»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 diakopter on 14 April 2013.
adu does anyone here use clang? 00:26
Tene I have a few times. 00:27
TimToady I think diakopter++ has used it. 00:28
sorear i've used it a few times as a black box 01:06
c code goes in, executables come out 01:07
japhb
.oO( C code goes in, sausage comes out )
 02:11
colomon TimToady! \o/ 02:17
adu Tene: I'm getting an undefined symbol, and ##C++ is saying it's off topic 03:08
gist.github.com/andydude/5456709 03:09
diakopter i.imgur.com/vXMuRqB.png # YAPC Austin registrations so far 05:37
tadzik for some reason I thought there's 120,000 06:07
diakopter o) 06:08
nwc10 diakopter: to two decimal places :-) 07:09
moritz \o 07:10
sorear ack, I need to get that dealt with pdq 07:18
labster Well, I spent the past few hours finding out what doesn't like to compile in core.setting. regex { }, BEGIN { }, and method handles trait all made my code die when I moved it into the setting. 07:20
moritz we should document that somewhere 07:37
diakopter nwc10: yes. :) Excel doesn't recognize they can be collapsed to integers... o_O 07:41
labster Well, I'm hoping that I'm wrong, at least for the rx// and handles cases, because code readability will suffer a lot. 07:55
moritz I have run into the same problem with 'handles' 07:57
masak good antenoon, #perl6 08:54
arnsholt 'lo masak 08:55
tadzik hello hello 08:57
jnthn morning, #perl6 09:01
hoelzro o/ jnthn 09:02
masak jnthn! \o/ antenoon!
hoelzro jnthn: are you busy right now? I found a "fun" behavior with NativeCall and thought I'd have you take a look 09:05
jnthn
.oO( what's it got against the noon? )
hoelzro: I can take a quick look but I don't want to get bogged down with NativeCall stuff today.
hoelzro ok
I'll just give you a link and you can take a look whenever you like 09:06
it's not urgent
jnthn ok 09:07
hoelzro jnthn: hoelz.ro/files/perl6-posix.tar.gz 09:11
if you run test.p6 in that tarball, there's a really odd error about unboxing and such
jnthn hoelzro: Will this work on Windows? 09:12
jnthn is guessing not
hoelzro also, when I started to remove functions from the module, I started to see segfaults or garbage values
I haven't tried
arnsholt I'll have a look
jnthn arnsholt: Ah, cool. :) 09:13
arnsholt I need to recompile NQP/Rakudo/Zavolaj so it'll be a while before I can actually run it thought =)
jnthn arnsholt: Well, I have a funny tangle of branches and stuff built here too, so... :)
arnsholt Heh
arnsholt hoelzro: That's odd 09:40
hoelzro yeah, I know!
my NQP fu isn't good enough to figure it out 09:41
I'm going to spend more time on Perl 6 stuff after I'm done with my coursera stuff
arnsholt jnthn: How much of the sixmodelisation replacement of Parrot types with NQP types has happened so far?
As a first guess, this might be the dyncall bits returning a Parrot STRING* where that isn't allowed any more 09:42
jnthn arnsholt: Very little.
arnsholt: str is still a Parrot STRING
arnsholt Hmm
jnthn arnsholt: Is it something to do with string memory getting corrupted? 09:43
arnsholt Nope
"This type cannot unbox to a native string"
jnthn Hmm 09:44
Do you know where it's coming from?
I mean, Perl 6 Str objects have to be unboxed like this
arnsholt The line info in the backtrace points at: nqp::getattr(nqp::p6decont($args), Capture, '$!list')
jnthn It's possible a non-string is being passed somewhere a Str is expected?
Huh! 09:45
arnsholt Which is part of a larger call: nqp::nativecall(nqp::p6decont(map_return_type($r.returns)), self, nqp::getattr(nqp::p6decont($args), Capture, '$!list'))
jnthn Oh...hmm.
So it could be on the inside of the call
arnsholt Although the actual error might be happening inside the C code
jnthn Yeah, that's my guess. What is being called? The argument really is a Str, yes?
arnsholt The sub decl is "our sub ttyname(int $fd) returns Str is native", invoked as "ttyname(0)" 09:47
jnthn Oh...very odd.
arnsholt Yeah. Grepping for "This type cannot unbox" in NQP, the only matching file is P6opaque.c
jnthn I'd have thought "unbox" would refer to calling, not handling result...and hat's the only Str in there. 09:48
arnsholt Yeah
dalek p: 8c9a25c | jnthn++ | src/vm/jvm/QAST/Compiler.nqp:
Map HLL interop ops (though not implemented yet).
