks like 6efc152b61 is the moar commit that broke panda's installer.t 15:08
This is perl6 version 2014.05-143-g408ebb9 built on MoarVM version 2014.05-10-g56a9b99 / ok 8 - moarvms will get installed after rest of the things 15:09
This is perl6 version 2014.05-143-g408ebb9 built on MoarVM version 2014.05-11-g6efc152 / Segmentation fault (core dumped)
rurban_ [Coke]: So we only tested for correctness. perl6 performance later. We expect something in between 2-4%, could be 5% or 0% also.
15:10 donaldh left
jnthn carlin: Interesting; thanks. 15:11
15:14 FROGGS left 15:16 donaldh joined 15:17 dmol left 15:21 bluescreen100 joined
[Coke] rurban_: someone's running benchmark stats, I presume they'll pick it up. 15:25
15:25 bluescreen10 left
rurban_ Our gsoc branch is not yet merged, still torture testing it 15:29
15:35 zakharyas left
timotimo p6weekly.wordpress.com/?p=175&s...725c065f6b - i'd really like some feedback 15:35
rurban_: i've mentioned your gsoc student't changes there, but i'm not sure if i described it correctly
also, i'd love more hints for interesting stuff that's happened :) 15:39
rurban_ Chirag Agrawal and Reini Urban had been improving parrot􏿽x92s pmc2c and PMCs by removing nested method calls and unneeded write barriers. It doesn􏿽x92t break the rakudo build, but performance measurements are still to be done.
timotimo that sounds good, thank you! 15:40
rurban_ PMC write barriers can now also be manually optimized
most of the parrot WB optimizations were done manually by looking at the control flow and remove duplicates
timotimo ah, hm 15:41
rurban_ See github.com/parrot/parrot/commit/46...2efdbc1eb5
I'm just benching rakudo-p now. parrot master against the new branch 15:43
So if you can wait 1639 secs :) 15:44
parrot -O2 was also not benched yet with rakudo-p
timotimo i can wait that long! 15:46
16:00 fhelmberger joined
dalek osystem: 0fa7aa8 | Matt++ | META.list:
Added the start of the Stats modules
 16:04
Ulti timotimo you could add that ^ it's not exactly great code and doesn't do much yet :S
it does however exist which is better than non existant code
*maybe* 16:05
timotimo ah, you've been talking about that for a while 16:06
Ulti well asking for help in doing it for a while, it was originally part of my bioinfo modules but was getting kind of weird to have it in there
timotimo mhm 16:07
Ulti not really ready to put that up yet, especially as the parallel parsing is super buggy and some of the time segfaults moar
timotimo oh 16:08
mhm
ChoHag I'm playing with threads in moar and the stdout/err of the (only) subthread is being squashed by the output of the parent thread. 16:09
In fact in linux the subthread's output doesn't appear at all. In freebsd it's random. 16:10
Ulti I'm on OSX which I guess is most similar to freebsd
16:10 denis_boyun_ joined
Ulti haven't really been printing to STDOUT in parallel though more building lists of objects 16:11
16:11 pdcawley left, pdcawley joined, treehug8_ joined
timotimo hm, there's a lock in there that'll serialize accesses to the stdout and such 16:11
not exactly sure why it'd omit output from subthreads 16:12
ChoHag Yeah I don't know that I plan to do that, but I'm just building the shape now with says all over the place.
16:12 fhelmberger left
timotimo can you paste code that shows that problem? 16:12
16:13 treehug88 left
ChoHag What's the channel's favourite paster? 16:13
moritz ChoHag: gist.github.com 16:14
timotimo aye, our evalbot can run code directly off of gist (but not secret gists)
16:15 FROGGS joined
ChoHag gist.github.com/anonymous/1606248d28b4f1430e88 16:16
timotimo hm, not actually sure what :every(0) will do 16:17
16:17 brrt left
ChoHag It's just multiplied by 1000 and passed to nqp::timer 16:19
timotimo is there a reason you're building your own TPS and then cue instead of just using a start block?
ChoHag Playing. 16:20
That is literally all of the code I have written so far. I'm still exploring what fits best.
timotimo OK 16:21
ChoHag And, as ever, don't quite know what I hope to achieve.
16:21 molaf_ joined
ChoHag I'd be a terrible project manager. "Just do the thing that's in my head, incomplete." 16:21
timotimo :)
PerlJam ChoHag: that's what I tell my minions! 16:22
ChoHag Does it work?
PerlJam ChoHag: oh, and I change my mind every so often too just to keep things lively. 16:23
ChoHag: It would work better if I had better minions.
