»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.
jnthn 'night, #perl6 00:01
BenGoldberg n: say 'alive' 00:02
camelia niecza v24-108-g17d73e4: OUTPUT«(timeout)[auto-compiling setting]␤»
00:06 jeffreykegler left, btyler left, lizmat joined 00:07 woolfy joined 00:08 beastd left 00:13 tgt left 00:24 dayangkun left 00:25 dmol left
dalek : 4983aa3 | (Elizabeth Mattijsen)++ | misc/perl6advent-2013/schedule:
sync titles of post to reality
 00:37
lizmat sadly, no one for tomorrow just yet
00:41 ajr_ left
dalek ast: 476ad78 | (Elizabeth Mattijsen)++ | S17-concurrency/ (2 files):
Some additional tests and variance in expressivity
 00:52
lizmat some leftovers while working on short p6 presentation at last Amsterdam.pm meeting 00:53
00:55 stevan_ joined 00:57 stevan_ left, stevan_ joined 01:00 rurban1 left 01:02 dayangkun joined 01:10 woolfy left
lizmat asleep in Paris& 01:17
01:25 rurban1 joined
dalek rl6-roast-data: 4f6d4ac | coke++ | / (4 files):
today (automated commit)
 01:33
01:33 zakharyas left
japhb__ OK, now there's an error I've never seen before: 01:59
$ git describe
warning: tag 'Malmö' is really 'Malmo􏿽xCC􏿽x88' here
Malmö-22-g262e600
japhb__ wonders how that even happens
geekosaur sees differences... normalized utf8 02:00
japhb In the right font, those do look different, BTW -- the dots are higher on the right in my terminal.
The thing that mystifies me is why the tag is being treated as anything but a bag of bytes. 02:01
geekosaur they're offset slightly here on the right, making me think that one uses a combining character instead of the unified one
japhb__ Most likely, yeah.
Still wondering how that happened in the first place .... 02:02
japhb I know, let's just insist on ASCII-only git tags! >.<
02:05 xinming left
japhb__ Has anyone seen this error? 02:06
./parrot -o runtime/parrot/include/parrotlib.pbc runtime/parrot/library/parrotlib.pir
error:imcc:No such file or directory 'interpinfo.pasm' in file 'runtime/parrot/library/parrotlib.pir' line 194
02:06 xinming joined
japhb__ I'm getting it in only one of my build trees. 02:06
As far as I can tell, at the FS level there are two differences: 1) the broken one is on a network filesystem, while the working one is local, and 2) the broken one has the `git describe` output in its path, while the working one has an ASCII-only path. 02:08
The failure is at a consistent spot, so it's not the network FS just being flaky. 02:09
japhb__ tries knocking out one of the differences.
02:13 bbkr left, bbkr joined
japhb__ It looks like ASCII path on network FS is working fine. 02:24
That's disappointing. Not sure if the failure is the net FS or parrot's build. But I suspect the latter. 02:25
02:29 jnap left, Ben_Goldberg joined 02:31 BenGoldberg left
DrEeevil sigh, MoarVM "build system" needs a cattleprod 02:33
y'all need to learn more about packaging ;)
diakopter DrEeevil: what about it is sucky? (feel free to lambast) :) 02:37
DrEeevil diakopter: it tries to run git needlessly
so the release tarball is useless
and then it is VERY confused about where to put files
also it needlessly bundles stuff that could be provided by the system, I'm a gonna try to patch that up 02:38
diakopter immagonna coin some vernacular
(teehee)
02:38 rurban1 left
diakopter DrEeevil: any suggestions/patches I'm sure would be welcomed 02:39
DrEeevil: what does it bundle that could be provided by the system
afaik all the bundled things have custom patches
DrEeevil diakopter: libtommath and libuv look like good candidates 02:40
diakopter libtommath definitely has custom patches
I know for a fact
japhb__ Yeah, I was about to say.
diakopter pretty sure libuv does too 02:41
DrEeevil: in both cases, however, the headers provided by the system versions don't expose enough
DrEeevil eew :( 02:42
that's so Google
diakopter turns to look at japhb__
02:45 rurban1 joined
DrEeevil oh well, I can tolerate the bundling ... just manually generating tarballs is annoying 02:47
and I still don't know how to get "make install" to do something reasonable
japhb__ looks around at his bus mates
diakopter DrEeevil: what prefix 02:48
japhb__ OOC, what does the "that's so Google" comment even mean in this context? 02:49
DrEeevil diakopter: eh, sucky - make install ignores it, needs to be set at configure time
japhb__: google patches and bundles everything, Chromium still has >50 bundled libs for no reason except ... google
"oh let's just patch boost and nail it in, it saves me some thinking"
diakopter well, optimizers mumble mumble mumble static linking mumble 02:50
DrEeevil it's very cowboy coding stupid thingy
diakopter: yeah ... err ... needing >6GB RAM to *link* ?!
diakopter well that's C++'s fault :D 02:51
DrEeevil no, it can be used properly
diakopter (joking)
DrEeevil but by doing hackpatchery "to make startup faster" ...
DrEeevil recommends increasing the medication dose ;)
diakopter double the max safe dose, I always say
DrEeevil here, have some opiates ;) 02:52
japhb__ LD50? Why not LD100? 02:53
As for so many bundled libs, I don't know Chromium's particular reasoning, but I do know from other projects that there is real, deep, horrible pain in trying to deal (as an upstream project) with version skew between distros when you need lots of libs. Because frankly, many libs have the compatibility levels of pre-industrialized manufacturing. 02:55
Which is to say, very little. 02:56
DrEeevil ... and nqp build system is ... uhm ... too much optimized 02:57
it should check if moar, parrot etc. are in PATH, maybe
japhb__: so hardcode it for specific versions - bundling is always wrong
japhb__: I have no trouble nailing dependencies down to specific revisions, but by bundling you just increase the pain level 02:58
japhb__ Heck, I remember that personally from working on the OpenGL bindings for Parrot. When Apple decided one day to change some fundamental types that had been agreed by everyone since OpenGL existed, I just about threw up my hands.
DrEeevil (see v8 - no api stability, and it changes every 6 weeks)
heh. I don't support legacy OSen
japhb__ DrEeevil: Nail to revisions? Really? How is that a win, if no distro has an actively supported package of your nailed revision for every lib you need. 02:59
DrEeevil japhb__: I'm packager. I can make that package-revision if needed 03:00
trust me, I play a doctor on the Internet ;)
japhb__ It's easy to say, but either A) you nail a git rev, and insist on people building from git checkouts of all libs, or you realize that not all the world is Debian and just bundle.
DrEeevil but - being that specific is usually a sign of fragile code, iow: you're not doing things sanely
s/Debian/Gentoo/ ;)
japhb__ DrEeevil: It's not the callers that are insane, it's the libs. BUGS. 03:01
DrEeevil so fix them
and by "fix" I don't mean local patching and that's it
japhb__ And if we could get even two distros to package a current version of any of our stuff, I would lend that more credence.
03:01 lizmat_ joined
DrEeevil you're part of an ecosystem, so ... behave properly 03:01
japhb__: am I the only distro-packager doing perl6 stuff?! :( 03:02
japhb__ DrEeevil: most end users do not keep the equivalent of an unstable repo in their repo list. So you can never assume the end user will actually have the fixed version available to them from the system -- so back to source build, or bundle.
DrEeevil such an arcane way of doing things 03:03
japhb__ DrEeevil: not the only. But possibly the only one trying to stay current right now, I don't know for certain.
03:04 lizmat left
japhb__ (Note: I have built packages for things in the past. I don't do it anymore, because it made me weep for the time I will never get back.) 03:04
DrEeevil (sigh. why does nqp release depend on features that are not in moarvm release ...) 03:05
DrEeevil cooks up a tarball from HEAD and pretends that's a release
japhb__ Gentoo, FreeBSD, etc. have the advantage of culturally building from source ... but not every distro is like that.
Because MoarVM support is iterating more rapidly than monthly releases. 03:06
DrEeevil it's the only way to be sane ;)
well, if nqp release relies on something, then moarvm should have an intermediate release - not like you run out of numbers
japhb__ Releases take time and energy from the very small pool of people we have to work on it. 03:07
Mind you, if you're volunteering to do rapid release cycles for MoarVM for us ... ;-)
DrEeevil a single script ...
DrEeevil is an automater
japhb__ There is more to a release than a script.
But if you're volunteering to put up a CI server and integrate with that, ... 03:08
DrEeevil well, you tag the current HEAD, make a tarball, run regression test, and if it works push tarball and VCS and that's most of it
japhb__ Seeing a common thread. ;-)
s/./?/
BTW, I don't mean to be difficult, or discouraging. 03:09
Just understand ... things are rarely so simple as they seem at first glance.
