»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.
raydiak mvuets: thanks for jumping in and fixing things...glad to have you here :) 00:04
tony-o invalid status 297373760 in GC orchestrate 00:12
00:28 cognominal joined 00:32 gfldex left
jnthn masak: In answer to "when a class A keeps calling into class B..." - feature envy. 00:43
mvuets sleep & 00:51
00:52 mvuets left 00:59 prime joined
timotimo tony-o: can you build your moarvm with ASAN? and before that, maybe turn on MVM_DISABLE_JIT (set it to some non-empty value) or MVM_DISABLE_SPESH 01:38
01:38 colomon left 01:44 Util joined 01:48 pmurias left 01:50 bjz left, bjz joined 02:01 rurban joined
[Coke] notes we have a bunch of flappy tests commented out of t/spectest.data 02:08
test -files-, I mean. 02:09
jdv79 flabby or flappy?
[Coke] we should probably re-enable those immediately after the release.
flappy == pass/fail "randomly" 02:10
dalek kudo/nom: 1385296 | coke++ | tools/update (2 files):
prefer -'s

  (most of the other tools use -, not _)
02:12
[Coke] jnthn++, moar looks much better today.
02:21 kurahaupo left
dalek rl6-roast-data: d275310 | coke++ | / (4 files):
today (automated commit)
02:36
02:45 JimmyZ joined, JimmyZ is now known as Guest84347, Guest84347 is now known as JimmyZ_ 03:12 perturbation left 03:15 noganex_ joined 03:18 noganex left 03:26 skaufman joined 03:29 PZt joined
tadzik moritz: we should synchronize one day and set up this service properly 03:37
04:03 mr-foobar left
ugexe language: perl 04:04
perl:
- '5.20'
env:
- BACKEND=moar PANDA_SUBMIT_TESTREPORTS=yes
- BACKEND=jvm PANDA_SUBMIT_TESTREPORTS=yes 04:05
- BACKEND=parrot
matrix:
allow_failures:
- env: BACKEND=jvm
- env: BACKEND=parrot
fast_finish: true
before_install: - git config --global user.name "TravisCI" - git config --global user.email $HOSTNAME":not-for-mail@travis-ci.org" - git clone git://github.com/travis-perl/helpers ~/travis-perl-helpers - source ~/travis-perl-helpers/init # helpers for parallel testing/testing dirs - git clone github.com/tadzik/rakudobrew ~/.rakudobrew
oops, sorry
TimToady ugexe: if you're using irssi you can set paste_verify_line_count = 1 and it'll ask if you really mean it 04:23
ugexe another notch on the ol irssi config 04:31
04:36 j4janicej is now known as j4jackj 04:37 JimmyZ_ left 04:41 j4jackj is now known as j4janicej 04:50 KCL_ joined 05:01 rurban left 05:10 BenGoldberg left 05:13 kaare_ joined 05:37 kurahaupo joined 05:59 gfldex joined 06:29 kurahaupo left 06:30 kurahaupo joined, kurahaupo left
moritz tadzik: first step: put the code into a git repo, give me the URL 06:34
06:36 kurahaupo joined 07:07 Mso150_p_l_w joined, Mso150_p_l left 07:25 gfldex left 07:46 JimmyZ_ joined 08:18 darutoko joined 08:30 Possum left 08:42 mr-foobar joined 08:43 Possum joined 08:48 Possum left
dalek p: 817019e | TimToady++ | src/QRegex/NFA.nqp:
delete SUBRULE edges after use

Added a better NFA dumper. Using this, I discovered that the NFA engine was spending 6-7% of its time just ignoring subrule edges. These edges are not used after subrules are merged, so they can be spliced out before the NFA engine ever sees them.
08:55
08:56 Possum joined
dalek p: b1cecf6 | TimToady++ | tools/build/MOAR_REVISION:
bump moar
08:59
kudo/nom: 25796d2 | TimToady++ | tools/build/NQP_REVISION:
bump nqp
09:00
09:06 JimmyZ_ left 09:08 molaf joined 09:11 JimmyZ_ joined
JimmyZ_ TimToady: the above commit increases Stage parse : ~62s -> 63.711s here 09:12
09:13 lizmat joined
TimToady do you have anything running in the background? ran faster here, once I stopped firefox... 09:17
JimmyZ_ TimToady: Nope, I exited all things 09:18
and ran it twice to confirm it 09:19
TimToady interesting... 09:20
maybe splice is slow on your machine for some reason...