ChoHag Well of course. The picture's incomplete and thus, when examined in too much detail, naturally rube-golber-esque.
PerlJam (so, no, it does not work)
FROGGS timotimo: I feedbacked your draft
ChoHag Then fixes that makes the inconsistency just pop up in a previously completed part. 16:24
timotimo FROGGS: thanks :)
16:24 molaf left
ChoHag fixing. My grammar's terrible sometimes. 16:24
timotimo: I remember now. I had a vague idea about controlling $!max_threads 16:25
timotimo ah
FROGGS jnthn: I get this in v5: Cannot find method 'package_at_key': no method cache and no .^find_method 16:26
timotimo well, if you get a "my $*SCHEDULER", you can use "start" and friends with your own scheduler
ChoHag Oh that's easier.
Can I just do that at the top of the class?
FROGGS jnthn: is that like the err message about .item where it should say instead that I have to declare the sub signature with Mu?
16:28 Rotwang joined
timotimo ChoHag: since $* are dynamic variables, it's important that it's "outside" whatever it's influencing "on the stack" 16:29
jnthn FROGGS: Perhaps so
timotimo that was a terrible description.
you should just set it in the BUILD i think
ChoHag Interestingly if I don't say anything except the two notes (or says) inside the inner-most loop, nothing is said under any circumstances. Even randomly in freebsd.
16:30 donaldh left
ChoHag If only the inner thread communicates, and starts doing so before the loop{}, its output gets through entirely. 16:31
timotimo huh. perhaps the thread actually threw an exception 16:32
and i have no idea what your :catch thing does
ChoHag Not work.
It's an exploratory wart. 16:33
timotimo it could very well be that you gave the job a thing that's supposed to be called when an exception happens, you give it a variable without a value and then somehow you cause an exception to be thrown 16:34
16:34 jnap left
jnthn The thing given to catch is supposed to be some piece of code that takes the exception and does something with it. 16:35
16:36 jnap joined 16:38 jnap1 joined, jnap left, atta left
ChoHag Ah I think that's it. 16:39
Looks like it's trying to instantiate the Channel inside the thread (or in both threads).
Unhandled exception in code scheduled on thread 34504231936
Internal error: inconsistent bind result
At $!c.receive
If I create $!c rather than relying on the lazy default, the problem seems to go away. 16:40
timotimo ah, it may do that after BUILD is run through? 16:41
in which case the thread will be unhappy if it trips over that
ChoHag Could be. 16:42
I can't remember if it happened before BUILD became BUILD.
Yes changing it back to a hand-called method seems to kill the problem too. 16:43
Well the thread shouldn't start running at BUILD time anyway. 16:44
timotimo yes, that sounds like a good idea :) 16:45
16:54 raiph joined 16:55 vendethiel joined 16:58 ajr_ left 16:59 ajr joined 17:00 ajr is now known as Guest15761, Guest15761 is now known as ajr_
timotimo TBH, i'd have written that code completely without the class, but that's just me :) 17:02
ChoHag Well yes. I considered removing it before pasting.
But then who wants code which does that?
It's a tad incomplete. 17:03
17:03 dakkar left 17:07 treehug88 joined 17:08 treehug8_ left 17:11 BeigeCougar joined 17:15 BeigeCougar left 17:16 brrt joined, spider-mario left 17:18 brrt left
timotimo rurban_: were you successful with the benchmarking yet? 17:19
17:22 Alina-malina left, spider-mario joined 17:24 jpwillis_ joined
flussence D uses Sets for something pretty clever: dlang.org/phobos/std_uni.html#Synopsis 17:25
17:25 BeigeCougar joined 17:27 BeigeCougar left, ajr_ left 17:30 rurban left, jnap1 left 17:32 rurban joined, rurban left 17:33 kurahaupo left
rurban_ timotimo: was in a meeting. I got 1642s down to 1639s, so no big deal yet 17:33
will try later when parrot PMC's we are better
17:38 erkan left 17:43 dmol joined 17:44 Alina-malina joined 17:46 treehug88 left
TimToady timotimo: you could add my two STD patches, also as lhf to move to rakudo 17:51
17:58 rurban joined 18:00 user6 joined, telex left 18:01 Alina-malina left 18:02 Alina-malina joined, rurban left, user6 left, telex joined 18:03 hoverboard joined, rurban joined 18:04 rindolf joined
rurban_ timotimo: Ulti? Not Util? 18:07
ChoHag Does perl 6 have method modifiers (before/after/etc) like Moose? 18:08
FROGGS ChoHag: no, but you can wrap methods 18:09
18:09 rurban left
jnthn Also, it's not hard to implement them in module space, given there's a MOP 18:11
Also there's deferal (nextsame/callsame/nextwith/callwith) that fill many of the things you might use method modifiers for.