DrEeevil well, I'll see if I can optimize things
japhb__ ++DrEeevil then. :-) 03:10
DrEeevil but it's slightly irritating to see typical beginner mistakes in every perl6-related package
so I'm trying to reduce the pain
japhb__ We are all beginners at something.
DrEeevil indeed
diakopter release every minute 03:18
DrEeevil GORAMMIT 03:23
you cannot compute a version from git in a release tarball
DrEeevil applies sed
diakopter wonders how you'll get sed for windows 03:24
DrEeevil I don't deal with legacy platforms
japhb__ DrEeevil: ... but we do. 03:26
Please don't berate design decisions we make because of our desire to be everywhere.
DrEeevil japhb__: I'm just pointing out how your "release tarballs" are worse than useless 03:27
which is a bit funny because it means that no one has ever used them 03:28
(and github makes it hard to change that, so ... double funny)
japhb__ Hard to change what?
DrEeevil to provide proper release tarballs 03:29
I'm getting really tired of having to work around that ... thing
so ... nqp+moarvm: bpaste.net/show/155958/ - so close, but no dice 03:30
japhb__ looks at DrEeevil quizzically
DrEeevil japhb__: github - when it works (3 days a week on average) it's slow, glitchy, has known issues that have been reported literally years ago
it makes sourceforge look good
nqp-2013.11, moarvm HEAD 03:31
japhb__ <eight-year-old>That's hyperbole.<eight-year-old>
DrEeevil their downloads were completely not-working for a few days last week 03:32
japhb__ Which release tarballs are you referring to? MoarVM? NQP?
DrEeevil right now MoarVM, the release tarballs have two fatal failures :)
japhb__ Ah. Well, I'm not surprised. MoarVM is used pretty much only by people working on it. It's not surprising that no one noticed the release tarball stop working. 03:33
(I should say, that's only the case right now because of the very active work on it to bring support for Rakudo.) 03:34
diakopter I didn't even know there was a release tarball targe 03:35
dalek ast: 4e2b3d7 | (Solomon Foster)++ | S05-grammar/protoregex.t:
Fudge for Niecza.
DrEeevil diakopter: automatically created if you have a tag 03:36
which is funny because it gives people a false sense of security
japhb__ Now, if NQP, Rakudo, or Rakudo Star release tarballs fail (especially R*), that's a bigger deal.
Oh, you're just talking about GitHub auto-generating tarballs? Then trusting those really is insane. 03:37
I thought you were talking about ones generated using our build process and uploaded. 03:38
dalek ast: 9c3354e | (Solomon Foster)++ | S12-construction/construction.t:
Fudge for Niecza.
japhb__ is about to turn into a pumpkin
DrEeevil japhb__: that's the trap of using github :)
I still refuse to interact with it more than strictly needed 03:39
lue That's the one thing I find funny about github: they refer to tags as "releases". Not that you'd ever use tags for anything else though :) .
DrEeevil lue: and half the time their release downloads are inaccessible
so ... screw that, use a professional service like BerliOS if you need free hosting
lue So I'm guessing now is that .5 of the time that the downloads are accessible, right? :) 03:40
.oO(at least the webfont'd private character next to "x Releases" is a picture of a tag)
 03:42
03:49 xinming left, rurban1 left 03:50 xinming joined 03:54 lizmat_ left 03:57 BenGoldberg joined, Ben_Goldberg left 04:01 Ben_Goldberg joined, BenGoldberg left 04:05 BenGoldberg joined 04:07 Ben_Goldberg left 04:14 BenGoldberg left 04:18 kaleem joined 04:22 preflex_ joined, ChanServ sets mode: +v preflex_ 04:23 preflex left, preflex_ is now known as preflex 04:33 kaleem left 04:50 rurban1 joined 04:54 rurban1 left 05:19 SamuraiJack_ joined 05:22 kaleem joined
raiph DrEeevil: do you have an account on feather? 05:26
05:26 jeffreykegler joined 05:27 [Sno] left
DrEeevil raiph: no 05:27
05:28 jeffreykegler left
raiph DrEeevil: have you ever done a jenkins setup or anything like it? 05:30
DrEeevil raiph: I've set up random automation taks, but not jenkins (and the java bits suggest that I shouldn't be near it) 05:32
05:33 TimK1 left
raiph DrEeevil: i don't have the skills to do it, but i'm pretty sure a lot of us would love to see p6 get ci automation 05:41
DrEeevil indeedly
06:06 zakharyas joined
moritz \o 06:17
FROGGS[mobile] o/ 06:18
06:33 rurban left 06:34 zakharyas left 06:35 [Sno] joined, rurban joined 06:54 FROGGS[mobile] left 07:09 darutoko joined 07:34 SamuraiJack_ left
Woodi morning :) 07:39
DrEeevil: our big problem is that distros do only one or two Perl6 packages a year... And each month we do releases for not-dev audience. Packaging just this would help a lot. 07:42
DrEeevil Woodi: I am too far removed from normality to understand this idea that stale == stable 07:43
Woodi DrEeevil: I would left packages of MoarVM alone for now becouse it have 'Hallo World' feature just 1 day old :)
DrEeevil Woodi: pff. nqp claims to support it as backend!11
Woodi DrEeevil: we thing Rakudo Star releases are moust usable
but nqp is still in half way to Pelt6 07:44
Perl6 of course
DrEeevil star is not really package-able for me
Woodi and Niecha had month releases, when was maintained...
DrEeevil yup, and it got some build system fixes to make it easier to build for me 07:45
that was quite nice while it lasted
07:45 dmol joined
Woodi I have 10 star releases packaged via graft :) can switch them in two console-commands :) 07:46
but must say I tried to do Debian package of R* via one-script but something stopped me, no remember what... :) 07:47
Rakudo on Parrot is kind strange becouse it is subsystem of Parrot, it is installed like plugin not separate app... 07:48
hoelzro morning #perl6! 08:03
Woodi: what's graft? 08:04
Woodi perl script to do packaging via links: peters.gormand.com.au/Home/tools/gr...graft-html 08:05
actually it compiles something, no idea what exactly it is internally :) 08:06
hoelzro interesting...
perl6 --rxtrace is supposed to trace *every* rule during parsing, right? 08:11
because for me, it stops at comp_unit
08:20 FROGGS joined 08:23 denis_boyun joined 08:24 fhelmberger joined 08:25 silug_ left, bbkr_ joined 08:26 cognominal left, stevan__ joined 08:27 fhelmberger_ joined, [Sno]_ joined, ggoebel112 joined 08:29 takesako_ joined 08:30 rom1504_ joined, FROGGS_ joined, PacoAir_ joined, ssutch_ joined, hummeleBop1 joined, djanatyn1 joined, cognominal joined, awwaiid_ joined 08:31 emma_ joined, Ulti_ joined, fhelmberger_ left, Vlavv_ joined 08:32 fhelmberger_ joined, SimonFelix joined 08:33 Shozan joined 08:34 cooper_ joined, cybernicus joined, ggoebel111 left, hummeleB1 left, ssutch left, FROGGS left, [Sno] left, stevan_ left, awwaiid left, djanatyn left, fhelmberger left, bbkr left, cooper left, raiph left 08:35 PacoAir left, Vlavv` left, Ulti left, SevenWolf left, roboticus left, [Sno]_ is now known as [Sno], cooper_ is now known as cooper, PacoAir_ is now known as PacoAir, cooper left, cooper joined 08:36 SimonFelix is now known as SevenWolf, cognominal left
Woodi I wonder is it possible to unbundle parrot and nqp from R* package. Maybe instead README can have "required" section with parrot/nqp version number and links. It will be good, at least, for me :) --build-gen options are for devs and devs switches to githun anyway... 08:39
DrEeevil Woodi: that would make life easier for us, but harder for everyone else 08:40
nwc10 I think that that defeats the entire point of the R* distribution - that it tries to be an "everything you need to get started" download 08:41
08:42 FROGGS_ is now known as FROGGS
hoelzro I can imagine breaking up rakudo-star into a package that depends on parrot and nqp 08:43
moritz well, for an OS package, R* would likely be a meta package 08:44
that depends on Rakudo and the modules
and Rakudo in turn depends on nqp
and nqp on parrot, the JVM and later Moar 08:45
JimmyZ # perl Configure.pl --backends=moar --prefix=../MoarVM/install/ 08:46
Missing braces on \N{} in regex; marked by <-- HERE in m/(?<v>[\d\._]+)\N <-- HERE +\n(?<n>\S+)/ at Configu
This is perl, v5.10.1 (*) built for x86_64-linux-thread-multi
hoelzro moritz: that's how I would do it
JimmyZ rakudo branch
moritz JimmyZ: is that in moar-support? 08:47
JimmyZ yeah
FROGGS JimmyZ: that was me
JimmyZ ;)
moritz FROGGS: simply use . instead of \N 08:48
FROGGS k
08:49 denis_boyun left
dalek kudo/moar-support: 16da194 | (Tobias Leich)++ | Configure.pl:
s/\N/./, JimmyZ++ and moritz++
 08:50
Woodi I always thinked R* do not need bundled Parrot (until it patch it...) a) ppls wantin to compile Perl6-thing can just download Parrot tgz with easy; b) ppls do not wanting to complie have .msi 08:51
08:51 FROGGS[mobile] joined
Woodi looks lazy ppls use .msi :> 08:51
JimmyZ hmm, make -j8 is broken in moar-support branch
Woodi in R* extracting knowledge what Parrot/nqp version is needed was paintfull, so bundling was usefull :) 08:53
hoelzro does anyone feel like "idiom-checking" some code I wrote last night?