JimmyZ_ Win32 on X86 without jit 09:21
let me try the old one again 09:22
09:22 JimmyZ_ left, woolfy joined 09:25 rindolf joined 09:31 mvuets joined 09:32 JimmyZ_ joined 09:34 JimmyZ_ left
TimToady zzz & 09:36
lizmat gnight TimToady
09:36 JimmyZ_ joined
lizmat fwiw, I'm not seeing much difference in parse times 09:36
JimmyZ_ TimToady: sorry, it's not the above commit, but I can't ~62s now 09:37
*can't get back
TimToady well, at best it would be about a 1% improvement, since the nfa is only about 10% of the parse time 09:39
but there's still a bunch of improvements that can be made, lots of useless epsilons and such that can be eliminated 09:40
zzz really &
JimmyZ_ good night 09:55
dalek kudo/newio: 1385296 | coke++ | tools/update (2 files):
prefer -'s

  (most of the other tools use -, not _)
09:56
kudo/newio: 25796d2 | TimToady++ | tools/build/NQP_REVISION:
bump nqp
kudo/newio: dd6fc0a | (Elizabeth Mattijsen)++ | tools/ (3 files):
Merge branch 'nom' into newio
kudo/newio: 0f4ec52 | (Elizabeth Mattijsen)++ | src/core/PIO.pm:
PIO.say should return True, not 1
10:00 cibs left 10:01 cibs joined 10:03 Isp-sec joined 10:04 jluis joined 10:08 psch joined 10:11 brrt joined, cibs left 10:12 jluis left 10:15 anaeem1 joined 10:18 cibs joined 10:23 jluis joined 10:25 anaeem1 left 10:26 KCL joined 10:29 KCL_ left 10:34 jluis_ joined 10:35 telex left, jluis left 10:36 telex joined 10:39 jluis__ joined 10:40 brrt left, jluis_ left 10:44 jluis__ left
masak antenoon, #perl6 10:46
jnthn: re "feature envy" -- I suspected so. but, so the "feature" here is the control flow itself? 10:47
10:50 spider-mario joined 10:52 smls joined
smls o/ 10:53
Happy Advent Sunday, #perl6! 10:54
10:56 jluis__ joined 10:58 Mso150_p_l_w left
moritz \o 11:00
so far we have only one sign-up for the advent calendar
that's too few
I can probably write a "State of Perl 6" for tomorrow
but we need more, otherwise it won't work 11:01
dalek : 069bb68 | moritz++ | misc/perl6advent-2014/schedule:
[perl6advent] claim day 1
mvuets \o/ 11:02
a pleasant surprise - didn't realize it starts so soon 11:03
moritz it only starts if people commit to contributing
11:03 rurban joined
smls
.oO( idea for an advent calendar post: the special global-ish dynamic variables $*OUT, $*ERR, $*CWD, $*TMPDIR, $*HOME, $*SPEC, $*SCHEDULER etc. and how they can make things more flexible/clear/convenient than in Perl 5 )
11:04
moritz smls: sign up and write about it! 11:05
smls I'm not qualified enough
moritz sure you are
just pick those (sub)topics that you know a bit 11:06
smls I'll think about it
moritz ++smls 11:07
timotimo i wish gtk-simple wasn't broken >_< 11:10
11:10 gfldex joined
smls How come the gtk-simple repo is part of the official perl6 org on github? What's so special about it? 11:12
masak enjoyed www.slideshare.net/ScottWlaschin/fp...ildstufflt 11:14
timotimo smls: it was part of a repository jnthn had made for one of his talks, i extracted it for him, but didn't think it'd be fitting to put it under timo/ 11:17
moritz smls: also people can put repos in the perl6 org if they want everyone[tm] to have write access 11:19
11:20 denis_boyun_ joined
mvuets what is the state of Test.pm? 11:22
is there a reference implementation?
moritz mvuets: there's a Test.pm shipped with rakudo
mvuets it spits warnings 11:23
wondering if worth fixing
moritz mvuets: what kind of warnings?
mvuets because: ## This is a temporary Test.pm to get us started until we get pugs's Test.pm / ## working. It's shamelessly stolen & adapted from MiniPerl6 in the pugs repo.
moritz seems it has evolved into a full one by now, with the comment being out of date 11:24
mvuets skip_rest emits this: use of uninitialized value of type Any in numeric context in sub skip_rest at lib/Test.pm:192
yeah, seems so. pugs' one does not even have subtest()
moritz still something more structured (see Test::Builder in p5) wouldn't be bad
mvuets: yes, worth fixing
mvuets so the testing api is defined in s24. does it mean Test implementation is up to language vendors? 11:25
moritz mvuets: correct 11:26
11:26 denis_boyun_ left
dalek kudo/nom: a9d4c84 | masak++ | lib/Test.pm:
[Test.pm] remove outdated comment
11:26
moritz mvuets: though of course language implementors are free to still from other language implementors (as far as license allows it)
mvuets hmm, do i understand it right that cannot be a Test distro on CPAN6 then?
dalek kudo/nom: 91d8995 | masak++ | lib/Test.pm:
[Test.pm] update copyright date
11:27
moritz mvuets: well, not right now, though we do plan to support loading modules by authority (not just name)
timotimo maybe it'd be good enough to hotfix the GTK::Simple module so that i can guarantee it works for the advent calendar
but what cool thing could i build with it? :S
mvuets another question: perl6/roast.git seems missing subtest() tests - shall i try writing some? 11:28
moritz mvuets: yes please 11:29
timotimo: do you know anything about the run time performance of r-j vs r-m?
mvuets and how much we care for having smartlinks (those nifty things linking tests back to specs and hence embedding code examples in synopsis-es)? 11:30
moritz mvuets: not very much, at the moment
timotimo moritz: no :(
i haven't touched r-j in a long time
mvuets ok. because i found much with --check and TODO actually mentions it 11:31
masak who's doing Dec 1 in the Advent calendar? :) 11:33
moritz masak: me
11:33 denis_boyun joined
masak great 11:35
ahh, but the schedule is far too empty for next week -- github.com/perl6/mu/blob/069bb68e7...4/schedule
moritz masak: agreed 11:36
masak preferably we should have all of next week, and preferably coming weeks, planned out by the end of today...
moritz masak: so, sign up :-)
masak ok, I can volunteer for one or two slots for next week.
just gotta figure out what to write about.
"Perl 5/Perl 6 comparison" sounds like a topic I could do. 11:38
moritz masak: also a summary of your current macro musings would be intersting 11:39
masak yes, but that would work better towards the end of the month, as I expect to have mused quite a bit more by then. 11:40
tadzik moritz: it's on github.com/perl6/ecosystem/tree/master/server 11:44
moritz tadzik: ok, great
tadzik I'll see how out of date it is, but probably not too much
it says 2 years ago 11:45
11:46 gfldex left
dalek osystem: 82861f8 | tadzik++ | server/ (2 files):
Update the module server
11:48
tadzik now it's up-to-date :)
11:50 JimmyZ_ left
moritz tadzik++ 11:51
11:51 JimmyZ_ joined
moritz huh, perl6-m startup time is 330ms here 11:53
wasn't it down to 150ms or some months ago? 11:54
JimmyZ_ moritz: yeah, re-added the eager de serialization due to some precomp bug
moritz :( 11:55
jnthn And introduced a different bug in the process, no? 11:56
jnthn doesn't know what to write about for advent
Don't think I can find time for a post in the first week (too much work/distractions) 11:57
But can do something after that.
moritz jnthn: you could always do a summary of what you worked on, or an example from some of the talks you gave, or a braindump what you're working on 11:58
JimmyZ_ jnthn: yeah
jnthn Is this the first advent we've had with Rakudo on Moar?
moritz it would be awsome if we could get our GSoC students to write about their projects for the p6advent
jnthn: yes 11:59
jnthn Hah!