ChoHag How does wrap work?
Specifically, what's the difference between those 4? 18:12
How do you call wrap, for that matter? 18:13
FROGGS m: sub wrapper(\a) { say "before"; callwith(|a); say "after" }; sub foo($a) { say "foo $a" }; my $wrapped = &foo.wrap(&wrapper); foo(42) 18:17
camelia rakudo-moar 0a4adc: OUTPUT«before␤foo 42␤after␤»
18:17 Mouq joined
FROGGS m: sub wrapper(\a) { say "before"; callwith(|a); say "after" }; sub foo($a) { say "foo $a" }; &foo.wrap(&wrapper); foo(42); &foo.unwrap; foo(21) 18:17
camelia rakudo-moar 0a4adc: OUTPUT«before␤foo 42␤after␤Not enough positional parameters passed; got 1 but expected 2␤ in method unwrap at src/gen/m-CORE.setting:3340␤ in block at /tmp/uuEhz8a6xM:1␤␤»
FROGGS m: sub wrapper(\a) { say "before"; callwith(|a); say "after" }; sub foo($a) { say "foo $a" }; &foo.wrap(&wrapper); foo(42); &foo.unwrap(&wrapper); foo(21)
camelia rakudo-moar 0a4adc: OUTPUT«before␤foo 42␤after␤Cannot unwrap routine: invalid wrap handle␤ in method unwrap at src/gen/m-CORE.setting:3341␤ in block at /tmp/heC8HUA7mZ:1␤␤»
FROGGS ahh
m: sub wrapper(\a) { say "before"; callwith(|a); say "after" }; sub foo($a) { say "foo $a" }; my $w = &foo.wrap(&wrapper); foo(42); &foo.unwrap($w); foo(21) 18:18
camelia rakudo-moar 0a4adc: OUTPUT«before␤foo 42␤after␤foo 21␤»
ChoHag Where are the meanings of \a and |a (and callwith etc.) defined/explained?
FROGGS S06:1235 18:19
synopsebot Link: perlcabal.org/syn/S06.html#line_1235
FROGGS err
ChoHag Close enough.
FROGGS perlcabal.org/syn/S06.html#Wrapping
18:20 pippo joined
FROGGS btw, is a QAST::WVal(Array) legit? I see one in a QAST::Stmts, even in perl6-m --target=ast -e 1 18:25
18:25 rurban joined 18:26 itz joined
pippo o/ #perl6 18:26
FROGGS hi pippo 18:27
18:27 kivutar left
pippo hi FROGGS 18:30
18:32 hoverboard is now known as moistcherry 18:49 itz left 18:51 Mouq left 18:52 guru joined, itz joined 18:53 guru is now known as Guest8194
pippo m: grammar G {rule TOP {^AA' '<one> <two> <three>$}; rule one {[\w+' ']*}; rule two {[\w+' ']*}; token three {[\w+' ']+}}; say G.parse("AA 111 11 11 3 3 333 "); 18:53
camelia rakudo-moar 0a4adc: OUTPUT«(Any)␤»
18:53 Guest8194 is now known as ajr_
pippo m: grammar G {rule TOP {^AA' '<one> <two> <three>$}; rule one {[\w+' ']*}; rule two {[\w+' ']*}; token three {[\w+' ']+}}; say G.parse("AA 111 11 11 2 3 3 333 "); 18:53
camelia rakudo-moar 0a4adc: OUTPUT«｢AA 111 11 11 2 3 3 333 ｣␤ one => ｢111 11 11 ｣␤ two => ｢2 ｣␤ three => ｢3 3 333 ｣␤␤»
pippo Anybody can explain me why the first does not work? It should since rule <two> is allowed to be empty. No? 18:54
18:55 moistcherry is now known as hovercraft 18:57 atta joined 18:58 kivutar joined
geekosaur but two is not actually empty in that case, since the trailing space is tied to the token so the extra spacing fails both two and three? 19:03
you want the spaces in TOP instead of the rules, I think 19:04
19:05 dwarring joined
geekosaur lemme rephrase that as I am not using token and rule consistently there 19:05
the extra spacing is only parsed if you parsed a \w+
19:05 kivutar left
geekosaur so if rule two does not match, your extra spacing interferes with token three matching 19:06
19:07 kivutar joined 19:09 Mouq joined
Mouq o/ #perl6 19:11
yoleaux 24 May 2014 05:42Z <lizmat> Mouq: I'll add tests then
Mouq Thanks lizmat!