I implemented a min heap, but it feels like I'm writing another language in Perl 6
if anyone's interested: gist.github.com/hoelzro/7820243 08:54
moritz (($current-index + 1) / 2).floor I'd write that as ($current-index + 1) div 2 08:55
also, I find $current-index to be too verbose :-)
I'd just name it $i
hoelzro ahhh
moritz also the outermost 'if' in method pop looks superfluous to me 08:57
I'd write that as my $value = @!values.pop; return unless @!values; # code from the else-branch here
hoelzro ok, seems reasonable 08:59
09:05 cognominal joined 09:06 cognominal left 09:22 lizmat joined
lizmat has arrived at the Carrefour Numérique a la Cité des Sciences 09:25
aka patch -p1 is starting 09:26
hoelzro nice
I wish I'd stop finding out about these things a few days before they happen =) 09:27
lizmat yes, the marketing of the patch -p1 left something to be desired 09:33
but I have no doubt there will be a patch -p2 sometime next year :-) 09:34
lizmat wonders how Chinese/Japanese developers use git tag 09:36
09:42 xinming left
lizmat will start with writing "Bagging the changes in the Set specification" advent post, so it can be used for tomorrow if no alternative presents itself 09:42
moritz lizmat++
lizmat: re git tag, "with Unicode"?
lizmat yes, as Malmö apparently creates problems as a tag in some situations :-( 09:43
moritz didn't have any 09:44
lizmat perhaps on ascii only filesystems ? 09:45
JimmyZ lizmat: for Chinese developers, they use english as git tag, or pinyin 09:47
and number
09:49 xinming joined
mathw oh patch -p1 is an event? 09:49
09:50 rindolf joined
mathw lizmat: maybe in non-UTF8 locales? 09:50
if such things still exist
lizmat patch.pm/p1
09:52 FROGGS[mobile] left
FROGGS hoelzro: I'd swap line 34 with line 35, and do: my int $left-index = $right-index - 1 09:55
09:55 xinming left
hoelzro why's that? 09:55
FROGGS hoelzro: so you dont do ($current-index + 1) * 2 twice 09:56
09:56 xinming joined
hoelzro ah, fair enough 09:56
FROGGS and you see at once that the left index is rigth index minus one
09:57 woolfy joined 10:09 SamuraiJack_ joined 10:11 tgt joined 10:14 xenoterracide joined 10:17 woolfy left 10:18 rurban1 joined 10:21 cognominal joined 10:22 rurban1 left 10:27 SamuraiJack_ left 10:28 SamuraiJack_ joined
hoelzro if I'm debugging in NQP, and I want to dump something's time, will nqp::say(nqp::what($obj)) do the job? 10:28
or do I have to do something like nqp::say(nqp::what($obj).name)?
lizmat I'm not sure, FROGGS might know 10:29
hoelzro also, if I want to say() something in NQP, and it tells me "cannot stringify this", can I add a method to the thing being said do enable stringification? 10:32
FROGGS nqp: my $a = "abc"; say($a.HOW.name($a))
camelia nqp-moarvm: OUTPUT«Assignment ("=") not supported in NQP, use ":=" instead at line 2, near " \"abc\"; sa"␤panic␤»
..nqp-parrot: OUTPUT«Assignment ("=") not supported in NQP, use ":=" instead at line 2, near " \"abc\"; sa"␤current instr.: 'panic' pc 16262 (gen/parrot/stage2/NQPHLL.pir:6020) (gen/parrot/stage2/NQPHLL.nqp:426)␤»
..nqp-jvm: OUTPUT«Assignment ("=") not supported in NQP, use ":=" instead at line 2, near " \"abc\"; sa"␤ in panic (gen/jvm/stage2/NQPHLL.nqp:379)␤ in infix:sym<=> (gen/jvm/stage2/NQP.nqp)␤ in !protoregex (gen/jvm/stage2/QRegex.nqp:746)␤ in infix (gen/jvm/stage2/NQPHLL.nqp…»
FROGGS nqp: my $a := "abc"; say($a.HOW.name($a))
camelia nqp-moarvm, nqp-jvm: OUTPUT«BOOTStr␤»
..nqp-parrot: OUTPUT«Can only use get_how on a SixModelObject␤current instr.: '' pc 53 ((file unknown):44) (/tmp/ObK3epAQW1:1)␤»
FROGGS parrot does not like natives
hoelzro I'm getting pretty tired of Parrot. 10:33
FROGGS nqp: my $a := "abc"; say(nqp::objprimspec($a)) # 0=obj, 1=int, 2=num 3=str or so
camelia nqp-moarvm, nqp-jvm, nqp-parrot: OUTPUT«3␤»
hoelzro and I've only been looking into Rakudo for a few months now =/
10:34 dayangkun left
FROGGS hoelzro: but that is our design, not parrot's fault 10:34
hoelzro doesn't matter! blaming Parrot!
FROGGS :P
hoelzro denouncing! 10:35
10:40 woolfy joined
FROGGS r: say $*VM<config><prefix> 10:47
camelia rakudo-parrot 262e60: OUTPUT«/home/p6eval/rakudo-inst-2␤»
..rakudo-jvm 262e60: OUTPUT«(Any)␤»
FROGGS Any? :(
jnthn FROGGS: What's under <config> is VM-specific at the moment, iirc 10:48
See terms.pm for how it gets the prefix out differently.
FROGGS r: say $*VM<properties><perl6.prefix> 10:49
camelia rakudo-jvm 262e60: OUTPUT«/home/p6eval/rakudo-inst-1␤»
..rakudo-parrot 262e60: OUTPUT«(Any)␤»
11:11 SevenWolf left 11:17 szabgab joined
szabgab hi, I just compiled * 2013.11 on OSX 11:18
11:18 rurban1 joined
szabgab and tried to panda list but it takes ages 11:18
actually after 5 or 10 min I stopped it without seeing any output 11:19
11:19 mberends left 11:20 denis_boyun joined
FROGGS seems like it is unable to fetch feather.perl6.nl:3000/projects.json 11:20
hmm, and so am I
szabgab my little resarch also showed that it gets stuck in the method update 11:21
that was supposed to fetch that file
FROGGS: but how did you see that so fast?
FROGGS szabgab: I know what it wants to do and that is the only thing that is likely to hang 11:23
11:23 rurban1 left
FROGGS and I've got that projects.json in my browser's history, so I tried 11:23
I can connect to feather.perl6.nl:3000, but I don't seem to get the json file :/ 11:24
jnthn Something not restarted after recent reboot, maybe?
FROGGS jnthn: no, I was able to fetch that file yesterday evening
jnthn oh... 11:25
11:29 atroxaper joined 11:32 autumn left 11:35 autumn joined 11:45 szabgab left 12:03 dayangkun joined
hoelzro grr 12:07
how much memory does it take to build jakudo?
jnthn Couple of gigs last I checked. 12:08
hoelzro curses! 12:09
12:09 denis_boyun left
nwc10 hoelzro: what was the answer you were hoping for? 12:10
hoelzro hmm
I guess I didn't know what else to expect
considering my VM with 2.5GB RAM kills the build process everytime 12:11
lizmat S32/Containers:1162 12:13
S32:1162 12:14
synopsebot Link: perlcabal.org/syn/S32.html#line_1162
lizmat wishes synopsebot would point S32/Containers:1162 correctly to perlcabal.org/syn/S32/Containers.html#line_1162 12:15
12:19 rurban1 joined
moritz S32::Containers:1162 12:21
12:22 [Sno] left
lizmat I think the regex should just also allow / 12:22
before the :
and all would be well, no?