moritz jnthn: I just checked, last year's "state" post said nqp was bootstrapped on MoarVM, but rakudo wasn't running on it
jnthn Then maybe I should write one about MoarVM :)
moritz jnthn: you should :-)
jnthn I know a little about it already, after all... :)
timotimo we could have a post that we can point annoying naysayers to who say "oh, when's perl6 going to be rewritten from scratch the next time? next week and the week after that again?" 12:02
as in "how do we know MoarVM is The One™" 12:03
moritz gist.github.com/moritz/7df64717d8dd772a87a0 # preview of the first advent post, feedback welcome
afk 12:04
dalek : 3095993 | masak++ | misc/perl6advent-2014/schedule:
[adv2014/schedule] claim two slots
masak others: please claim advent slots for next week!
masak does Uncle Sam pointing-to-YOU pose 12:05
12:05 rindolf left
timotimo moritz: i wouldn't say moarvm is significantly faster at run time against the JVM. i think the jvm gets pretty good when it's heated up. especially for native stuff and tight code 12:05
masak moritz: s/de-factor/de facto/
moritz: s/supported as a backend/supported as backends/ 12:06
12:07 ab5tract joined
ab5tract o/ #perl6 12:07
yoleaux 24 Nov 2014 18:59Z <tony-o> ab5tract: fixed the HTML::Parser::XML module, it works as expected and it works with the advent sample ugexe provided
smls moritz: "make the list-y types more transparent, list iteration faster, and more transparent when a list will flatten" -- maybe avoid repetition of 'transparent'
timotimo would suggest not using "more transparent" twice
%)
masak maybe not repeat "transparent" in the sentence? :) 12:08
masak suggests less repetition
ab5tract so i was reading about the shortfall of advent entries
in the logs
smls moritz: s/will in some details/will fill in some details/ 12:10
timotimo ab5tract: would you like to contribute? :)
ab5tract i have some stuff written about the difference between set coercers vs composers in relation to sets
ab5tract suggests to himself less repetition for himself 12:11
timotimo: indeed. my goal was to have an async scraper for the advent calendar with a ncurses interface 12:12
but i've been repeatedly stymied on that front. it boils down to weird things happening at the HTTP::UserAgent layer.. I could try replacing it with LWP::Simple 12:13
masak ab5tract: please claim a slot in github.com/perl6/mu/blob/master/mi...4/schedule
timotimo ugh, ncurses 12:14
ab5tract timotimo: :) ... but it's a well-known library that can highlight NativeCall while also potentially wow-ing folks if one were to put an async-y layer over the top 12:16
timotimo mhm mhm
don't forget ncurses is not reentrant :)
ab5tract i doubt ncurses would ever let me forget it :) 12:18
masak: pull request sent
itz lose the Oxford Comma :)
timotimo that reminds me i wanted to build a nice high-level console UI library thingie
dalek : a907f24 | ab5tract++ | misc/perl6advent-2014/schedule:
Update schedule
: 0323a69 | 唐鳳++ | misc/perl6advent-2014/schedule:
Merge pull request #5 from ab5tract/patch-1

Update schedule
timotimo with all the features of ANSI5
ab5tract timotimo: pure perl? 12:19
audreyt moves quickly :)
another thing i can discuss are the (potential) changes to the set comparison operators, of which my patch is sitting in limbo because i get distracted easily 12:20
timotimo pure perl, aye
ab5tract timotimo: that sounds like fun. hoelzro and i were discussing the repl environment, and "wouldn't it be nice to have an simple plugin interface" 12:22
timotimo to be honest, "pure perl" tends to mean "moarvm with asynchronous IO and multithreading" these days :P 12:23
ab5tract timotimo: i thought that was the whole point of perl 6 :P 12:24
ah, i see what you mean.. the moarvm bit
timotimo that or JVM
but as i said, i haven't touched that in a while
psch wonders if it makes sense to make perl6-bench use the evalserver 12:25
ab5tract it feels a bit undernourished, that backend. but that's a surface level observation.. i've never seen a blog post about NativeCall on JVM, for instance. i thought it would be fun to write an interface to Processing, but didn't get much farther than imagining it
timotimo psch: perl6-bench already subtracts start-up and compile times from benchmarks 12:28
psch timotimo: okay. but vm-warmup is a big-ish thing on jvm, isn't it? i guess i'd have to look how perl6-bench does it's benching in the first place 12:29
JimmyZ_ jnthn.net/papers/2014-yapceu-performance.pdf said while-empty was 4x slower than perl5, I think now they are the same :P
psch i guess that's actually the point of good benchmarks, to have code that will be optimized well eventually 12:30
12:33 denis_boyun left
psch huh, ./bench fails with "Missing or wrong version of dependency 'src/gen/m-BOOTSTRAP.nqp'" but the same perl6 from my env works when called directly..? 12:34
JimmyZ_ make clean in rakudo ...
and then rebuild 12:35
psch yeah, i've nuked install/ and ran git clean -xdf
timotimo psch: running a rakudo-j benchmark with all tests takes a LONG time. 12:36
12:36 kjs_ joined
psch timotimo: oh, i expect that - i don't really plan to run everything, i just want to have a hint re: mvm vs jvm on one or two tests 12:37
12:38 smls left
timotimo ah, so like --tests-tagged or --tests? 12:38
ab5tract timotimo: if it takes so long, it surely should use the evalserver ? 12:39
psch uh, probably? i haven't looked too closely yet at how i have to use perl6-bench... :)
timotimo the problem with the evalserver is that jit warmup is potentially shared 12:40
which would make the calculations be off
ab5tract ah 12:41
timotimo it'd be impossible to remove it just by subtracting timings and startup timings etc
ab5tract speaking of the spectest... is verbose output stored anywhere? 12:42
psch heh, maybe advocating the use of jakudo as a service is a good idea, then we can publish evalserver p6bench and claim that's real-world-use... :)
s/service/daemon/
timotimo you can run the "mumblemumble_summary" script to get verbose output 12:43
ab5tract find . | grep mumble # nothing 12:45
timotimo yeah, "mumble" stands for "something i don't remember" 12:49
ab5tract :P
mvuets haha 12:50
12:50 haroldwu left 12:51 haroldwu joined
ab5tract what's the invocation for running a single test with harness? the quicker i can get the set comparator patch working, the sooner i can write a "how to start hacking on rakudo" advent post :) 12:51
timotimo you can make t/spec/S123-blah/foobar.t 12:52
12:54 colomon joined, jluis__ left
ab5tract timotimo: doh.. that makes sense. i was trying against the file which failed in spectest, which was moar specific. thanks 12:54
12:55 kjs_ left 12:56 denis_boyun_ joined 12:57 gfldex joined 12:58 spider-mario left
bartolin ab5tract: you can use "perl t/harness --fudge --moar t/spec/S123-blah/foobar.t" 13:01
ab5tract bartolin: for the fudge :) 13:02
bartolin ab5tract: oh, that was not what you wanted. sorry for the noise ;-) 13:03
ab5tract not at all! that probably should have been written as "thank you for the fudge" :) 13:04
bartolin *g*
dalek : c259106 | nige123++ | misc/perl6advent-2014/schedule:
Update schedule
13:06
: c9594ac | moritz++ | misc/perl6advent-2014/schedule:
Merge pull request #6 from nige123/patch-1

Update schedule
psch re: advent calender: i guess struggling with understand code-gen for the jvm backend is not quite worthy of a post, but that's about what i can think of i could write about
moritz ab5tract: please /msg me your email address so that I can invite you to the advent calendar
psch: I'd like to read that
psch moritz: alright, i guess i have to pick a day then 13:08
and take more extensive notes as well...