19:11 darutoko left
Mouq What's the usual --prefix people on Debian/Ubuntu use for their Perl 6 configuration, if any they use one? 19:12
s/any//
Mouq just goes with /usr/local for now 19:14
19:15 sqirrel joined 19:16 thrakcattack joined
pippo geekosaur: but in the TOP rule the space between <two> and <three> should be getting those? 19:20
geekosaur: no? 19:21
19:21 hovercraft is now known as siriustar
Mouq pippo: I believe <one> and <two> eat up the "111 11 11 " and "3 3 333 " respectively 19:24
pippo: And then <three> doesn't have anything to match
pippo: I'm not sure if it's supposed to backtrack in this case
pippo o/ Mouq. I have put rules in there to allow backtracking indeed... 19:25
o/ Mouq. I have put rules and not tokens in there to allow backtracking indeed...
Mouq pippo: Perhaps you want 'regex' instead
PerlJam pippo: neither rule nor token backtrack
Mouq You can try regex TOP { :s ... } if you want sigspace 19:26
pippo OH! So do I have to declare TOP as a regex? I'll try... 19:27
PerlJam pippo: or use :!ratchet if you declare it as a rule.
Mouq m: say "aabbccd" ~~ token { b .* cc } 19:29
camelia rakudo-moar 0a4adc: OUTPUT«Nil␤»
Mouq m: say "aabbccd" ~~ token { :!ratchet b .* cc }
camelia rakudo-moar 0a4adc: OUTPUT«｢bbcc｣␤␤»
Mouq m: say "aabbccd" ~~ token { :!r b .* cc }
camelia rakudo-moar 0a4adc: OUTPUT«｢bbcc｣␤␤»
Mouq Sweeet
I expected NYI
PerlJam (honestly I wasn't sure myself, but I felt there was a > 60% chance it was implemented :) 19:30
pippo m: grammar G {rule TOP {:!r ^AA' '<one> <two> <three>$}; rule one {[\w+' ']*}; rule two {[\w+' ']*}; token three {[\w+' ']+}}; say G.parse("AA 111 11 11 3 3 333 "); 19:31
camelia rakudo-moar 0a4adc: OUTPUT«(Any)␤»
pippo m: grammar G {rule TOP {:!r ^AA' '<one> <two> <three>$}; rule one {:!r [\w+' ']*}; rule two {:!r [\w+' ']*}; token three {[\w+' ']+}}; say G.parse("AA 111 11 11 3 3 333 ");
camelia rakudo-moar 0a4adc: OUTPUT«(Any)␤»
pippo :-( 19:32
geekosaur afaik you need an explicit space the same way one is already in TOP? note that ' '
and similar in rule two and token three
unquoted space is not relevant as part of a pattern; perl 6's regexes are like perl 5's /x regexes, except that things like ' ' work to quote (in perl 5 you have to use \s) 19:34
basically perl 6 regex is an actual language
Mouq geekosaur: Not in the case of 'rule'
pippo: I don't know what the problem is there... 19:35
Mouq needs to figure out if there's a way to do copy+paste in xterm without a mouse wheel...
19:37 anaeem1 joined
geekosaur mouq: default for xterm is left or right to select, middle to paste. it is difficult to work around this 100% sanely, but blog.ssokolow.com/archives/2011/05/...rms-paste/ 19:39
19:39 mr-fooba_ left
geekosaur hope that url's still valid... 19:39
probably have to modify it if you want it to use PRIMARY that way 19:41
Mouq geekosaur: Thanks! :)
PerlJam pippo: what exactly are you trying to do btw?
Mouq m: "AA 111 11 11 3 3 333 rule {^AA [\w+' ']* 19:42
camelia rakudo-moar 0a4adc: OUTPUT«===SORRY!=== Error while compiling /tmp/iEGqZw1BCk␤Two terms in a row␤at /tmp/iEGqZw1BCk:1␤------> 111 11 11 3 3 333 rule {^AA [\w+' ']*⏏<EOL>␤ expecting any of:␤ postfix␤ infix stopper…»
Mouq Ackl
m: "AA 111 11 11 3 3 333 " ~~ rule {^AA [\w+' ']* [\w+' ']* [\w+' ']+$} 19:43
camelia ( no output )
Mouq m: say "AA 111 11 11 3 3 333 " ~~ rule {^AA [\w+' ']* [\w+' ']* [\w+' ']+$}
camelia rakudo-moar 0a4adc: OUTPUT«｢AA 111 11 11 3 3 333 ｣␤␤»
Mouq m: say "AA 111 11 11 3 3 333 " ~~ rule {^AA (\w+' ')* (\w+' ')* (\w+' ')+$}
camelia rakudo-moar 0a4adc: OUTPUT«｢AA 111 11 11 3 3 333 ｣␤ 0 => ｢111 ｣␤ 0 => ｢11 ｣␤ 0 => ｢11 ｣␤ 1 => ｢3 ｣␤ 1 => ｢3 ｣␤ 2 => ｢333 ｣␤␤»
ChoHag Mouq: left+right button simultaneously simulates the middle button. 19:48
Traditionally.