12:23 rurban1 left 12:32 p5eval joined
lizmat S03:3311 12:32
synopsebot Link: perlcabal.org/syn/S03.html#line_3311
lizmat
.oO( just abusing the synopsebot to get links )
 12:33
jnthn Isn't that what it's for? :P
colomon used to abuse the ir clog for links... 12:40
moritz too 12:48
dalek ecs: 9fdb85e | (Elizabeth Mattijsen)++ | S32-setting-library/Containers.pod:
I don't think we ever had a MixBag
 12:56
lizmat provisionally finished Day 12 of advent: gist.github.com/lizmat/7823431/raw...tfile1.txt 13:04
comments welcome: moritz, masak, jnthn ? 13:05
lue? :-)
13:08 cybernicus left 13:12 denis_boyun joined 13:15 brrt joined 13:20 xenoterracide left, denis_boyun left 13:21 tgt left, rurban1 joined, brrt left 13:23 kaare_ joined 13:28 tgt joined 13:36 kaleem left
lizmat ok, scheduled the post for Dec 12, 0:00. 13:38
if we need it for tomorrow, it's just a matter of changing the title (s/12/07) and the publish date 13:39
FROGGS lizmat++
13:42 [Sno] joined 13:53 sqirrel joined 13:56 ajr joined 13:57 ajr is now known as Guest8998, Guest8998 is now known as ajr_
ajr_ Is the equivalent of $. implemented in P6 yet? 13:58
13:59 V_S_C joined
moritz p: my $dummy = get(); say $*ARGFILES.ins 13:59
camelia rakudo-parrot 262e60: OUTPUT«1␤»
14:02 V_S_C left
ajr_ Thanks, moritz 14:24
14:25 rindolf left 14:26 SamuraiJack_ left 14:44 p5eval left 14:49 thou joined, pecastro left 14:55 denisboyun joined
woolfy I just read and re-read lizmat++ 's advent calender day 12 contribution ( gist.github.com/lizmat/7823431/raw...tfile1.txt ) and I liked it (not understood it completely, but hell) 14:56
PerlJam woolfy: maybe that's a sign that lizmat++ needs to expand on a couple of things to make them more understandable? 15:03
lizmat fwiw, it already felt as a long advent blog post
PerlJam is reading now 15:04
lizmat apparently, there is a preview available at perl6advent.wordpress.com/?p=1938&a...eview=true
PerlJam ah, even better 15:05
woolfy PerlJam: no, it's a sign that I should study Perl 6 better and harder and more in-depth and that at home I should ask more 'stupid' questions to lizmat++ 15:07
PerlJam you know ... I'm not sure who our audience is with the advent calendar. For instance, "implemented as dispatchers" and "expense of another call frame" may not make enough sense to our readers without links to somewhere that explain 15:08
lizmat ah, good point 15:09
hmmm....
PerlJam however, lizmat++ for an excellent advent post!
woolfy Our audience? We should all mention the advent calendar on our facebook, in our tweets, in our own blogs, on this channel and on other channels and mailing lists. That might easily double to audience from 10 to 20 readers!
(or what that number demotivational) 15:10
lizmat woolfy: it's more than that already (moritz has the numbers)
PerlJam woolfy: do you have a wordpress login for the advent calendar?
moritz we've had between 600 and 1k page views per day since Dec 01 15:11
woolfy ow cool
moritz today "only" 320, but today isn't over yet
colomon best ever 11,000 views.
woolfy I do have a wordpress login.
moritz woolfy: if you tell me the login name, I can invite you to that blog
colomon: that was when slashdot had us on the front page 15:12
PerlJam The residual 50ish views each day in Nov are interesting too :)
woolfy moritz: my login name is wendyga
( wendyga.wordpress.com/ ) 15:13
moritz woolfy: invitation sent
woolfy Which reminds me that I have to write a next blog about Fosdem... we have a devroom and we need speakers...
moritz: thanks
PerlJam Has anyone "advertised" the advent calendar around the internet this year? 15:18
I see there's a perlmonk post that mentions is. 15:19
er it
15:19 rurban1 left
woolfy Moritz: "invited you to become an editor"??? ow... I just accepted the invitation... now I have to write too... I don't wanna write about something that I know not enough about. 15:20
15:21 fhelmberger joined
PerlJam woolfy: you don't *have* to write something ... but if you have something interesting to say ... :-) 15:21
lizmat PerlJam: made the reference to dispatcher and call frame a bit less incrowdy
PerlJam++ for the suggestion 15:22
woolfy PerlJam: maybe at Fosdem at the Perl-booth... not on a blog about Perl 6.
moritz woolfy: as an editor you have the possibilities to preview other people's posts
woolfy: which is why I chose that permission level -- feel free not to use the extra power :-)
15:24 fhelmberger_ left
woolfy I just want to read and maybe I will comment by email. 15:25
moritz that should work with the current permissions :-)
15:25 fhelmberger left
colomon is afraid power will go to woolfy++'s head… ;) 15:28
PerlJam is scheming how to get an advent post out of woolfy for next season ;)
jnthn "An illustrated guide to the tuits of 2014" :D 15:30
lizmat I think an illustrated guide to the tuits and buttons of 2013 would already be possible :-) 15:33
jnthn aye :)
lizmat including some pictures of the biggest tuit of them all
jnthn will take some pretty Camelia tuits with him to his Build Stuff talk next week :)
woolfy You nutty people stop scheming to get me writing advent calender stuff. I will not do it. Hmm, maybe. No no no! Aw come on... NOOOOOO!!!! 15:34
lizmat please? 15:35
jnthn haha :)
woolfy oO(how bad would cancelling my whole Wordpress-account be?)
jnthn
.oO( Could cause locuts plagues and a nuclear winter... )
 15:36
moritz PerlJam: start scheming your own advent post!
lizmat
.oO( how many levels of reverse psychology would be needed to get woolfy to write an advent post? )
PerlJam moritz: I've been scheming, but nothing has gelled yet. :( 15:37
woolfy oO(I am sleepy and want to go into hibernation for the rest of winter)
jnthn r: sub woolfy() { say 'psychology'.comb.reverse.join; woolfy }; woolfy 15:38
15:38 raiph joined
colomon lolcats plagues? 15:38
camelia rakudo-parrot 262e60: OUTPUT«(timeout)ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp…»
..rakudo-jvm 262e60: OUTPUT«ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp␤ygolohcysp…»
PerlJam lizmat, woolfy: Have you guys seen www.youtube.com/watch?v=Xe1TZaElTAs ? That the tuits are "woodne nickels" resonates with the shinto shrine that Shirky talks about :) 15:39
15:41 sqirrel left
woolfy Read about that movie and about the shrine and the Unesco response. Watching the movie right now. 15:41
jnthn: it's not working! it's not working! it's not working! it's not working! it's not working! it's not working! it's not working! it's not working! it's not working! it's not working!
15:41 denisboyun left
lizmat :-) 15:42
r: %(:a :b).perl.say
camelia rakudo-parrot 262e60, rakudo-jvm 262e60: OUTPUT«("a" => Bool::True).hash␤»
lizmat n: %(:a :b).perl.say
camelia niecza v24-108-g17d73e4: OUTPUT«(timeout)[auto-compiling setting]␤»
lizmat std: %(:a :b).perl.say
camelia std 3b262af: OUTPUT«ok 00:01 124m␤»
nwc10 jnthn: that's not too bad. It's not like you said "hoardes of Zombies"
lizmat n: %(:a :b).perl.say 15:43
camelia niecza v24-108-g17d73e4: OUTPUT«(timeout)[auto-compiling setting]␤»
15:43 jnap joined
jnthn lizmat: I *think* that's adverbing the adverb. It's oddly confusedly dropping it somehow, I guess... 15:44
lizmat afaics, it's *not* getting passed as a named parameter to e.g. Hash.new 15:45
jnthn No, I doubt it will be
I suspect it's just getting lost
lizmat seems it's getting lost in the AST building
moritz lost in adverbian
jnthn Probably 'cus nothing notices it's there, or thinks to look.
I don't know exactly where that'd want fixing, off hand...
lizmat well, I would think we want that to be passed as a named parameter, and then have e.g. Hash.new complain about additional named parameters ? 15:46
15:46 ajr_ left
lizmat or maybe we should force multiple named parameters to have no whitespace between them 15:47
and make it a TTIAR error ?
the fact that forgetting a comma makes code silently forget, feels pretty bad to me 15:48
15:48 ajr joined, ajr is now known as Guest83859, Guest83859 is now known as ajr_
jnthn lizmat: I'm not sure exactly what should happen here... 15:48
I know that in:
foo(:a :b) 15:49
It's meant to be have like
foo(:a, :b)
It's certainly not meant to just be silently forgotten.
15:50 thou left, rurban1 joined
lizmat n: sub foo { say %_.perl }; foo(:a :b) 15:53
camelia niecza v24-108-g17d73e4: OUTPUT«(timeout)[auto-compiling setting]␤»
lizmat r: sub foo { say %_.perl }; foo(:a :b)
camelia rakudo-parrot 262e60, rakudo-jvm 262e60: OUTPUT«("a" => Bool::True).hash␤»
lizmat
.oO( is niecza on the blink? )
15:54 denis_boyun joined
lizmat jnthn: what does "QAST::Op(call &foo) foo(:a :b)" mean 15:55
is the "foo(:a :b)" just comments ?