i'm thinking an overview of "what does running perl6-j do and how did i get to understand it" is somewhat of a decent title/summary 13:10
looking at topic-brainstorming that should probably come before NativeCall-jvm, if anyone writes about that 13:11
13:13 noganex joined
psch i'll try to pin down a date until this evening, got some other stuff to do about now o/ 13:13
13:13 spider-mario joined 13:16 noganex_ left
dalek : c3bb8dc | (Elizabeth Mattijsen)++ | misc/perl6advent-2014/schedule:
Commit to doing an advent piece on 5 dec
13:20
moritz anybody else in need of a perl6advent.wordpress.com invite? 13:25
lizmat having claimed 5 dec, please note that I'm not stuck to that date and could do it later :-) 13:28
colomon we don't have a day 1 on the schedule? 13:29
moritz we have
or did I forget to push?
bartolin colomon: you probably only looked at the diff? 13:30
colomon yes
bartolin++
moritz that shows only 3 lines of context :-)
colomon moritz++ 13:31
moritz colomon: gist link for a preview is in the backlog
13:31 PZt left 13:43 bjz left 13:44 bjz_ joined 13:53 denis_boyun_ left, ggoebel111111114 joined 13:57 dj_goku left
ab5tract hmm.. i get a fair number of failed tests when running make spectest against nom. are there a specific set of steps that need to be taken when switching between branches on rakudo.git in order to have a "clean" run of the spec suite? 13:59
s/are there/is there/
i'm already rm'ing all .moar files in t/spec before running make spectest after switching branches. but should i be doing a distclean or something? 14:01
timotimo what symptom are you seeing? 14:02
lizmat ab5tract: are you on OS X ? 14:04
tony-o timotimo: gonna rebuild moar with ASAN, the two envirovars didn't stop the errors
ab5tract timotimo: gist.github.com/ab5tract/211fd789b8071fce9599
lizmat: yup
timotimo tony-o: OK, thanks
lizmat welcome to my world :-(
ab5tract :/ 14:05
lizmat spectests on OS X can only be described as "flaky" at the moment :-(
tony-o timotimo: should i just do --moar-options='ASAN=1' ?
ab5tract ok..
timotimo i think it has to be --moar-options=--asan
tony-o ok
timotimo [Coke]: if i write an article about GTK::Simple - maybe in combination with cairo - could you help me make sure everything works on OSX, too? IIRC you got it to work on your machine
tony-o i have gtk working on my machine too .. 14:06
(osx)
dalek kudo/newio: b0605fc | (Elizabeth Mattijsen)++ | src/core/IO (2 files):
Add missing .spurt methods
kudo/newio: 8b565a6 | (Elizabeth Mattijsen)++ | src/core/ (2 files):
Streamline spurt functionality
kudo/newio: db00491 | (Elizabeth Mattijsen)++ | src/core/IOU.pm:
Set that on successful open/mkdir/spurt
ab5tract lizmat: so that begins to explain why they don't even necessarily appear consistent
lizmat indeed...
ab5tract makes chasing down the bugs pretty difficult .... :( 14:07
timotimo tony-o: did you have to change anything, like the library name in "is native"?
ab5tract but, at least my set comparator changes pass the set related tests :D
lizmat it's some memory corruption that seems related to the way OS X does ASLR (en.wikipedia.org/wiki/Address_space...omization)
timotimo or build gtk with brew or macports?
tony-o timotimo: yea just leave off the .dylib or .so
lizmat that happens on moment X, but isn't showing up until Y (which is much later in process terms) 14:08
tony-o iirc we checked it in debian and it worked that wre without the .whatever
ab5tract lizmat: joy :( :(
[Coke] timotimo: sure, I'll be your os x guinea pig.
timotimo thank you
lizmat whether this is a problem of OS X, or simply more prominently visible on OS X due to something like ASLR
who knows
tony-o s/that wre/there/
lizmat jnthn at least now has access to an OS X machine so he can look at it 14:09
timotimo i know i can make GTK::Simple compile by removing the "use GTK::Gdk" line, maybe it'll also work to just put the GDK module's contents into the same file
[Coke] does not like this dizzy symptom. yikes.
tony-o are you sick too [Coke] ?
ab5tract lizmat: that is indeed good news
timotimo jnthn already found a few things on os x and made them better 14:10
dalek kudo/newio: a9d4c84 | masak++ | lib/Test.pm:
[Test.pm] remove outdated comment
kudo/newio: 91d8995 | masak++ | lib/Test.pm:
[Test.pm] update copyright date
kudo/newio: 0bdd9f3 | (Elizabeth Mattijsen)++ | lib/Test.pm:
Merge branch 'nom' into newio
lizmat was pretty sick yesterday and this morning...
some sleep here at the hackathon helped me a lot
fortunately, the office here has a nice area for a quick nap
tony-o i got sick yesterday on a bike ride still about 2 hours away from home
lizmat (which wasn't so quick in my case) 14:11
tony-o: that sucks...