geekosaur that needs to be configured in xorg, it's not a default 19:49
(need an xorg.conf)
ChoHag xorg have raped tradition to the point of stockholming it.
geekosaur er? that was always an option that had to be enabled separately 19:50
(except in old versions where chording wasn't supported at all)
and this goes all the way back to X386
19:51 siriustar is now known as hoverboard 19:52 lue left 19:58 araujo left 19:59 araujo joined
FROGGS \o/ 20:00
timotimo so ... i published the post
FROGGS v5 written in pure Perl 6 can now say $^O again!
timotimo thank you all for your input
FROGGS it cannot say string now though, but Ints and Rats do work
and... I have to turn off the optimizer for some weird reason 20:01
colomon timotimo: url?
20:01 sqirrel left
Mouq timotimo++ FROGGS++ \o/ 20:02
flussence (Last time I encountered a ChordMiddle option was in a log message stating it'd been auto-disabled because I clicked the real middle button...)
20:02 Mouq left 20:03 Mouq joined
Mouq geekosaur++ 20:04
pippo PerlJam: I am parsing a file whre there are no column delimiters. Sometime one columnt is empty. The first column is the one following 'AA '. The last one is the one before the end of line. The second one is between the first and the third. Only the last column is never empty.
geekosaur huh, so they did make it default? nice of them 20:05
Mouq pippo: Do the columns have a specific length? 20:06
*per column
timotimo colomon: p6weekly.wordpress.com/ as always :) 20:07
Mouq perl6 -e 'say $*DISTRO; say $*KERNEL' 20:08
linux (3.2.0.4.amd.64)
use of uninitialized value of type Version in string context
use of uninitialized value of type Str in string context
()
pippo Mouq: no. They contain text that is interleaved by a single space.
colomon timotimo++
dalek ast: d659f4c | (David Warring [email@hidden.address] | integration/advent2011-day10.t:
reorder tests
ast: 12e781e | (David Warring [email@hidden.address] | integration/advent2011-day11.t:
adding 2011 advent day 11
Mouq Not sure what's wrong with $*KERNEL here
20:08 kaare_ left
PerlJam pippo: there's always 3 columns or 2 columns? 20:10
timotimo i don't know if it's the weather or something, but i feel pretty drained
20:11 woolfy1 joined
pippo PerlJam: Only one column is never empty the last one. But for parsing I will consider a line such this one as having three clomuns the first are empy. 20:12
PerlJam: Only one column is never empty the last one. But for parsing I will consider a line such this one as having three clomuns the first two are empy.
20:13 lue joined, woolfy left
lizmat Mouq: what OS are you on? 20:13
timotimo for example, it would be absurdly easy to just bind a few more of the path operations for cairo and build some example code with that to create some pretty images 20:14
but i can't bring myself to do it
PerlJam pippo: I was going to advocate a simple approach using split rather than a grammar/regex. But I guess you'd have difficulty distinguishing between the case when there's only no column 1 and when there's only no column 2
lizmat timotimo: it's not just you
pippo PerlJam: I have these configs:
1 2 3
1 3 20:15
2 3
20:15 itz left
lizmat Mouq: seeing the same thing here, investigating 20:15
20:16 itz joined
Mouq m: grammar G {rule TOP {^AA' '<one>' '{say "here"}<two>' '<three> $}; token one {[\w+] *% ' '}; token two {[\w+] *% ' '}; token three {[\w+] *% ' '}}; say G.parse("AA 111 11 11 3 3 333 ") 20:16
camelia rakudo-moar 0a4adc: OUTPUT«here␤｢AA 111 11 11 3 3 333 ｣␤ one => ｢111 11 11｣␤ two => ｢｣␤ three => ｢3 3 333｣␤␤»
lizmat m: say $*KERNEL 20:17
camelia rakudo-moar 0a4adc: OUTPUT«use of uninitialized value of type Version in string context␤use of uninitialized value of type Str in string context␤ ()␤»
dalek ast: 94985a8 | (David Warring [email@hidden.address] | integration/advent2011-day11.t:
tighten test to work at compile time
lizmat m: say $*KERNEL.name; say $*KERNEL.version; say $*KERNEL
camelia rakudo-moar 0a4adc: OUTPUT«linux␤v30.Ubuntu.SMP.Wed.May.1.16.35.23.UTC.2013␤linux (30.Ubuntu.SMP.Wed.May.1.16.35.23.UTC.2013)␤»
pippo I can distinguish between 1 3 and 2 3 because column one starts just after 'AA '. If there is nothing starting just after it means it is column 2.