FROGGS yes 15:56
it is just what in $/ was when calling the action method
lizmat ok 15:57
diffing -uw the --target=ast
+ - QAST::Op+{QAST::SpecialArg}(p6bool)
+ - QAST::IVal(1)
+ - QAST::Op+{QAST::SpecialArg}(p6bool)
+ - QAST::IVal(1)
is what seems to be added for the case with comma
I guess the "QAST::Op+{QAST::SpecialArg}(p6bool)" is a stringification that loses the name of the key, right? 15:58
FROGGS it does not show the key, never 15:59
15:59 Psyche^_ joined
lizmat ah, ok 15:59
FROGGS p: say (:a :b)
woolfy So, question... I am reading the Perl 6 Advent Calendar: wordpress.com/read/blog/id/10740073/ and I see '30 followers'. If I would use this URL on my Facebook page, would my friends be able to read the contributions?
camelia rakudo-parrot 262e60: OUTPUT«"a" => Bool::True␤»
moritz patches to make the QAST dumper more informative or clear are always welcome :-)
16:00 kaleem joined
lizmat hmmm... so the name of the keys can never be seen in the AST dump ? 16:00
PerlJam woolfy: aye, they should.
moritz decommuters
FROGGS PerlJam: not as far as I know 16:01
jnthn lizmat: Knowing how those work, I think it's likely they don't get shown, yeah.
PerlJam hmm.
FROGGS btw, the ":a :b" parses as a single term:sym<colonpair> according to viv 16:02
16:02 Psyche^ left
FROGGS like in: 16:02
16:02 rindolf joined
FROGGS p: use Test:a :b 16:02
camelia rakudo-parrot 262e60: OUTPUT«===SORRY!===␤Error while importing from 'Test': no such tag 'b'␤»
FROGGS ohh
interesting 16:03
p: use Test:a:b
camelia ( no output )
jnthn I suspect it does in Rakudo too; it's the EXPR that parses the other one...
Well, it's that infix can parse a colonpair...
16:05 denis_boyun left, kaleem left 16:10 FROGGS left
woolfy PerlJam: ok, thanks, I will mention it 16:11
16:11 denis_boyun joined 16:13 thou joined
lizmat n: sub foo { say %_.perl }; foo(:a :b) 16:25
camelia niecza v24-108-g17d73e4: OUTPUT«(timeout)[auto-compiling setting]␤»
lizmat I guess niecza *is* on the blink
[Coke] lizmat: first time you use niecza, it auto compiles the setting. try again. 16:28
n: sub foo { say %_.perl }; foo(:a :b)
camelia niecza v24-108-g17d73e4: OUTPUT«(timeout)[auto-compiling setting]␤»
colomon :( 16:29
n: say "Hi"
[Coke] whoever has the camelia code to build niecza, if you run a simple hello world after the build, it'll do that step for us.
camelia niecza v24-108-g17d73e4: OUTPUT«(timeout)[auto-compiling setting]␤»
colomon seems like something is wrong here?
[Coke] it can take a little while, but 5m seems excessive. wonder if we're screwing it up byasking repeatedly. ;) 16:30
[Coke] heads back to dayjob.
16:30 sqirrel joined
dalek kudo-star-daily: 858b52a | coke++ | log/ (5 files):
today (automated commit)
 16:30
[Coke] github.com/coke/rakudo-star-daily/...odules.log - don't see any deprecation warnings or failures. whee. 16:31
lizmat [Coke]++
so, just over 6 hours until the next advent blog and no takers yet
16:32 xinming left
lizmat around an hour from now, patch -p1 will shut down for the day and I will be most likely offline for the evening 16:32
[Coke] I am dayjobbed out. I might be able to take one later in the month, after a weekend.
lizmat maybe I should just schedule my day 12 presentation for tonight ?
16:33 xinming joined, kst` is now known as kst 16:34 ssutch_ left
PerlJam lizmat: +1 (especially since you've already got it written :) 16:35
[Coke] lizmat: it would save a kitten.
jnthn +1
.oO( Every time you write a Perl 6 advent post last minute, God kills a kitten? )
 16:36
lizmat ok, scheduled for 7 Dec 2013, UTC 0:00
[Coke] lizmat++
dalek : b0e7c18 | (Elizabeth Mattijsen)++ | misc/perl6advent-2013/schedule:
Move Day 12 to Day 7
 16:37
lizmat which means our first open day is Dec 12 now
[Coke] I'll take the 12th. 16:38
put me down for something. :|
dalek : 778f99c | (Elizabeth Mattijsen)++ | misc/perl6advent-2013/schedule:
Putting coke up to something
 16:39
lizmat :-)
PerlJam lizmat++ (you were reading my mind apparently) 16:40
lizmat great minds think alike :-)
[Coke] subroutine signatures? Doesn't look like we've had a post specifically on that.
ah. ok, leave that one, put me down for another one on the 21st. 16:41
lizmat about subroutine signatures ?
16:42 tipdbmp left 16:43 djanatyn1 is now known as djanstyn, bluescreen10 joined, djanstyn is now known as djanatyn
dalek : 74fe36c | (Elizabeth Mattijsen)++ | misc/perl6advent-2013/schedule:
Putting coke up for some more on the 21st
 16:44
16:44 tipdbmp joined 16:46 pmurias joined
lizmat gives up on foo(:a :b) for now 16:49
16:50 yeltzooo left
woolfy OK, when I log out of WordPress and try to view wordpress.com/read/blog/id/10740073/ it wants me to sign in before allowing me to read it. 16:53
16:53 cooper left
woolfy So that URL is not something I can tell my friends to go read. Most of them have no account on WordPress. 16:53
Anybody a better idea? 16:54
16:54 robinsmidsrod left
lizmat sorry, /me is a complete Wordpress noob 16:54
16:54 cooper joined, cooper left, cooper joined 16:55 TimK1 joined, yeltzooo joined, obra joined
[Coke] as opposed to perl6advent.wordpress.com/ ? 16:56
16:56 jeffreykegler joined, robinsmidsrod joined
[Coke] ok, dayjob ferrealz. 16:56
lizmat woolfy: perhaps perl6advent.wordpress.com/2013/11/ ? that gives you a table of contents 16:57
16:58 denis_boyun left
lizmat sees perlpilot editing the toc :-) 16:59
woolfy lizmat++ that's much better thank you 17:01
[Coke]++ thank you, would have worked too 17:02
17:03 kivutar joined 17:06 atroxaper left 17:10 robinsmidsrod left, jeffreykegler left, robinsmidsrod joined, btyler joined, ajr_ left
woolfy So, advertised the advent calendar on my FB page. 17:11
17:11 jeffreykegler joined 17:12 ajr joined, ajr is now known as Guest25690, Guest25690 is now known as ajr_
lizmat cool! 17:14
dalek kudo/nom: dcf13b5 | (Elizabeth Mattijsen)++ | src/vm/jvm/core/Promise.pm:
Fix param snafu in Promise.then
 17:19
17:19 denis_boyun joined 17:24 ssutch joined, sqirrel left
dalek ast: f7caca4 | (Elizabeth Mattijsen)++ | S17-concurrency/promise.t:
Add Promise.allof test that previously failed
 17:25
lizmat patch -p1 dismantling for today, see you guys later& 17:26
17:26 lizmat left, cognominal left 17:28 jeffreykegler left 17:30 FROGGS[mobile] joined
moritz idea for an advent post: rewrite the code on leonerds-code.blogspot.co.uk/2013/1...day-1.html and the following posts in p6 :-) 17:31
woolfy says I love you all with a huge great love and until tomorrow 17:32
17:32 woolfy left 17:36 kivutar left 17:37 robinsmidsrod left 17:39 robinsmidsrod joined 17:40 zakharyas joined 17:42 robinsmidsrod left 17:44 ajr_ left, robinsmidsrod joined 17:49 FROGGS joined, SL-RHanson left 17:56 jnap left, robinsmidsrod left 17:57 robinsmidsrod joined 17:59 sftp left, sftp joined 18:04 denis_boyun left, robinsmidsrod left, robins joined 18:05 rhr left 18:13 rhr joined 18:17 rindolf left 18:18 robins left, robinsmidsrod joined 18:21 vky joined 18:30 darutoko left 18:35 aindilis` left, xinming left
timotimo i spent pretty much all day playing starbound with two friends 18:36
now for some relaxing java debugging
18:36 xinming joined 18:39 jnap joined, aindilis` joined 18:43 jnap left 18:51 sqirrel joined 18:55 FROGGS[mobile] left, denis_boyun joined 18:56 kivutar joined 18:59 denis_boyun left 19:02 SamuraiJack_ joined 19:03 btyler left 19:05 telex left 19:07 telex joined 19:08 sqirrel left
lue lizmat++ excellent post 19:09
.oO(So you're saying if we toss the Norwegian Blue we can use literal Unicode in the setting like civilized people? I'm OK with that.)