tony-o timotimo: i get those errors with --asan and both env vars set 14:12
lizmat: being sick is always gross 14:13
ab5tract it is also good news that i've been worrying about my set changes breaking things, when in fact its endemic
lizmat as long as the test file runs ok by itself, you should be ok 14:14
at least, that's what I've been going by
jnthn Wow, so sickness... Hope y'all get well soon... 14:19
Yes, the OSX failures are a darn nuisance. Things are far cleaner for me than they were after a couple of patches, but still not clean :/ 14:20
14:22 dj_goku joined 14:26 dj_goku left 14:27 dj_goku joined, dj_goku left, dj_goku joined 14:31 guru joined, anaeem1_ joined, guru is now known as Guest20826, Guest20826 is now known as ajr 14:32 ajr is now known as ajr_
[Coke] tony-o: vomit and dizziness, that's me. 14:38
ab5tract jnthn: i'm thinking that these may also explain some of the seemingly random failures i've been having when working on the Adventoratuer 14:40
which, in the context of memory, start to make more sense... 'Malformed UTF-8', 'Could not open file: {$impossible-location}', etc 14:42
JimmyZ_ that's different thing.. 14:43
cosimo mvuets: pong
14:48 Vlavv left 14:53 zakharyas joined
dalek kudo/newio: 8c062d1 | (Elizabeth Mattijsen)++ | src/core/IOU.pm:
Pass any extra .new parameter on through
14:54
kudo/newio: c73c0b1 | (Elizabeth Mattijsen)++ | src/core/IO/Dir.pm:
A child of Dir is not necessarily a Dir
kudo/newio: 54d2711 | (Elizabeth Mattijsen)++ | src/core/CompUnit (3 files):
Initial IO::Path sanitation of CompUnit and friends
kudo/newio: d027745 | (Elizabeth Mattijsen)++ | src/core/CompUnit.pm:
Remove superfluous reference to $*SPEC
14:56
14:57 Possum left
ab5tract JimmyZ_: oh? 14:59
15:00 Vlavv joined, denis_boyun_ joined
lizmat ab5tract: fwiw, what you describe, sounds familiar to me 15:01
15:01 Possum joined 15:03 kjs_ joined 15:12 rindolf joined 15:14 ptc_p6 joined 15:17 anaeem1_ left 15:18 isleofmax joined 15:21 isleofmax left, anaeem1_ joined 15:25 smls joined
b2gills m: sub example ( $ --> Int | Failure ){ ... } # Is there a way to get this to work? 15:26
camelia rakudo-moar 91d899: OUTPUT«===SORRY!=== Error while compiling /tmp/kBW6J1HBRU␤Missing block␤at /tmp/kBW6J1HBRU:1␤------> sub example ( $ --> Int ⏏| Failure ){ ... } # Is there a way to g␤ expecting any of:␤ statement list␤…»
lizmat m: sub a(--> Int) { fail "foo" }: my $a = a' 15:27
camelia rakudo-moar 91d899: OUTPUT«===SORRY!=== Error while compiling /tmp/FTtSlZZom9␤Invocant colon not allowed here␤at /tmp/FTtSlZZom9:1␤------> sub a(--> Int) { fail "foo" }:⏏ my $a = a'␤»
lizmat m: sub a(--> Int) { fail "foo" }; my $a = a'
camelia rakudo-moar 91d899: OUTPUT«===SORRY!=== Error while compiling /tmp/D1b8x9o8yZ␤Two terms in a row␤at /tmp/D1b8x9o8yZ:1␤------> sub a(--> Int) { fail "foo" }; my $a = a⏏'␤ expecting any of:␤ argument list␤ postfix␤ …»
lizmat m: sub a(--> Int) { fail "foo" }; my $a = a 15:28
camelia rakudo-moar 91d899: OUTPUT«Type check failed for return value; expected 'Int' but got 'Failure'␤ in any return_error at src/vm/moar/Perl6/Ops.nqp:649␤ in sub a at /tmp/fjTCKn7wdu:1␤ in block <unit> at /tmp/fjTCKn7wdu:1␤␤»
15:29 kjs_ left
lizmat hmmm.. that feels like a bug, as I seem to recall a Failure should survive a type check like that... 15:29
jnthn It's up for debate. It really hoses a good bit of compile time optimization if you can't actually trust your types. 15:31
So I'm generally against such magic.
The Nil discontinuity in assignment is already bad enough.
masak (fewer exceptions in the type system)++ 15:32
sounds to me like Failure wants to be a bottom type, though.
b2gills So the only course of action is to fail immediately in that case 15:35
m: sub a(--> Int) { my $FATAL = True; fail "foo" }; my $a = a 15:36
camelia rakudo-moar 91d899: OUTPUT«Type check failed for return value; expected 'Int' but got 'Failure'␤ in any return_error at src/vm/moar/Perl6/Ops.nqp:649␤ in sub a at /tmp/T1aTovFSDM:1␤ in block <unit> at /tmp/T1aTovFSDM:1␤␤»
b2gills m: sub a(--> Int) { my $*FATAL = True; fail "foo" }; my $a = a
camelia rakudo-moar 91d899: OUTPUT«foo␤ in sub a at /tmp/oVOsj4nTzY:1␤ in block <unit> at /tmp/oVOsj4nTzY:1␤␤»
15:36 virtualsue joined 15:42 vendethiel joined 15:43 KCL left
vendethiel o/, #perl6 15:45
dalek kudo/newio: d8fe25a | (Elizabeth Mattijsen)++ | src/core/CompUnitRepo/Locally.pm:
Make sure we get a Str for a CompUnitRepo
15:46
kudo/newio: 31040e6 | (Elizabeth Mattijsen)++ | src/core/CompUnit.pm:
Handle path -> abspath migration
15:47 anaeem1_ left, anaeem1 joined 15:48 anaeem1 left
lizmat Patch -p2 hackathon dwindling down... 15:48
15:48 anaeem1_ joined 15:51 kjs_ joined 15:52 daxim left 15:56 Possum left 15:57 kjs_ left 16:01 kurahaupo left 16:02 Possum joined 16:07 kshannon_ left, Possum left 16:08 kshannon joined
smls moritz: Would "Data munging in Perl 6 vs Perl 5" be acceptable as a perl6advent topic? 16:10
Rather than focus on any particular Perl 6 feature, I'd do one or two small case studies for contrieved-but-practical usecases that involve reading stuff from a file, putting it into a data structure, transforming said data structure, and printing output. 16:11
I.e. the kind of quick-and-dirty scripting that sysadmins/powerusers/students/scientists etc. tend to use Perl 5 for in practice (as opposed to "professional programmers building full-fledged applications").