lizmat Mouq: Systemic.gist was too lazy, fixing and adding tests 20:18
Mouq lizmat++
pippo Mouq: Your example works!! 20:19
Mouq lizmat: Oh, sorry, missed earlier question. Debian
20:19 itz left
lizmat turns out, that wasn't important 20:19
subtle difference between $!name and $.name :-) 20:20
Mouq lizmat: Ahh :)
lizmat .gist was using the attribute directly ($!name) instead of calling the accessor ($.name)
vendethiel I really have a hard time with $!. Same with ? ? in ruby ...
Mouq pippo: Hopefully that's extensible enough for whatever you're doing ^_^ 20:21
lizmat $.name is short for self.name
$!name accesses the attribute directly
PerlJam vendethiel: good! That's to prod you to use the accessor :)
vendethiel PerlJam: that doesn't seem very relevant
pippo Mouq: but... uses a fixed space between one and two and three. :-( 20:22
lizmat vendethiel: in this case it was: the attribute is lazily filled by the accessor method
in the $*KERNEL case
20:22 treehug88 joined 20:23 Rotwang left
lizmat the gist method was directly accessing the attribute, so it didn't get filled 20:23
and produced nonsense and warnings
vendethiel lizmat: ?? 20:24
I just mean I think $! is hard to read.
lizmat ah, ok -)
Mouq m: grammar G {rule TOP {^AA' '<one> <two>?? <three> $}; token one {[\w+] *% ' '}; token two {[\w+] *% ' '}; token three {[\w+] *% ' '}}; say G.parse("AA 111 11 11 3 3 333 ")
camelia rakudo-moar 0a4adc: OUTPUT«｢AA 111 11 11 3 3 333 ｣␤ one => ｢111 11 11｣␤ three => ｢3 3 333｣␤␤»
timotimo in that case, just write it as nqp::getattr(self, $?CLASS, '$!field')
... %)
20:25 oetiker left
lizmat should we really encourage nqp::getattr usage outside of the core ? 20:25
that's pretty implementation specific....
dalek kudo/nom: f05e57b | (Elizabeth Mattijsen)++ | src/core/Systemic.pm:
Make .gist/.Str allow lazy setting of attributes
 20:27
jnthn No, don't use nqp::getattr outside of core...
Mouq m: class C { has $!attr; method test { say MY::.keys } }; C.new.test
camelia rakudo-moar 0a4adc: OUTPUT«self %_ $! $/ $_ $*DISPATCHER &?ROUTINE RETURN␤»
pippo Mouq: What I am trying to get is two => ｢｣. Why I cannot get that? :-) 20:28
timotimo lizmat: i was not serious at all :) 20:29
20:29 oetiker joined
Mouq pippo: Because you're trying to make a regex that is greedy and lazy at the same time 20:30
pippo: Or at least that's how I see the problem
pippo Mouq: So there is no way out here?!! :-(( 20:32
PerlJam pippo: you could also turn the problem into something more procedural: get and remove the thing at the beginning, get and remove the thing at the end and whatever is left is thing 2.
20:33 aindilis` left, Tene left, cxreg left, Tene joined, Tene left, Tene joined, zamolxes joined, mhasch joined
PerlJam That sounds like a couple of s/// and a few calls to trim 20:33
20:33 nwc10 joined, xiaomiao joined, cxreg joined, Grrrr joined, ashleydev joined, logie joined, cosimo joined
Mouq pippo: I'm sure there is :) it's just tricky 20:33
20:33 yeltzooo joined 20:34 itz joined
pippo Mouq: PerlJam: geekosaur: Thank you very much for your help and patience. I'll try to 20:34
Mouq: PerlJam: geekosaur: Thank you very much for your help and patience. I'll try your suggestions.