 09:54
p: 906e635 | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/ (2 files):
Implement nqp::settypehll[role]? on JVM.
p: b18c101 | jnthn++ | src/vm/jvm/ (2 files):
Implement nqp::isbig_I.
p: f28e949 | jnthn++ | src/vm/jvm/ (2 files):
Implement nqp::getpayload.
kudo/jvm-support: b51e852 | jnthn++ | tools/build/Makefile-JVM.in:
Start building Perl6::World in JVM build.

It'll need a few more changes and depends on a couple of missing ops, but it does now cross-compile.		 09:56
cognominal I am implementing permutations using gather/take but I am getting something wrong. get_attr_str() not implemented in class 'Coroutine' gist.github.com/cognominal/5458754 10:05
btw, the rest of the error message is useless : current instr.: 'print_exception' pc 111585 10:06
jnthn That looks like an error reporting fail rather than the actual error.
arnsholt Yeah
jnthn Try running with --ll-exception to see if it gives anything more helpful
arnsholt Looks like the exception printer expects all callables to have some property, which Coroutines don't have 10:07
jnthn Doing return unless @l; inside a gather/take looks highly suspect though.
cognominal jnthn: I thought so to but what is the alternative to stop the recursion
jnthn cognominal: Put the return outside of the gather block? 10:08
cognominal ok, no error now but no output either but I can go from there 10:09
afk& 10:14
kresike hello all you happy perl6 people 10:40
dalek p: adfa9c8 | jnthn++ | src/vm/parrot/ (2 files):
Add nqp::isint/nqp::isnum/nqp::isstr for Parrot.
 10:52
p: 9c79d1f | jnthn++ | src/vm/jvm/ (2 files):
Add nqp::isint/nqp::isnum/nqp::isstr for JVM.
dalek kudo/jvm-support: 26dc105 | jnthn++ | / (2 files):
Eliminate final pir:: in Perl6::Pod.
 10:56
kudo/jvm-support: 70c0109 | jnthn++ | / (3 files):
Eliminate final pir:: in optimizer.
kudo/jvm-support: 97f7855 | jnthn++ | tools/build/Makefile-JVM.in:
Add Perl6::Pod to the JVM build.
colomon \o/ 11:03
FROGGS jnthn: btw, do you "like" the way the options are passed to load_module? gist.github.com/FROGGS/59b7f7a890381005eda7 11:12
jnthn FROGGS: Passing them as names seems reasonable. 11:15
*nameds 11:16
FROGGS thanks
jnthn :$name?, :$file?, :$from?, :$auth?, :$ver?
Rendundant use of ?
Named args are optional by default
FROGGS ahh, yeah, now that you say it *g*
<jnthn> Rendundant use of ? <-- you are already speaking like perl 11:17
tadzik :D 11:19
Useless use of ? in named context
FROGGS I would like to pass --ll-exceptions sometimes to jnthn though :P
moritz named params are optional by default 11:20
FROGGS ohh, I got it already, thanks :P 11:21
dalek p: 12b0eb8 | jnthn++ | tools/build/Makefile-JVM.in:
Start building/installing P5Regex library on JVM.
 12:16
[Coke] jnthn++ 12:28
[Coke] jnthn: No registered operation handler for 'isstr' (building rakudo-jvm) 12:32
jnthn: No registered operation handler for 'hllizefor' 12:33
jnthn [Coke]: I didn't put the stuff in to do NQP version checks on the JVM side yet; it's probably just a not-new-enough nqp-jvm (I added that op this morning)
[Coke] ah, I forgot about that arms race, roger. 12:34
cognominal stuck anew with the same message; no return in the gather though. gist.github.com/cognominal/5458754 12:41
moritz curious, I get "get_attr_str() not implemented in class 'Coroutine'" 12:46
with cognominal's code 12:47
nwc10 jnthn: your NQP changes have not yet broken the Rakudo build for Parrot. Please try harder :-) 12:47
cognominal can we use gather in a recursive setting?