16:11 JimmyZ_ left
smls And hopefully show that Perl 6 can be a good replacement for Perl 5 in that area... :) 16:11
vendethiel smls: I actually have bad slides for a talk like that :) 16:12
smls I'd go through the Perl 5 and Perl 6 solutions for the case study side-by-side and step-by-step, noting not just superficial syntax differences, but how one would most idiomatically approach each step in each language.
vendethiel (that's pretty much 99% of my P6 use at $work)
smls vendethiel: Are the slide public? :) 16:13
*s
vendethiel smls: no, they're terrible :P
16:13 Possum joined 16:15 denis_boyun_ left
smls I've already come up with a tiny case-study that allows me to show off «.lines» (as a lazy list!), «.map», a sigspace'd regex, a hash, «.grep», «.classify», «.comb», «.sort», and string interpolation. 16:16
lizmat smls++
smls (in 12 LOC, some of which are boring 'say' statements...)
But I'd maybe like to have a second case-study for showing off hyper/meta ops and maybe Sets/Bags/Mixes etc. 16:17
lizmat Patch -p shutting down&
16:17 lizmat left
vendethiel smls: gist.github.com/vendethiel/69bad550355ec5455ebd 16:19
16:19 dwarring left
vendethiel few examples if you want 'em 16:19
smls vendethiel: Thanks 16:20
16:25 woolfy left 16:26 zakharyas left 16:30 zakharyas joined 16:35 spider-mario left 16:38 raiph joined 16:43 pecastro_ joined 16:44 pecastro left
timotimo .tell azawawi will you write a post on the advent calendar about farabi6? 16:46
yoleaux timotimo: I'll pass your message to azawawi.
timotimo a blog post about HTTP::Server::Async could be pretty neat, too 16:49
16:56 Possum left
smls Here's another idea for a perl6advent post I'd like to see: "Language Consistency & Correctness (a.k.a. Less Special Syntax & Magic)" 17:00
But someone more knowledgeable than me should write that one... :)
Although I did come up with some initial bullet points: gist.github.com/smls/116c4e55de020e17829a 17:01
^^ timotimo? moritz? lizmat? TimToady? :) 17:02
b2gills masak: Perhaps `Failure` could be a Role.
17:04 KCL_ joined
smls I know that topic of P6's goal of consistency/correctness (and by extension, extensibility) has been touched on multiple times already, but it would be nice to have a focused blog post written by an expert, that shows just how much that idea permeates P6's design... 17:04
17:05 atta left, atta_ joined 17:15 kurahaupo joined, kurahaupo left 17:19 dwarring joined
moritz smls: when I started to realize the reasons and implications around these factors, I wrote the 5-to-6 blog post series to try to explain them 17:21
17:22 kurahaupo joined
moritz smls: though I'm not sure I can summiarze it into a single blog post (espeically without feeling that I'm just rehashing what I wrote on my blog, and on perlmonks and mailing lists) 17:23
smls I don't think rehashing is necessarily bad 17:27
Lots of little bits information on a topic floating around various places on the net, don't have the same effect as a dedicated (and concise) blog post covering that topic specifically (and trying to explaining the underlying intentions and general patterns) 17:30
timotimo smls: i like the idea and the bullet points you've found so far :) 17:34
smls timotimo: Then adopt it :) 17:35
timotimo moi?
i'm hard at work thinking about potential topics as well 17:36
what do the cool kids do with async http servers these days? 17:37
17:37 zakharyas left
vendethiel
.oO( they scale )
17:37
timotimo well, we don't. we just crash :P 17:39
tony-o timotimo: what did the asan thing do? 17:47
or what does it do? 17:48
17:50 ptc_p6 left 17:51 ptc_p6 joined 17:52 FROGGS__ joined 17:55 Sqirrel left, Sqirrel joined
tony-o 17:56
17:56 FROGGS_ left, ptc_p6 left 18:03 perltricks joined
tony-o timotimo: also, jvm seems to work fine - i did 25k requests in 5m with wrk (10 threads) 18:06
dalek ake: 11580c8 | (Arne Skjærholt)++ | src/Snake/ (2 files):
Handle circumfix parens.
18:07
ake: 5570b0b | (Arne Skjærholt)++ | src/Snake/Grammar.nqp:
Update a TODO.
18:10 ab5tract left 18:12 xinming_ joined
moritz p6advent post for tomorrow scheduled for publishing 18:13
tony-o a little less than half of what i get from nginx ..
18:15 xinming left
vendethiel niiiiice 18:16
I'm so excited :3
colomon \o/ 18:17
perltricks ohhh cool. Is there still time to contribute an rticle? 18:18
vendethiel sure :)
perltricks vendethiel: who should I send it to? 18:19
vendethiel perltricks: github.com/perl6/mu/blob/master/mi...4/schedule
tony-o ooh the moar server beats out the nginx example in a slightly unfair comparison 18:20
perltricks vendethiel: thx :)
vendethiel "slightly unfair" hahaha 18:21
tony-o the nginx serves a static file while the moar version just writes out the socket
it server 48k requests in 10 minutes tho
with only 1 error
sorry, 5 minutes ^^ 18:22
jnthn tony-o: Wow... 18:25
So serve.
tony-o++
smls moritz: I just added a comment to your advent post gist (gist.github.com/moritz/7df64717d8dd772a87a0), is it too late? 18:26
tony-o hmm still getting this tho: Unhandled exception in code scheduled on thread 4476928000 18:27
Unhandled exception: When invoking is_bindable, Provided outer frame 0x7fb1b5158bf0 (MVMStaticFrame bind_one_param) does not match expected static frame type 0x7fb1b51589b0 (MVMStaticFrame )
only happens in moar
jnthn Yeah, I've run into that one in my photo app 18:29
I need to hunt it down
bboiab...