TimToady let's wait until we have moar options 19:10
PerlJam This channel always makes me smile :-) 19:17
lue TimToady: But of course :)
.oO(We just need to give MoarVM an animal mascot, and name it Eric.)
 19:18
19:18 vky left 19:19 japhb__ left
jnthn Hm, a few of us pondered a Moar mascot at YAPC::EU and I think a Chimera was what we came up with... :) 19:19
.oO( anagram of "eric ham"... )
 19:20
TimToady with eric internalizing occuaptional therapy, you have erOTic, which is OT 19:22
PerlJam As soon as someone puts a chimera on the flip side of a tuit, it'll be official :)
19:23 btyler joined
lue jnthn: sounds great, you just need to get a license for Eric now... 19:23
TimToady
.oO(Better Eric the Red than Eric the Dead...)
TimToady liefs now before someone swings a battleax at him 19:24
FROGGS .oO( battleaux - Is that french? ) 19:25
TimToady wouldn't you vike to know? 19:27
19:28 araujo left
PerlJam TimToady: you must've been exercising to need all of that stretching 19:29
19:29 kivutar left
PerlJam :) 19:29
TimToady it's already a long boat 19:30
dalek kudo/moar-support: 84e430f | jnthn++ | src/Perl6/Actions.nqp:
Avoid another p6bindassert with Mu.
 19:32
19:32 gjb__ joined, gjb__ is now known as japhb 19:33 araujo joined 19:35 SamuraiJack_ left 19:37 avuserow joined
raydiak I can't seem to recall where in the synopses I read about this: How do I call a method in a parent class from my method which overrides it? 19:56
19:56 rurban1 left
jnthn self.LeParentClass::method(...) to be very explicit 19:56
But normall one of callsame/callwith/nextsame/nextwith is enough. 19:57
*normally
19:58 jnap joined
raydiak thanks jnthn++ ...I should locally clone the specs so I can grep them 19:59
japhb Is panda supposed to work properly for installing to both the parrot and jvm backends when you have a single rakudo tree with both backends installed? 20:01
20:08 kivutar joined
jnthn Don't think so... 20:09
diakopter .
dalek kudo/opts: eb9cd18 | jnthn++ | src/Perl6/Optimizer.nqp:
Make paramless block opt on JVM a level 3.

Breaks some spectests. Need better analysis before it can become a level 2. Doesn't break anything in the setting itself, though, so we can keep the benefit from better optimizing that.		 20:10
japhb jnthn: Sigh.
OK, at least I know to stop banging my head against it. 20:11
jnthn japhb: We only had a build system that could build/install both at once for a couple of months. We only had a second backend that was useful for stuff 4-5 months... It's not so surprising Panda doesn't know how to handle this case yet. 20:13
20:15 vky joined 20:16 vky left
jnthn nqp: class A { has $!a; method a() { $!a } }; say(nqp::isnull(np::create(A).a)) 20:20
camelia nqp-jvm: OUTPUT«Can not invoke this object␤ in (/tmp/1vjTjeiUJ7:1)␤ in (gen/jvm/stage2/NQPHLL.nqp:1100)␤ in eval (gen/jvm/stage2/NQPHLL.nqp:1086)␤ in evalfiles (gen/jvm/stage2/NQPHLL.nqp:1292)␤ in command_eval (gen/jvm/stage2/NQPHLL.nqp:1196)␤ in command_line (ge…»
..nqp-parrot: OUTPUT«invoke() not implemented in class 'NQPMu'␤current instr.: '' pc 138 ((file unknown):76) (/tmp/_gs92h975A:1)␤»
..nqp-moarvm: OUTPUT«Cannot invoke this object␤frame_name_0␤»
jnthn nqp: class A { has $!a; method a() { $!a } }; say(nqp::isnull(nqp::create(A).a)) 20:21
camelia nqp-moarvm: OUTPUT«1␤»
..nqp-jvm, nqp-parrot: OUTPUT«0␤»
TimToady Doesn't putting Python references into our advent postings violate our anti-anti-lol policy somehow? 20:22
20:26 denisboyun joined 20:27 rurban1 joined 20:28 rurban2 joined 20:31 rurban1 left 20:33 rurban2 left
jnthn um...seems nqp-p.bat on windows ends up containg a shell script. wtf. 20:34
lue jnthn: maybe it's a BATch of shell scripts? :) 20:35
jnthn I can't actually figure out where gen\parrot\nqp_launcher comes from... 20:38
fill_template_file('src/vm/parrot/nqp.sh', 'gen/parrot/nqp_launcher', %config); 20:39
...oh.
lue ? wouldn't/shouldn't that be parrot's built-in nqp there? 20:41
jnthn huh? 20:42
No, it should have a conditional and use the nqp.bat on Windows.
20:47 raiph left 20:51 raiph joined
sjohnson anti-lol = no laughing 20:51
20:51 denisboyun left
dalek p/indy-args: ee45ffe | jnthn++ | / (2 files):
Fix NQP runner on Windows, so make test works.
 20:53
p/indy-args: ed3c300 | jnthn++ | src/QAST/Node.nqp:
Fix build on Parrot after indy-args work.
colomon rn: say 5 +< 8 20:57
camelia rakudo-parrot dcf13b, rakudo-jvm dcf13b: OUTPUT«1280␤»
..niecza v24-108-g17d73e4: OUTPUT«(timeout)[auto-compiling setting]␤»
20:57 avuserow left
lue r: my $a = 0x80; my int8 $b = 0x80; say ($a +< 1).fmt("%02X"); say ($a +< 1).fmt("%02X"); 21:00
camelia rakudo-parrot dcf13b, rakudo-jvm dcf13b: OUTPUT«100␤100␤»
21:05 SevenWolf joined 21:16 Entonian joined
dalek Heuristic branch merge: pushed 28 commits to nqp by jnthn 21:22
lue
.oO(What is a heuristic branch merge anyway?)
 21:27
21:29 rurban1 joined 21:31 xenoterracide joined
timotimo turns out: no jvm work until now :P 21:32
21:34 rurban1 left
colomon is trying to puzzle out enough of sorear++'s regex handler to fix sigspace 21:35
dalek kudo/nom: afb0511 | jnthn++ | src/Perl6/Optimizer.nqp:
Start trying to use callstatic where possible.
 21:36
kudo/nom: ceb7f13 | jnthn++ | src/Perl6/Optimizer.nqp:
Avoid double-walking trees in sink context.

Bump NQP_REVISION.
21:36 dalek left 21:37 dalek joined, ChanServ sets mode: +v dalek, Entonian left
lee_ i'm getting an error trying to build the latest nqp-jvm (after a make clean) 21:46
gist.github.com/leedo/0243ac58685541a7bc1e
hope that is of some help
i should try a clean nqp checkout i guess
japhb lue: It means dalek saw a big pile of commits, heuristically assumed they were from a branch merge, and rather than spam the channel, it just said it looked like a branch merge happened.
21:47 ajr joined
jnthn lee_: Ah...that looks like an error reporting fail due to the bootstrap using an old op, rather than a real failure... 21:47
21:47 ajr is now known as Guest89226
lee_ trying a clean checkout 21:47
21:47 Guest89226 is now known as ajr_
dalek Heuristic branch merge: pushed 61 commits to rakudo/moar-support by jnthn 21:48
jnthn uh, ratehr than *the* real failure
lue japhb: I figured something like that, though perhaps "28 commits pushed to nqp by jnthn. That's quite a bit" or something would avoid having to guess things :)
jnthn Anybody pushing that much at once is probably merging a branch. OR they just had a really long hacking vacation in a place without internets :) 21:49
lue come to think of it, "28 commits pushed ... . Last of the commits is:" followed by the last commit in traditional dalek style might be a nice improvement.
japhb It's getting harder and harder to have a hack vacation without internets.
jnthn japhb: True that... :) 21:50
lue (because the Heuristic message provides no links or anything to view what happened. Not that it's unbearable, but would be nice otherwise.)
jnthn Anyway, the thing I just pushed was just sync'ing moar-support with the latest work from nom. 21:52
21:52 kaare_ left
timotimo and nqp's master branch now has all the goodies? 21:52
jnthn Yeah
timotimo yays
jnthn indy-args is merged and gone.
lue
.oO( <dalek> 37 commits to branch foo by bar. EXPLAIN! )
jnthn++ # merging work into master is always worth something :) 21:53
jnthn And the Rakudo opts branch is merged into nom and gone.
timotimo did moar-support get goodies, too? 21:55
ah, yes
cool
jnthn I merged opts into nom before I merged nom into moar-support, yes.
git++ # making this stuff easy 21:56
I guess main question for the future is when we want to merge moar-support :) 21:58
Probably when sanity test passes...
timotimo sounds sane 21:59
jnthn :P
21:59 dmol left, dmol joined
timotimo huh, weird: No rule to make target `tools/build/create-moar-runner.pl', needed by `m-install'. Stop. 22:00
22:01 jeffreykegler joined
jnthn You're trying to m-install it? :P 22:02
timotimo i'm trying to -j3 install
Heap corruption detected: pointer 0x2ae949585cf8 to past fromspace so that wasn't fixed yet? :( 22:03
jnthn No; didn't reproduce it locally here yet either.
timotimo OK. i'll try to go at it again without --optimize 22:04
jnthn Something happens when we build with --optimize.