[Coke] jnthn: upped nqp, tried to build nqp-jvm, got: Unknown compilation target 'classfile' 12:48
cognominal r: sub a($n) { gather { take $n; take $n-1 if $n != 0 }}; say a(2); 12:51
camelia rakudo 37c995: OUTPUT«2 1␤»
cognominal This simpler recursion case works 12:52
masak moritz: I'm getting the same. 12:53
minimizing. 12:54
nwc10 [Coke]: I just tried and I get that too
moritz cognominal: iirc [Coke] has hit another limitation of gather/take recentely; if you reuse the same gather/take block, the first iterator has to be exhausted before the second one is started 12:54
cognominal arf 12:56
jnthn [Coke], nwc10: cd nqp ; perl Configure.pl --prefix=..\install-jvm ; make install ; cd .. ; perl ConfigureJVM.pl ; make
Argh
[Coke] Aye, used to work in b.
jnthn [Coke], nwc10: cd nqp ; perl ConfigureJVM.pl --prefix=..\install-jvm ; make install ; cd .. ; perl ConfigureJVM.pl ; make
nwc10 jnthn: this is nqp failing to build itself 12:57
cognominal afk for ~45 minutes
thx moritz
jnthn nwc10: Oh.
I need to see how it fails in a bit more detail, then. 12:58
Since it works here :)
dalek p: d228c92 | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/Ops.java:
Error reporting tweak.
p: 2ac90f3 | jnthn++ | src/vm/jvm/ (2 files):
Implement neg_I and abs_I for JVM.
p: 115b38c | jnthn++ | src/vm/jvm/ (2 files):
Implement nqp::radix_I.
nwc10 pasta.test-smoke.org/489
[Coke] One moment.
jnthn It sounds like it's somehow invoking a Parrot NQP, not a JVM one. 12:59
[Coke] what nwc10 said.
ah. "nqp --target=c"
pulls from nqp in path.
you're on windows, so that that probably JFW there. 13:00
jnthn Ahh
masak r: sub p(*@l) { gather { if @l == 1 { take @l } else { @l.shift; for p(@l) -> @p {} } } }; p 1..4 13:01
camelia rakudo 37c995: OUTPUT«Nominal type check failed for parameter '@p'; expected Positional but got Int instead␤current instr.: '' pc 470 ((file unknown):228) (/tmp/rbAFU_hQSA:1)␤called from Sub '' pc 279978 (src/gen/CORE.setting.pir:122782) (src/gen/CORE.setting:5741)␤called from Sub 'reif…
masak g'ah.
I got this error locally during some of the experimenting.
oh, no, sorry. it's consistent.
hold on.
nwc10 ./nqp --target=classfile --output=NQPP5QRegex.class src/stage2/gen/NQPP5QRegex.nqp
works
moritz for p(@l) -> @p
masak r: sub p(*@l) { gather { if @l == 1 { take @l } else { @l.shift; for p(@l) -> @p {} } } }; say p 1..4
camelia rakudo 37c995: OUTPUT«get_attr_str() not implemented in class 'Coroutine'␤current instr.: 'print_exception' pc 111628 (src/gen/CORE.setting.pir:50166) (src/gen/CORE.setting:9743)␤called from Sub '' pc 481 ((file unknown):229) (/tmp/Ey81jJgi4k:1)␤called from Sub '' pc 279978 (src/gen/COR… 13:02
masak there we go.
masak submits rakudobug
moritz: yes, I know.
[Coke] so, looks like the def for RUNNER is wrong.
moritz 'for' flattens, so if p(@l) returns a list of arrays of integers, that's what you get
masak moritz: the internal error seems to happen before that, though.
jnthn r: sub x() { gather { die 'oops' } }; x
camelia rakudo 37c995: OUTPUT«oops␤current instr.: 'throw' pc 358206 (src/gen/CORE.setting.pir:156326) (src/gen/CORE.setting:9557)␤called from Sub 'die' pc 36476 (src/gen/CORE.setting.pir:14282) (src/gen/CORE.setting:578)␤called from Sub 'die' pc 36374 (src/gen/CORE.setting.pir:14222) (src/gen/…
jnthn r: sub x() { gather { for 1..10 { take 1; die 'oops' } } }; x 13:03
camelia rakudo 37c995: OUTPUT«oops␤current instr.: 'throw' pc 358206 (src/gen/CORE.setting.pir:156326) (src/gen/CORE.setting:9557)␤called from Sub 'die' pc 36476 (src/gen/CORE.setting.pir:14282) (src/gen/CORE.setting:578)␤called from Sub 'die' pc 36374 (src/gen/CORE.setting.pir:14222) (src/gen/…
jnthn Hm.