20:36 kst` joined 20:37 thrakcattack left 20:39 breinbaa1 joined
dalek ast: 42a632d | (Elizabeth Mattijsen)++ | S02-magicals/ (4 files):
Test Systemic features a bit more
 20:39
20:40 kivutar left
pippo Mouq: PerlJam: is it possible to set two => ｢｣ if it doesnt match (whith a code block)? For example in the Mouq code above where we have <two>??. 20:41
itz oh "xanadu" is finally out 20:42
lizmat remembers reading the book way back in the seventies 20:43
20:43 pecastro left
itz I have the late 80s Microsoft Press (!) reprint 20:44
dalek kudo/nom: c8e91cb | (Elizabeth Mattijsen)++ | t/spectest.data:
Add advent2011-day11 tests
20:45 dayangkun_ joined 20:47 molaf_ left
PerlJam m: grammar G {rule TOP {^AA" "<one> <two>?? <three> $ {make [~$<one>,~($<two>//""),~$<three>]} }; token one {[\w+] *% " "}; token two {[\w+] *% " "}; token three {[\w+] *% " "}}; say G.parse("AA 111 11 11 3 3 333 ").made.perl; 20:52
camelia rakudo-moar 0a4adc: OUTPUT«["111 11 11", "", "3 3 333"]␤»
PerlJam pippo: you could do something like that.
dalek ast: 6dafb80 | (David Warring [email@hidden.address] | integration/advent2011-day11.t:
test trusted vs untrusted inheritence
 20:53
ast: 37ce61e | (David Warring [email@hidden.address] | S02-magicals/ (4 files):
Merge branch 'master' of github.com/perl6/roast
20:54 treehug88 left, rindolf left, jpwillis_ left, raiph left, bluescreen100 left, [Sno] left, carlin left, virtualsue left, kst left, muraiki left, dayangkun left, yakudzo left, breinbaas left 20:57 jpwillis1 joined 20:58 carlin joined, lizmat_ joined, woolfy1 left 20:59 Mouq left, lizmat left 21:00 woolfy joined
pippo PerlJam: Wow! It what I need. Thank you very much! :-) 21:00
PerlJam: Wow! It is what I need. Thank you very much! :-)
21:03 brrt joined 21:06 donaldh joined, vendethiel left, skids left, itz left 21:08 treehug88 joined, raiph joined, bluescreen100 joined, [Sno] joined, virtualsue joined, muraiki joined, yakudzo joined
dalek kudo/nom: 1a3706c | (Elizabeth Mattijsen)++ | lib/lib.pm6:
Make sure 'use lib @dirs' works correctly
 21:11
21:11 Mouq joined
dalek ast: 287670c | (David Warring [email@hidden.address] | integration/advent2011-day11.t:
braino
 21:12
21:21 anaeem1 left
lizmat_ is trying to grok what our sub EXPORT is supposed to return 21:21
21:22 lizmat_ is now known as lizmat
lizmat FROGGS: any pointers? 21:22
the specs are not really clear in it
timotimo it's supposed to return a hash
well, a stash really
lizmat but what would be the keys and values ?
timotimo the keys are the names for symbols to be installed, the values are things to be installed under those symbols
cognominal what the diff between a hash and stash? 21:24
lizmat that's what I thought, but I can't seem to get it to work 21:25
timotimo i think stashes are immutable?
i think stashes are an implementation detail
rurban_ [Coke]: what is driving the aprrot blog? Django? 21:27
parrot
Ah drupal
[Coke] sounds right. 21:29
21:35 ajr_ left 21:39 woolfy1 joined
lizmat sits on the terrace, listening to the rain and the thurderstorm that is getting closer and closer 21:39
brrt has been waiting for rain and thunderstorm to no avail for 3 days now 21:40
geekosaur ran is here now; thunderstorms more likely tomorrow 21:41
brrt's was last weke here
*week
21:41 woolfy left
brrt well, i'm still waiting 21:42
21:47 brrt left
lizmat .tell FROGGS does it make sense to have CUR::Local::File instances if which the directory does not exist ? 21:47
yoleaux lizmat: I'll pass your message to FROGGS.
21:48 xragnar left 21:49 xragnar joined
lizmat .tell FROGGS or CUR::Local::Installation for that matter ? 21:51
yoleaux lizmat: I'll pass your message to FROGGS.