*bbiab
tony-o bbiab?
smls "be back in a bit" 18:30
tony-o ah
jnthn: i've tracked mine down to here but i'm not sure what i could do differently to avoid the problem: github.com/tony-o/perl6-http-serve...nc.pm6#L48
18:35 silug left 18:43 ajr_ left, Mso150 joined 18:44 rindolf left 18:45 rindolf joined, zakharyas joined
moritz smls: not too late, but maybe a bit too much; IMHO that's a topic for another post 18:46
smls: a "Where is rakudo fast? Where is it slow?"-post would be very welcome 18:47
18:47 silug joined
timotimo smls: are you sure we're not much faster than moose - at least on moarvm? 18:51
maybe something's not being rooted when we do something or other ... 18:52
yeah, that's unhelpful
19:01 spider-mario joined, rindolf left, Possum joined, rindolf joined
smls timotimo: Nah I'm not sure about any performance claims as I haven't done any benchmarks myself; it's just things I heard from jnthn's and lizmat's talks etc. 19:04
19:07 zakharyas left
colomon moritz: two more comments (much smaller, thankfully) since smls's. ;) 19:08
timotimo we don't really have benchmarks that directly contrast the performance of classes in p5 moose vs p6
smls in any case, the start-up time of perl6 scripts tend to be similar to what I'm used to from perl+moose scripts, so at least in that area the comparison seems valid... :) 19:10
19:11 rindolf left
smls Of course, few things can compete with the 0.00 seconds that perl takes to parse, compile & start a medium-sized Perl 5 script that doesn't use any modules... :P 19:11
dalek : 146dee4 | (David Farrell)++ | misc/perl6advent-2014/schedule:
added new proposed advent article
19:13
: 574c5fe | colomon++ | misc/perl6advent-2014/schedule:
Merge pull request #7 from sillymoose/master

added new proposed advent article
rl6-roast-data: ed98fd7 | coke++ | / (4 files):
today (automated commit)
19:19
perltricks colomon: thx :) 19:20
colomon perltricks++
19:24 Mso150_s joined, Mso150 left, Alula left 19:25 PZt joined
colomon just tried out parallel mandelbrot 4 times in a row, getting one success and three different failures. :( 19:25
19:26 Alula joined
moritz perltricks: /msg me your email address to get an invitation to the p6advent blog 19:35
tony-o jnthn: ping 19:40
mvuets cosimo: hi! i had troubles with LWP::Simple tests yesterday. kind folks helped to overcome problems, and i tried to adjust the test that failed. just sent you a pr, could you please take a look and say if that makes sense 19:41
19:42 romel joined 19:43 perturbation joined 19:50 anaeem1_ left 19:51 anaeem1_ joined 19:55 anaeem1_ left 19:56 denis_boyun_ joined 20:04 mvuets left 20:07 telex left 20:08 telex joined
[Coke] .tell mvuets using .tell is more likely to get heard later. 20:12
yoleaux [Coke]: I'll pass your message to mvuets.
20:13 ab5tract joined 20:15 kurahaupo left 20:17 ab5tract left 20:21 darutoko left 20:27 FROGGS__ left 20:33 guru joined, guru is now known as Guest13254 20:34 Guest13254 is now known as ajr_, ajr_ left 20:35 kurahaupo joined, blackbolt joined 20:39 FROGGS__ joined 20:42 Mso150_s left 20:43 Mso150_s joined
jnthn tony-o: pong 20:43
btyler nativecall question: for making bindings to APIs that heavily use structs to define options for a given function, is there any way around defining said structs in p6? 20:45
arnsholt Nope
btyler libgit2 does a lot of that, and many "foo_options" structs embed other structs, which also need defining...all the way down
arnsholt Yeah, that sucks 20:46
Also, are the embedded structs pointers, or just embedded?
('Cause we don't support embedded structs yet)
20:47 vendethiel- joined
btyler all the ones I've bumped into so far are pointers 20:47
arnsholt Good
btyler so I've been stubbing with OpaquePointer until suddenly something tries to read from the pointed-to struct and ...boom :)
arnsholt But yeah, the struct defining sucks
20:48 vendethiel left
btyler I had fleeting thoughts of something to automate it, since it's a fairly mechanical process, but I suspect implementing something on that level of magic would take me a long, long time 20:48
and be quite fragile in any case
jnthn Well, an 80% solution might still be a time-saver 20:49
arnsholt Definitely
I think it should be reasonably doable with something like clangs parser hooks 20:50
jnthn Something that spits out Perl 6 code with XXX comments in places it thonks you'd really better review
*thinks
btyler sounds fun -and- educational
arnsholt I even have a name for a project like that, but not the tuits to implement it 20:51
20:52 smls left
btyler on the topic of nativecall, I defined helpers '&-in-c' and '*-in-c' that feel like they might belong at a lower level than my binding: github.com/kanatohodets/p6-git2/bl...it2.pm#L10 20:53
20:54 raiph left
btyler but I don't know if there's a way to do it that doesn't surprise people with messages about CArrays that they're not expecting 20:55
or needing to define native subs that take CArray[thing]
leedo_ this is a stupid question, but whats the difference between a prefix operator and plain sub in p6?
masak leedo_: not a stupid question. 20:56
psch m: (1, 2, 3).map(!) # boom 20:57
tony-o jnthn: i've tracked mine down to here but i'm not sure what i could do differently to avoid the problem: github.com/tony-o/perl6-http-serve...nc.pm6#L48
camelia rakudo-moar 91d899: OUTPUT«===SORRY!=== Error while compiling /tmp/UrZmrQ5J5M␤Negation metaoperator not followed by valid infix␤at /tmp/UrZmrQ5J5M:1␤------> (1, 2, 3).map(!⏏) # boom␤»
masak leedo_: hm, let's see. a prefix operator can be symbolic, like `+`
leedo_: also, what psch said.
jnthn Prefix operators can have different precedences also
psch although i supposed ! is different than usual prefixes
? might have been a better example
masak psch: `!` isn' that different. 20:58
jnthn "foo 1, 2, 3" always means "foo(1, 2, 3)", but depending on precedence the foo might just bind to the 1
tony-o jnthn: without that $*SCHEDULER.cue bit, i end up with no segfaults or thread errors on moar
leedo_ ok, that all makes sense :) just seemed unclear why &-in-c woudl be a prefix and not a sub
masak leedo_: it's really a matter of taste and intended usage. 20:59
psch masak: i was mostly going on the error message, it states "metaoperator" and we can do != but not ?=
btyler (disclaimer about &-in-c: I'm learning p6 as I go, no best practices to be found in my code)
jnthn tony-o: I'm not at all convinced your code is to blame. 21:00
moritz btyler: they looko fine, but in the long run I'd really prefer a dedicated pointer type
*looks
btyler moritz: that was my general feeling, seemed like something that belongs outside a given binding 21:01
jnthn tony-o: It looks like some kind of race condition...somewhere. but I don't think it's in what you've written.