nwc10 is camelia still doing (historical) nqp-jvm, but not (yet) real nqp (on the JVM) ? 13:04
moritz nwc10: correct
timotimo std: [.perl] 1, 2, 3, 4 13:05
camelia std 86b102f: OUTPUT«===SORRY!===␤Two terms in a row (preceding is not a valid reduce operator) at /tmp/Q9GdpfQt5T line 1:␤------> [.perl] ⏏1, 2, 3, 4␤ expecting any of:␤ infix or meta-infix␤ infixed function␤ statement modifier loop␤Parse
..failed␤F…
[Coke] nwc10: look for "nqp.bat" in ConfigureJVM.pl
nwc10 jnthn: got test failures, but they are out of date as teh goalposts moved: pasta.test-smoke.org/490
timotimo er, that also doesn't make sense 13:06
things you come up with when you're kind of asleep
dalek p: 14266b9 | jnthn++ | tools/build/Makefile-JVM.in:
Try to unbust Linux/Unix build.

  nwc10++ and [Coke]++ for reporting/diagnosis.
jnthn nwc10: Yeah, I'm currently focusing on the Rakudo porting rather than clearing those up. But I'll get them eventually. :) 13:08
nwc10: Already incidentally fixed something there today anyway.
nwc10 jnthn++ # yep, "works" on "my" machine again 13:14
fails those 5 tests. ie - build works once more.
jnthn Time for a break/walk; bbiab 13:16
[Coke] jnthn++
masak timotimo: you can't put method names inside reduction metaops like that. 13:17
oops, talking to the backlog.
[Coke] with that, rakudo-jvm now builds 4 jvm files.
masak std: (1, 2, 3, 4)>>.perl 13:18
camelia std 86b102f: OUTPUT«ok 00:00 43m␤»
masak timotimo: maybe you meant this?
timotimo i was kind of derping what a reduction does, but i was also at that moment napping and decided to type what came to my mind into the calculating machine 13:20
PerlJam
.oO( What would an infix:<perl> op do? )
 13:21
moritz today's mini challenge: define a custom meta operator that can take an identifier, and applies the string as a method to all args 13:22
[Mperl] 1, 2, 3 would be the same as (1, 2, 3)>>.perl 13:23
moritz is not quite serios
*serious
PerlJam moritz: that would be sort of like a meta-map (but the reduction doesn't make enough sense there) 13:25
masak moritz++ # mini-challenge
my (pessimistic) prediction: defining metaops will not work in Rakudo. :/ 13:26
maybe in Niecza.
colomon I'd be surprised if it worked in Niecza.
masak timotimo: have you seen the great quote by Charles Babbage? :)
moritz also note that it's not quite a meta op
because it doesn't take an op as argument, but an identifier
colomon unless you mean in the grand sense that it's reasonable to implement code that allows you to define metaops. 13:27
masak timotimo: 'On two occasions I have been asked, — "Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?" [...] I am not able rightly to apprehend the kind of confusion of ideas that could provoke such a question.' en.wikiquote.org/wiki/Charles_Babbage
moritz should have termed it a "macro-challenge" :-)
PerlJam masak: that last bit of the quote "I am not rightly to apprehend ..." comes to mind *all the time* on IRC ;) 13:28
timotimo masak: i have
masak timotimo: it's like the 19th-century version of 'Two different users asked me: "Will the program get the output right if I get the input wrong?", and I'm like "WTF!"'
timotimo the output may be right even if the input is wrong. but it'll be right for the wrong inputs. or if the inputs are truly bogus, they'll be wrong or - hopefully! - an exception or some defined error state at least 13:29
hoelzro I love that quote
colomon PerlJam: on the internet in general! 13:30
arnsholt hoelzro: Could you do me a favour and add your problem to the Zavolaj issues? 13:38
I'm a bit too busy at work ATM to deal with it right now, sadly =/ 13:39
timotimo what's a good way to let a user supply a filter to a method that will be passed to .grep? i tried method frob($crit) { $something.something.grep({$_<frob> ~~ $crit}).something.something }, but that didn't seem to work properly. is that because of a bogus $crit or something else?