22:06 lizmat joined 22:07 woolfy joined 22:09 bluescreen100 joined 22:10 bluescreen100 left, bluescreen100 joined
lizmat good *, #perl6! 22:11
I just git pulled and tried to make install, and get:
Error while compiling op call: Method 'jast' not found for invocant of class 'NQPMu'
22:12 bluescreen10 left
lizmat trying from a reconfigure again now 22:12
jnthn You'll need an NQP update too, if you didn't already... 22:13
I did bump NQP_REVISION, though...
lizmat well, a simple make install didn't seem to pick that up
timotimo yeah, that check only hppens when you run configure 22:14
lizmat should it have?
ah, ok
22:14 zakharyas left
jnthn Ah, I thought we'd got a makefile dependency on it... 22:16
22:18 zakharyas joined
lizmat compiled and installed ok now 22:19
running spectest now 22:20
oops, only configured ok until now... making now 22:21
jnthn lizmat: For me, the latest stuff shaved nearly half a minute or so off spectest on JVM.
lizmat: Hopefully it's a little faster for you too :) 22:22
lizmat hopes so too
timotimo the stage after optimize for moarvm takes reeeeally long if you're not optimized 22:23
lizmat btw, how do you run the spectests on JVM, without using the evalserver ?
timotimo Unhandled exception: Cannot invoke this object - this is the one i should see, right? for Pod::To::Text? 22:24
well, i get to a hello world \o/
lizmat if I just do 'make j-spectest' , after about ~30 test files, it starts to say 'no subtests run'
jnthn lizmat: No, I just make spectest 22:28
timotimo: Yeah, though I think it's for Test.pm...
lizmat with an TEST_JOBS setting ?
timotimo you are right.
raydiak has to run and may be gone for many hours, but just so someone knows, day 8 advent post is in good shape and will be ready on time 22:29
22:30 rurban1 joined 22:31 denisboyun joined
jnthn lizmat: Yeah, 6, iirc 22:31
lizmat ah, not 8 then 22:33
(which is what I use)
22:34 rurban1 left
lizmat t/spec/S05-mass/properties-script.rakudo.jvm ................. No subtests run :-( 22:35
runnint with TEST_JOBS=6 now 22:36
hoelzro timotimo: ping 22:42
lizmat Java HotSpot(TM) 64-Bit Server VM warning: Exception java.lang.OutOfMemoryError occurred dispatching signal SIGINT to handler- the VM may need to be forcibly terminated 22:44
Exception in thread "Thread-0"
Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "Thread-0"
jippie!
:-(
so I still can't do a JVM spectest completely 22:45
diakopter exceptionhandler threw an exception
not very handled
22:47 BenGoldberg joined
timotimo hoelzro: pong 22:48
hoelzro timotimo: is Starbound any good?
22:51 filly joined 22:52 woolfy left
TimK1 r: sub f(Array[Int] $a) { }; my Int @a = 1,2,3; f(@a) 22:56
camelia rakudo-jvm de5c8f: OUTPUT«Nominal type check failed for parameter '$a'␤ in block at /tmp/TmgGpkNk2u:1␤ in any eval at gen/jvm/stage2/NQPHLL.nqp:1086␤ in any evalfiles at gen/jvm/stage2/NQPHLL.nqp:1292␤ in any command_eval at gen/jvm/stage2/NQPHLL.nqp:1196␤ in any com…»
..rakudo-parrot de5c8f: OUTPUT«Nominal type check failed for parameter '$a'; expected Array[Int] but got Array[Int] instead␤ in sub f at /tmp/tMT7em56P6:1␤ in block at /tmp/tMT7em56P6:1␤ in any at /tmp/tMT7em56P6:1␤ in any at gen/parrot/stage2/NQPHLL.nqp:1146␤ in any…»
TimK1 Is that me or Rakudo getting confused? (Or a little of each?) 22:57
jnthn Rakudo.
TimK1 OK. Thx.
timotimo hoelzro: i like it. i've been playing with two friends
jnthn sub f(Int @a) { ... } works
timotimo the linux port is fine
hoelzro I saw it like a week ago; the idea intrigued me
might have to pick it up =)
timotimo there's a post on their blog telling you when the right moment to join the beta is 22:58
lizmat parrot spectest ok
timotimo at the moment you can expect frequent wipes of your worlds/characters, i bet
and crazy loads of balancing changes
lizmat seems to have gone down from 480 seconds to 440 seconds wallclock
hmmm... that was testjobs=6
hoelzro ouch
lizmat trying again with testjobs=8
timotimo 10%?
22:59 Piers_ joined
Piers_ So... if a chap is interested in playing with Perl 6 and concurrency again, is it worth trying to play with MoarVM, or am I better off with the Java build? 23:00
timotimo you'd have to wait a bit longer to play with concurrency on moarvm i'm afraid
we just got to hello world this week
Piers_ grins.
Congrats!
timotimo (though hello world on rakudo means a whole lot already) 23:01
hoelzro isn't sure if 10% was referring to a Perl 6 thing or a Starbound thing
Piers_ Yeah.
timotimo hoelzro: that was the speed improvement
a bad estimae, though. would be more like 8%
hoelzro ah, ok
timotimo if you'd like, we can hop onto mumble or something and i'll host a game (but i'm out of the house right now) 23:02
hoelzro maybe another night
almost bed time for me =) 23:03
jnthn Yes, JVM is the thing to use to play with concurrency stuff
timotimo sure
Piers_ Any timetable for getting concurrency into Moar? 23:05
jnthn Piers_: I'm currently thinking that by March/April is realistic. 23:06
lizmat down from 480 to 420, so about 12.5% faster in the spectest (for rakudo)
Piers_ jnthn: Brilliant.
Can't remember if I said, but the Perl 6 stuff at YAPC::Eu this year was about the only stuff I thought was exciting. 23:07
jnthn will try to have even more exciting things for YAPC::EU 2014 :) 23:08
colomon Piers_! o/
jnthn: please don't wait that long to be exciting. ;)
Piers_ Hmm... do I need a specific nqp branch to build rakudo@nom? Just did 'perl ./Configure.pl --gen-nqp --backends=jvm' and had it barf with NoSuchMethodError 23:09
lizmat is tired of parler franglais all day, with some double dutch intermixed
sleep&
colomon Piers_: should be automatic.
I always do it perl ./Configure.pl --gen-parrot --backends=jvm 23:10
Piers_ Hrm... hang on, I'll nopaste the error.
23:10 btyler_ joined
colomon (and yes, that makes no sense to me.) 23:10
jnthn colomon: you...what... :)
colomon jnthn: is that not how you're supposed to do it?
jnthn colomon: I'm just a bit surprised at the mix of --gen-parrot and --backends=jvm :)
colomon: Does it add parrot into the backends list automatically? 23:11
Piers_ Error: gist.github.com/pdcawley/41c4e3b44bebea9cef5f
colomon jnthn: … I don't think so?
Piers_ Arse. Meant --gen-nqp
lue
.oO(The Norweigian Blue 'asn't 'ad its coffee this mornin'!)
Piers_ (which is what I actually _typed_)
colomon jnthn: trying to build from scratch now
jnthn Piers_: Hm, somebody else got that earlier... 23:12
Piers_: Was this a clean checkout?
Piers_ Think so. I'll zap it.
23:12 btyler left
hoelzro so, I've been looking at that odd assignment issue I discovered yesterday 23:16
I discovered that the --target=parse output differs, which is...odd
now I'm searching for a debugging technique to print out what matched a rule at the end of a rule in Perl6::Grammar
any pointers?
jnthn If it's just the one, then throw in a { say($/) } at the end of it... 23:17
Piers_ jnthn: That seems to have helped, btw - it seems to have got past where it fell over last time.
perigrin isn't used to Piers_ being named Piers_ ... *visibly* 23:18
hoelzro ok, wasn't sure if that work 23:20
*would work
maybe I should stop working...