21:53 treehug88 left 21:54 hummeleB1 left 21:57 bluescreen100 left
pippo good night #perl6 21:59
21:59 pippo left
Mouq \o pippo! 22:07
22:07 lizmat left
Mouq lizmat: A directory may not exist at the time of `use lib` but could at the time of `use Foo` afterward 22:09
rurban_ We also identified several unneeded write barriers in 6model and are removing it now for the new parrot 6.5.0. Could be quite a perf. improvement 22:15
22:18 lizmat joined 22:20 skids joined
lizmat Mouq: hmmm... "use" is compile time: does it make sense to assume a directory will appear out of thin air during runtime? 22:22
and if so, wouldn't that be a serious potential security hole
Mouq lizmat: I have no idea 22:23
rurban_: Woo!
rurban_ new parrot numbers in a minute
dalek ecs: e8d23f6 | (Elizabeth Mattijsen)++ | S99-glossary.pod:
Add "mischan" lemma
 22:24
lizmat fwiw, I feel that whenever a CompUnitRepo::Local object is created, it should *fail* when the directory to which it applies, does not exist or is not readable 22:25
22:29 SevenWolf joined
timotimo rurban_: very good :) 22:30
22:33 spider-mario left 22:41 [Sno] left, [Sno] joined 22:44 hoverboard left
Mouq .ask FROGGS With your slang example, what is matched by `use COBOL; test ` is `; test `. Is the inclusion of the semicolon intentional? 22:46
yoleaux Mouq: I'll pass your message to FROGGS.
timotimo wow, quite loud thundering outside 22:47
www.blitzortung.org/Webpages/index....;page_0=30 - a fun website
jpwillis1 I can't stop moofing. 22:49
raiph i've quit moofing for a mo and have an unrelated question
yoleaux 9 Jun 2014 08:53Z <moritz> raiph: re perl6/p5-modules: you could contact the authors of the invidial modules, tell them to fork the repo (or extract subdirs) if they're still interested
9 Jun 2014 09:17Z <moritz> raiph: i've updated rakudo.de a bit
22:50 grondilu joined
raiph .tell moritz thanks re gh/perl6 repo message 22:50
yoleaux raiph: I'll pass your message to moritz. 22:51
grondilu is a bit intrigued/amuzed to read about lots of storms in US as he noticed lots of storms here in France as well lately.
cognominal orage, ho désespoir 22:52
neige donc! 22:53
raiph cognominal: i'll raise your neige a moof
grondilu looks up moof in the wiktionary 22:54
geekosaur expect more and more violent storms
grondilu founds nothing
geekosaur knows it as the sound a dogcow makes
(apple in-joke of sorts)
en.wikipedia.org/wiki/Dogcow 22:56
geekosaur notes the alternative meaning however... 22:57
jnthn haha 22:58
I just noticed that :)
rurban_ nqp-p is now 4.3% faster in the new pmc2c_orig branch
for nqp: perf stat -r2 prove -r --exec "./nqp-p --vmlibs=nqp_dyncall_ops" t/nqp t/hll t/qregex t/p5regex t/qast t/serialization 22:59
timotimo i don't know what moofing is :(
23:00 virtualsue left
dalek p/pmc2c_orig: bb6a1c6 | rurban++ | src/vm/parrot/pmc/ (7 files):
pmc2c_orig: optimize 6model pmcs for the new pmc2c

Annotate several PMC methods for unneeded or manual GC write barriers This requires pmc2c from parrot 6.5.0 (i.e. rurban/pmc2c_orig2-gh1069)
This is ~4.36% faster for the nqp testsuite.
  $ perf stat -r2 prove -r --exec "./nqp-p --vmlibs=nqp_dyncall_ops" t/nqp t/hll t/qregex t/p5regex t/qast t/serialization		 23:01
tadzik wow, rurban_++ :) 23:04
23:04 donaldh left
rurban_ testing now rakudo speedup 23:04
timotimo way cool, ~4.5% is pretty nice
rurban_ 6model had several unnecessary GC write barrier calls 23:05
mostly duplicate WBs in nested method calls 23:06
in tail position. so could remove that now with the new pmc2c
23:13 rurban left
timotimo sounds great! 23:13
23:23 rurban joined 23:25 Mouq_ joined 23:27 Mouq left 23:31 jpwillis1 left 23:32 Mouq_ is now known as Mouq 23:34 rindolf joined 23:44 hoverboard joined 23:47 carlin left
timotimo i'm not sure about this "why perl didn't win" blog post 23:52
perl6 didn't really have such a bad influence on perl5, did it?
23:56 denis_boyun_ left 23:59 rindolf left
moritz no, it's just a convenient scape goat 23:59
yoleaux 22:51Z <raiph> moritz: thanks re gh/perl6 repo message

Powered by Cro and the Raku® Programming Language.

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

Thank you!