PerlJam timotimo: I don't even understand what you're asking for. 13:43
hoelzro arnsholt: sure! 13:44
colomon timotimo: the natural way is just frob($crit) { $something.something.grep($crit).something.something } unless you want to not smartmatch $crit against the entire object passed to grep
FROGGS timotimo: if grep accepts a block, why dont you just pass a block?
hoelzro I could even post more tarballs that illustrate different problems later tonight
timotimo because the most common usage would be to pass in just a number 13:46
or a list
but sometimes you want "everything but the number 1"
FROGGS I think you can do that only by passing a pointy block 13:47
FROGGS and maybe wrap the param if it isnt a block already 13:47
colomon r: say (1..20).grep(&is-prime) 13:48
camelia rakudo 37c995: OUTPUT«2 3 5 7 11 13 17 19␤»
colomon r: say (1..20).grep(not &is-prime)
camelia rakudo 37c995: OUTPUT«␤»
colomon r: say (1..20).grep({not $_.is-prime})
camelia rakudo 37c995: OUTPUT«1 4 6 8 9 10 12 14 15 16 18 20␤»
timotimo is there a swartzian transform for grep yet? :)
PerlJam timotimo: so you want multis; one for the common case and another for the general case?
colomon r: say (1..20).grep({!$_.is-prime})
camelia rakudo 37c995: OUTPUT«1 4 6 8 9 10 12 14 15 16 18 20␤»
colomon I really don't think you'll easily do better than that
timotimo that sounds kind of like what i want
colomon r: sub test-numbers($crit) { (1..20).grep($crit); }; say test-numbers(&is-prime) 13:49
camelia rakudo 37c995: OUTPUT«2 3 5 7 11 13 17 19␤»
colomon r: sub test-numbers($crit) { (1..20).grep($crit); }; say test-numbers({ !$_.is-prime })
camelia rakudo 37c995: OUTPUT«1 4 6 8 9 10 12 14 15 16 18 20␤»
pmichaud good morning, #perl6 13:51
moritz good am, pm
masak rn: multi negate(&F) { return -> $x { !&F($x) } }; say (1..20).grep(negate &is-prime)
isBEKaml pmichaud: good morning
camelia rakudo 37c995, niecza v24-37-gf9c8fc2: OUTPUT«1 4 6 8 9 10 12 14 15 16 18 20␤»
masak HOP FTW, people. 13:52
colomon \o
masak pmichaud! \o/
moritz note that this HOP is even built into Perl 6
colomon HOP++
moritz r: say grep !*.is-prime, 1..20
camelia rakudo 37c995: OUTPUT«1 4 6 8 9 10 12 14 15 16 18 20␤»
masak ooh
colomon moritz++
jnthn back
colomon feels stupid for not thinking of that.
isBEKaml likes HOP in what he read so far. :) 13:53
but too many books, the list just keeps piling up. :|
[Coke] ETOOLITTLETIME 13:59
isBEKaml exactly 14:00
PerlJam sleep less ;)
isBEKaml nah, been there - only to get panic attacks in the middle of the day. :/
moritz you'll automatically sleep less as you become older 14:06
PerlJam I don't think people actually sleep less as they get older. They only sleep less *at night* 14:08
moritz might be
masak like cats. 14:10
spider-mario speaking of cats: www.youtube.com/watch?v=BkgOzxKPJXQ 14:12
:>
[Coke] .meow 14:33
yoleaux DeadlineExceededError: Deadline exceeded while waiting for HTTP response from URL: edgecats.net/random
isBEKaml .cats 14:41
FROGGS .meow 14:57
yoleaux moar.edgecats.net/cats/tumblr_lk64a...o1_500.gif
FROGGS \o/
FROGGS boring 14:58
.meow
yoleaux moar.edgecats.net/cats/z1fLa.gif
FROGGS ohh, this one is good :oD 14:58
dalek p: 1fd7f29 | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/Ops.java:
Correction to freshcoderef.
 14:59
p: 1c43791 | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/sixmodel/SerializationWriter.java:
Improve error reporting.
p: e8b076b | jnthn++ | src/NQP/Actions.nqp:
Fix compiling our-scoped subs.

Not sure why we ever got away without this on Parrot.
dalek kudo/jvm-support: a996710 | jnthn++ | / (3 files):
Perl6::Ops is backend-specific.
 15:01
kudo/jvm-support: 32f7240 | jnthn++ | / (2 files):
Add a partly translated Perl6::Ops for JVM.

We'll have some Perl 6 specific ops in JVM land. This translates the mappings, though of course the ops all still need implementing. This just means we can compile things that refer to them.
kudo/jvm-support: 8c05ec6 | jnthn++ | src/Perl6/Actions.nqp:
nqp::concat => nqp::concat_s

They are just synonyms by now.
kudo/jvm-support: 72ab250 | jnthn++ | / (2 files):
Add Perl6::Actions to the JVM build.