23:20 Piers_ is now known as pdcawley
pdcawley perigrin: better? 23:20
jnthn pdcawley: Phew. :)
pdcawley Hmm... configure is now failing with an OutOfMemory error trying to build QAST.jar 23:21
colomon jnthn: just rebuilt from scratch with perl Configure.pl --gen-parrot --backends=jvm, seems to have worked perfectly, no obvious signs that it created a parrot 23:23
japhb Oh panda, why are you not doing anything at all?
perigrin pdcawley: *much* :) 23:24
23:25 btyler_ left
pdcawley I'm 'Piers' on the work IRC 'cos there's a pdm on there as well and boss's 'type two letters and hit tab' practice falls over hard in the face of that. So, I've accommodated him. 23:26
jnthn japhb: Earlier, it was 'cus it couldn't reach the module list on feather... 23:27
pdcawley tries setting JAVAC to 'javac -Xms500m -Xmx2g' before running Configure.pl to see if that helps. 23:29
Arse... no it won't 'cos it's running 'java'
23:30 denisboyun left
colomon pdcawley: what version of java are you running? 23:30
japhb jnthn: Oh, interesting. I wonder if it's having that problem again ....
Is there a way to get it to be verbose enough to tell?
(Because I can't install the debugger in order to use it, and panda seems to be locking up on startup) 23:31
jnthn japhb: I'd guess if left long enough it *should* get a network timeout?
japhb: Not sure how well debugger installs for JVM yet, btw. I did most of the other things it needed... 23:32
japhb: Bad news is that it doesn't know anything about threads yet so it gets REALLY confused.
japhb Oh, I bet!
jnthn Well, my 2012 YAPC::EU talk was "a debugger for Perl 6" and my 2013 one was "concurrency stuff", so I guess 2014 can be "a debugger with support for concurrent stuff in Perl 6" :) 23:33
japhb heh 23:34
jnthn Having the debugger actually be clueful about promises, channels, etc is probably a good idea. :)
japhb Oh yeah, I can connect to feather.perl6.nl:3000 just fine, but sending the GET /projects.json HTTP/1.0<ret><ret> just sleeps
WTH?
jnthn I've no idea how that's actually set up... 23:35
japhb Methinks there ought to be a fallback. Redundancy for the win. 23:36
pdcawley colomon: 1.7.0_25 23:37
on a mac.
pdcawley tweaks nqp/tools/build/Makefile-JVM.in
tries again.
hoelzro so...
pdcawley jnthn: There's a perl 6 debugger? 23:38
hoelzro <EXPR('e=')>
pdcawley awesome!
23:38 Grrrr left
hoelzro that's expressions with a precedence greater than e? 23:38
23:38 Grrrr joined, tgt left
pdcawley Yay! That built it. 23:39
colomon pdcawley: that's actually more …. yay!
what did you change?
pdcawley tweaked Makefile-JVM.in to add -Xms500m -Xmx2g to the 'JAVA' variable. 23:40
lue hoelzro: no, that just means precedence level 'e'.
pdcawley Pretty much cargo culting from README.md
hoelzro lue: so what does that actually match? 23:41
jnthn finally found time to scribble a short blog post: 6guts.wordpress.com/2013/12/07/a-fe...k-updates/
23:42 mtk left
lue Any operator registered as having a precedence level of 'e'. In rakudo's grammar, this would be anything that falls under the %list_prefix category (search for the INIT { } block) 23:42
hoelzro I see
hmm
23:43 pecastro joined
lue I'm sure jnthn or others can tell you if there are such things as e> or d<= . I'm guessing not :) . 23:45
hoelzro I don't see anything tagged with list_prefix =/
jnthn The thing past to EXPR is actually a precedence limit.
23:45 mtk joined
jnthn Encountering aAnything looser causes EXPR to consider its work done. 23:46
23:46 mtk left 23:47 mtk joined
hoelzro I see 23:47
it seems like parsing this one line alters some state in the parser somehow 23:48
and makes things angry.
jnthn Which one line?
23:48 thou left
hoelzro rt.perl.org/Public/Bug/Display.html?id=120704 23:49
"@!values[0] = 1;"
I've been trying really hard to fix this one
but not making a lot of headway =/
jnthn That's not a parsing issue at all.
hoelzro it's not?
hoelzro headdesks
jnthn I don't think so.
hoelzro or head tables
jnthn I think it correctly gets interpreted as 23:50
my $value = (@!values[0] = @!values.pop);
So the assoc is OK
But the problem is then when it gets sent down for code-gen.
hoelzro I assumed it was, considering that perl6 --target=parse produces wildly different output if that line is commented out
jnthn We emit the code to evaluate @!values[0] before the code to evaluate @!values.pop. For most operators this doesn't matter. For = it rather does... 23:51
Thing is, that worked at some point.
hoelzro then why does it work if I comment out line 12?
jnthn Line 12 of...? 23:52
r: my @values = 1,2; my $value = (@values[0] = @values.pop); say $value; say @values; 23:53
23:53 PacoAir left
camelia rakudo-parrot de5c8f, rakudo-jvm de5c8f: OUTPUT«2␤2␤» 23:53
lue jnthn: the bug report has a odd.t attachment
jnthn Oh...
colomon 's Mac compiles the JVM stage parse faster than his Linux box does. I think the Linux box wins handily compiling Parrot stage parse. Weird.
hoelzro sorry, maybe I should've been more specific 23:54
the standalone example works
(which lead to a lot of headscratching)
lue r: class Foo { has @!values = 1,2; method Bar { my $value = (@!values[0] = @!values.pop); say $value; say @!values; } }; Foo.Bar; 23:55
camelia rakudo-jvm de5c8f: OUTPUT«Cannot look up attributes in a type object␤ in method Bar at /tmp/bk9dWyUPXD:1␤ in block at /tmp/bk9dWyUPXD:1␤ in any eval at gen/jvm/stage2/NQPHLL.nqp:1086␤ in any evalfiles at gen/jvm/stage2/NQPHLL.nqp:1292␤ in any command_eval at gen/jvm/…»
..rakudo-parrot de5c8f: OUTPUT«Cannot look up attributes in a type object␤ in method Bar at /tmp/vbFuQ07kG5:1␤ in block at /tmp/vbFuQ07kG5:1␤ in any at /tmp/vbFuQ07kG5:1␤ in any at gen/parrot/stage2/NQPHLL.nqp:1146␤ in any eval at gen/parrot/stage2/NQPHLL.nqp:1133␤…»
jnthn um...oh...wtf.
lue r: class Foo { has @!values = 1,2; method Bar { my $value = (@!values[0] = @!values.pop); say $value; say @!values; } }; Foo.new.Bar;
camelia rakudo-parrot de5c8f, rakudo-jvm de5c8f: OUTPUT«2␤2␤»
colomon pdcawley: I don't have any difficulties compiling JVM Rakudo from scratch on my Mac. 10.8, Java 1.7.0_04-b20
lue r: class Foo { has @!values = 1,2; method Baz { @!values[0] = 5; }; method Bar { say @!values, "---"; my $value = (@!values[0] = @!values.pop); say $value; say @!values; } }; Foo.new.Bar;
camelia rakudo-parrot de5c8f, rakudo-jvm de5c8f: OUTPUT«1 2---␤2␤2␤» 23:56
jnthn hoelzro: OK, so...I see what you're seeing with commetning that line out.
OK, there is more going on there then. wtf.
hoelzro ok, good
I'm not crazy!
lue Oh Oh Oh! hoelzro: get rid of the parens around (84, 85), what happens then?
hoelzro hmm
hoelzro tries
still broken =/
pdcawley colomon: Strange. 23:57
lue r: class Foo { has @!values = (1,2); method Baz { @!values[0] = 5; }; method Bar { say @!values, "---"; my $value = (@!values[0] = @!values.pop); say $value; say @!values; } }; Foo.new.Bar;
camelia rakudo-parrot de5c8f, rakudo-jvm de5c8f: OUTPUT«1 2---␤2␤2␤»
hoelzro I've tried adding a bunch of output to the grammar/actions to try and understand it better, but I don't think I have enough experience to find out what's really going on.
pdcawley colomon: 10.9 here.
hoelzro I'd like to fix this myself
pdcawley Anyway, it built now.
hoelzro because it "seems" simple enough for me to fix
23:58 mtk left
jnthn hoelzro: Interestingly, if you just mention @!values on the last line of that method it's enough. 23:58
lue hoelzro: does it happen on the first pop or the second? /me copies over the file now locally
jnthn uh, not
hoelzro lue: it should be screwed up for both
jnthn Yeah, there has to be the assignment there
hoelzro but it works for length 1, I believe 23:59
because @!values[@!values.end] == @!values[0] there
so I probably could omit the second call to pop
this came up while I was implementing a min heap for fun

Powered by Cro and the Raku® Programming Language.

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

Thank you!