21:03 Mso150_s left
tony-o jnthn: yea i definitely have a race condition but i'd think that the connection close would just fail and move on 21:04
21:05 Mso150_s joined
jnthn tony-o: I mean, something internal to Rakudo has the race. 21:05
tony-o jnthn: ah
jnthn I've seen the same thing happen in a completely different code-base using the same primitives you are, but though doing a totally different set of things. 21:06
tony-o think it's a problem with channels? 21:07
jnthn I doubt it; I'm not using those. Just promises and supplies. 21:08
The thing about wrong outer smells of a race in taking closure clones...
21:10 kaare_ left 21:23 PZt left
colomon jnthn: for what it's worth, the parallel mandelbrot code, which is just very straightforward promises, I think, is not stable on moar or jvm. though it's more stable-ish on jvm, it seems. 21:32
that's with rakudos from earlier this afternoon.
21:36 pmurias joined
dalek p-js: fe08dda | (Pawel Murias)++ | src/vm/js/QAST/Compiler.nqp:
Add a .log method for logging stuff to nqpjs.log when the env var NQPJS_LOG is set.
21:36
p-js: 224785c | (Pawel Murias)++ | / (12 files):
Add a mini-nqpmo (a proper nqpmo will be generated properly in future commits).

The following code works now: my class NQPMu {
  }
my class Foo {
   method foo() {
   say("hi");
   }
  }
Foo.foo;
arnsholt pmurias++ # nqpjs 21:38
21:38 ohcamacj-2 joined 21:42 [Tux] left 21:48 ron1230 joined
ron1230 Have been to the department of energy (doe), nuclear regulatory commission (nrc), and national institute of health (nih) web sites and hate what I see. 21:49
I am hungry, tired, have no clean clothing and wish you all would go to hell. I called the Democratic Peoples Republic of Korea. 21:50
moritz hugs ron1230 21:51
ron1230 Call to DPRK was this afternoon. Am ronaldxs on github. bye. 21:52
moritz: do you want me to hug you back? You are wasting my time. You all look like assholes. 21:54
btyler sounds like someone left a screen unlocked
moritz ron1230: no; it merely sounds like you need a hug 21:55
ron1230 moritz: no you are lying
Timbus lol
21:56 spider-mario left
ron1230 You are accessories to murder and other crimes by the way and may be appropriate for people to come after you for that. 21:57
Anyone have a sensible comment.
?
masak ron1230: you look like you need several hugs.
masak hugs ron1230 21:58
moritz hugs ron1230 again
TimToady hugs ron1230
ron1230 masak: drop dead
21:58 [Tux] joined
Timbus no comment 21:58
masak hugs ron1230
ron1230: welcome to #perl6.
ron1230: ooh, I see you contributed to November back in 2011. 22:01
ron1230++
22:01 ab5tract joined
ron1230 (doe) www.energy.gov/science-innovation/national-labs (nrc) www.nrc.gov/ (nih) www.nih.gov/health/NIHandSleeplessinAmerica/ 22:02
big trouble
bye
masak bye, ron1230. 22:03
22:03 ron1230 left
moritz points out that NIH is also Not Invented Here 22:03
masak the "Not Invented Here" acronym always gives me cognitive dissonance.
probably because it's a slogan used to *avoid* certain products. 22:04
jnthn Would you like it better if you'd invented it? :P
moritz masak: I experience NIH as a disguised lack of domain knowledge
masak jnthn: if I had invented it, "NIH" would mean that a company or group *loved* things they didn't invent themselves! 22:05
moritz I think "this library is total overkill for solving $X, I only need a tiny subset that I can implement quicker than the new dependency is worth"
and then I start implementing stuff 22:06
22:06 ab5tract left
moritz and realize why the library is a lot bigger than what I'm building: because the domain actually needs it 22:06
anyway, time to sleep 22:07
jnthn 'night, moritz
masak sleep well, moritz++ 22:08
22:14 ab5tract joined 22:18 ab5tract left 22:20 ron1230 joined
ron1230 Hating to waste all of our time and trying to move on. DOE seems to have changed a bit www.energy.gov/articles/national-la...-forward-0 (look at all slides) and www.energy.gov/articles/womeninstem...dvancement 22:23
22:24 Mso150_s left, blackbolt left
psch progress \o/ 22:38
i'm now installing a method that lists the number of candidates existing in the foreign class for the shortname 22:39
now i just gotta figure out how to handle the dispatch logic instead of writing that method
i guess that's where i can borrow a bit from IndyBootStrap.methcall_noa and IndyBootStrap.methcallResolve_noa, although variable arity is something i'm not yet sure how to handle 22:43
jnthn psch: If just having the args as an array is helpful, I think there's an asVarArgsCollector or somehting in the MethodHandles combinator set 22:45
psch jnthn: noted, i'll look at that. i'm still unsure how arguments from p6land end up in jvm-land 22:53
the IndyBootStrap.*resolve methods get an Object slurpy, which gets handed to ArgsExpectation.invokeByExpectation, but that's already after the arguments got gathered into the Object[] i gather 22:54
22:54 molaf left
jnthn psch: Yes, but it then updates the callsite with something more specialized for the follow-up invocatin. 22:54
*invocation
iirc
A lot of the things in IndyBootstrap work like that. 22:55
psch ooh, methcall_noa is mapped to QAST::Op(:op<callmethod>) 22:56
that should maybe have been somewhat obvious 22:57
well, i'll look closer tomorrow. i also have quite a few notes i'll expand to an advent post, which might actually focus on "java multis now prettier" if i have enough tuits to finish the implementation and the post soonish 22:59
with that, g'night o/ 23:00
23:00 psch left 23:01 gfldex left 23:04 Isp-sec left 23:09 KCL_ left 23:24 FROGGS__ left 23:37 Ugator left 23:46 araujo left 23:53 rmgk_ joined, rmgk left, rmgk_ is now known as rmgk
ron1230 Hope all ate well at Thanksgiving and are in good health. Good night. 23:59
23:59 ron1230 left