FROGGS "Not sure why we ever got away without this on Parrot." <-- I really do like these kind of weirdness 15:02
jnthn I don't 'cus it tends to tell me I've missed something somewhere... 15:03
moritz www.perlmonks.org/?node=1030688 # nice algorithm question; I'd appreciate an answer with Perl 6 code :-) 15:07
(though I'm not the OP)
kresike bye folks 15:12
[Coke] ~~ 15:12
tadzik this "not sure why we got away with this" is sometimes paired with "oh look, for reasons unknown this weird bug is now fixed" :P 15:19
FROGGS that is exactly what I was thinking
isBEKaml hey, tadzik!
tadzik \o/ 15:30
hey hey
isBEKaml tadzik: how goes GUInification? One GUILib to rule them all? :)
tadzik: Still picking leaves of the great forest? :P 15:32
s/of/off/
pmichaud nqp's ohloh page was finally updated. www.ohloh.net/p/nqp 15:34
jnthn .. 15:37
is mostly written in C
with a well-commented source code
:D
pmichaud jnthn: you should claim your projects. www.ohloh.net/people?q=jnthn
jnthn hm, do I have an account there...
oh, seems so 15:38
tadzik isBEKaml: haha. No, I don't remember thinking about it at all
timotimo of course a site such as ohloh would have trouble classifying such a niche language as nqp or even as perl6 15:40
jnthn "4510 total commits to 6 projects" :) 15:42
arnsholt hoelzro: Thanks for the bug report! More examples would be great as well. Might be useful to figure out WTH is going on 15:43
moritz claimed a few of the projects on ohloh 15:48
isBEKaml tadzik: All in good time. :D 15:53
dalek p: 7853513 | jnthn++ | src/vm/jvm/ (2 files):
Greatly shrink lexical name list code size.

Can further optimize this later as many times the names are not needed at all in execution.		 16:24
dalek kudo/jvm-support: a79969d | jnthn++ | tools/build/Makefile-JVM.in:
Add Perl6::Grammar to JVM build.
 16:25
kudo/jvm-support: 491bf5e | jnthn++ | tools/build/Makefile-JVM.in:
Add Perl6::Optimizer to the JVM build.
kudo/jvm-support: 2877033 | jnthn++ | tools/build/Makefile-JVM.in:
Add Perl6::Compiler to JVM build.
[Coke] so I peeked at olhol for the first time in ages. doesn't seem to be a way to say "all the logins by coke at github are me", instead I have to go project by project and claim them. 16:28
nwc10 that seems a bit, well, suboptimal on their part
[Coke] claims 2 or 3 and give sup. 16:31
dalek kudo/jvm-support: 982d91e | jnthn++ | / (2 files):
Add compiler frontend program to the JVM build.
 16:40
kudo/jvm-support: 911d56b | jnthn++ | .gitignore:
Update .gitignore.
pmichaud p6: my $a = 1; $a.++; say $a; 17:16
camelia niecza v24-37-gf9c8fc2: OUTPUT«2␤»
..rakudo 37c995: OUTPUT«===SORRY!===␤CHECK FAILED:␤Undefined routine '&postfix:<.>' called (lines 1, 1)␤»
pmichaud rakudobug (maybe already reported)
moritz long-known NYI feature 17:20
at least long known to those that look at roast fudges :-) 17:21
jnthn bbl; dinner 17:23
diakopter japhb_: welcome. 17:24
_sri the rust folks are redesigning their I/O API, maybe of interest to some here mail.mozilla.org/pipermail/rust-de...03746.html 17:25
cognominal moritz: where are roast results dipslayed?
japhb_ o/
cognominal … if they are
moritz on your terminal, when you run 'make spectest' 17:28
diakopter _sri: thanks for the link. If I were imagining undertaking such a thing for a language-dedicated VM (well, "runtime") such as Rust's, I'd steal libuv from node, including exposing the event loops from the VM, port node's JS api directly to a language such as NQP, then implement the currently-standard nqp:: IO ops on top of that "node" API (for short-term compatibility), in addition to adding new ones for async, then implement(/redesign/adapt) ... 17:35
... the Perl 6 IO API on top of that, then gradually eliminate the middle layers over the next months/years...
_sri diakopter: i believe that's exactly where they started
diakopter hah.
_sri but now they are moving in a different direction to take advantage of threads 17:36
diakopter do you know why they couldn't use threads with libuv? 17:37
_sri it's quite interesting, apparently they have something very similar to parrot tasks (green threads mapped to a number of real threads)
it could still use libuv internally, don't know details yet i'm afraid 17:39
the goal seems to be a synchronous looking API that scales really well with dirt cheap green threads (through non-blocking I/O) github.com/brson/rust/blob/io/src/...mod.rs#L11 17:41
very very close to how parrot did it 17:42
diakopter eh 17:45
except parrot has no native event loops or async IO...? 17:46
.. which makes it completely different...
_sri afaik it does internally 17:46
diakopter as far as you know what 17:46
_sri whiteknight explained it to me once 17:47
diakopter parrot has native event loops or async IO? no.
_sri implicit yes
i'm not sure if it ever made it into a real release
diakopter implicit?
_sri but it was supposed to use non-blocking I/O for everything 17:48
diakopter *cough*
lots of things were supposed to happen
_sri *all* I/O operations would look blocking, but behind the scenes use non-blocking I/O
diakopter okay, well, I'm not sure why we're talking about some theoretical parrot that's completely different from the current [or ever existing] one 17:49
_sri i think it might have been part of the parrot release that added threads
in the end it's just an optimization 17:50
diakopter hm, you seem to have interpreted my msg "except parrot has no native event loops or async IO...?" as me asking whether it had them.. instead of what I meant, which was "it doesn't have those... why do you think it does?"
_sri ah, ok 17:51
like i said, i'm just interpreting the things i heard from whiteknight and reini, but don't actually know the code 17:52
diakopter ok. such malarkey activates my no-spin-zone persona 17:54
dalek rl6-roast-data: b807d88 | coke++ | / (4 files):
today (automated commit)
 18:05
[Coke] niecza has been dirty for 282 days. pugs has been clean for 4 days. rakudo has been dirty for 1 day. 18:06
all in S32-temporal. 18:07
moritz will probably be fixed the timezone pull request 18:15
[Coke] is all excited that his new office digs have a window. 18:26
Hello9 Hello! :D 18:34
Wanted to share a new programming forum: thedevshed.net/forum/
feel free to join
:)
diakopter Hello9: why? 18:37
diakopter Hello9: that forum has ZERO non-admin/moderator posts. how could it possibly be helpful to anyone here? 18:40
timotimo well, he did say it's new ... 18:42
diakopter hm, true.
or she. 18:43
timotimo indeed. 18:44
diakopter anyone here work somewhere that uses google apps for business/gmail and formerly used Exchange/Outlook? 18:45
Hello9 The forum is new
launched today
thedevshed.net/forum/announcements/...tml#post68
Anyways, feel free to join :) 18:46
timotimo hm, the only part that would refer to perl6 at all would be the "perl and ruby development" subforum of "web development"
timotimo not sure i agree with that choice 18:46
both ruby and perl are useful for much more than web development
diakopter :)
Hello9 Yes
But where should it be placed then?
"Web Development" is one of it's uses 18:47
timotimo "programming"?
python's in "programming", too, even though it has quite a few famous web frameworkey bits
Hello9 :) 18:48
timotimo i wonder what the filter bubble you come from is, such that you would name one of those subforums "visual basic and .net framework", i always thought c# was the "^H'poster child' of the .net family of languages and technologies
flussence
.oO( I feel like I've tripped over a crack in the universe and landed face-first in 2002... )
 18:50
diakopter er, 1998
timotimo there was .net in 1998? 18:51
flussence diakopter: I know, but I didn't start doing PHP until '02. I quickly found out people promoting sites like this were a dime a dozen (and they almost always went nowhere :) 18:52
timotimo .o( "this reddit thing? it won't go anywhere" ) 18:53
GlitchMr rn: say '一二三四五六七八九十' ~~ m:g/<[一..十]>/ 18:54
camelia niecza v24-37-gf9c8fc2: OUTPUT«===SORRY!===␤␤Regex modifier g not yet implemented at /tmp/xQ6IjuQfXo line 1:␤------> say '一二三四五六七八九十' ~~ m:g⏏/<[一..十]>/␤␤Unhandled exception: Check failed␤␤ at /home/p6eval/niecza/boot/lib/CORE.setting line 1443 (die …
..rakudo 37c995: OUTPUT«False␤»
GlitchMr rn: say '一二三四五六七八九十'.comb:/<[一..十]>/ 18:55
camelia niecza v24-37-gf9c8fc2: OUTPUT«===SORRY!===␤␤Confused at /tmp/Nm1SRRn1Xa line 1:␤------> say '一二三四五六七八九十'.comb⏏:/<[一..十]>/␤␤Parse failed␤␤»
..rakudo 37c995: OUTPUT«===SORRY!===␤Two terms in a row␤at /tmp/3MWBeUJwKH:1␤------> say '一二三四五六七八九十'.comb⏏:/<[一..十]>/␤ expecting any of:␤ method arguments␤ postfix␤ infix stopper␤ infix or meta-infix␤ statement en…
GlitchMr rn: say '一二三四五六七八九十'.comb(/<[一..十]>/)
camelia rakudo 37c995, niecza v24-37-gf9c8fc2: OUTPUT«一 二 三 五 六 七 八 九 十␤»
GlitchMr It doesn't match 四... 18:55
jnthn .u 三四五 18:57
yoleaux No characters found
jnthn ...what? :)
jnthn r: say "三四五".ords 18:57
camelia rakudo 37c995: OUTPUT«19977 22235 20116␤»
GlitchMr .u 三 18:57
yoleaux No characters found
GlitchMr lolwhat
n: say +"三" 18:58
camelia niecza v24-37-gf9c8fc2: OUTPUT«Unhandled exception: Cannot parse number: 三␤ at /home/p6eval/niecza/lib/CORE.setting line 1504 (die @ 5) ␤ at /home/p6eval/niecza/lib/CORE.setting line 3620 (ANON @ 10) ␤ at /home/p6eval/niecza/lib/CORE.setting line 3622 (NumSyntax.str2num @ 5) ␤ at /h…
moritz 4E09 三 CJK UNIFIED IDEOGRAPH-4E09
GlitchMr That's really good name for a character.
It's three in Chinese. 18:59
It matched everything except for four.
But, I guess it makes sense
After all, I cannot do that:
n: say +"three"
camelia niecza v24-37-gf9c8fc2: OUTPUT«Unhandled exception: Cannot parse number: three␤ at /home/p6eval/niecza/lib/CORE.setting line 1504 (die @ 5) ␤ at /home/p6eval/niecza/lib/CORE.setting line 3620 (ANON @ 10) ␤ at /home/p6eval/niecza/lib/CORE.setting line 3622 (NumSyntax.str2num @ 5) ␤ a…
GlitchMr Because it isn't an actual number, just a word for it. 19:00
skids wonders if anyone ever took the ascii/shift_jis art tools to the unicode extreme. 19:06
GlitchMr skids, xahlee.info/comp/unicode_ascii_art.html 19:32
for some example 19:33
Unicode ASCII art Camelia could be interesting. 19:34
arnsholt I guess »ö« is a bit too small scale? =) 19:35
GlitchMr This is ISO-8859-1 art. 19:37
arnsholt I guess 19:38
Not if you write ö with a combining umlaut =) 19:39
GlitchMr That's cheating
I could also use fullwidth characters in ASCII art.
But IMO, that is cheating.
arnsholt Hehe, true. It'll still normalise to the ISO subset
masak writes t2 blog post 20:06
[Coke] attempts to run rakudo-jvm and gets a stack trace. progress! 20:13
jnthn++
masak lol, I blug! strangelyconsistent.org/blog/t2-rectangle-haikus 21:02
moritz: FINALLY
now please start nagging me about t3 instead :P 21:03
jnthn masak: When's t3 coming? :P 21:07
labster masak: typo: part-of-speec 21:10
masak labster: thanks, fixing. 21:10
jnthn: within two months :P
labster Well, I'm off to make my code less elegant, because regex and rx { } don't seem to work when I add them to the setting. 21:12
masak living in the undistributed future sometimes requires making one's code less elegant.
labster: just be glad it isn't still 2008 :P 21:13
jnthn What's being added to the setting, ooc? 21:14
labster File::Spec. It's needed for IO::Path to work properly. 21:15
I'm probably going to add it as IO::Spec instead.
labster .oO ( 'ooc' still means 'out of character' to me, too much roleplaying habit ) 21:16
masak "What's being added to the setting, out of character?" :P 21:17
raiph more on pinto/stratopan: perlmaven.com/videos/jeffrey-thalha...-and-pinto # relevant to panda? relevant to lizmat and other #p6p5ers re p6 & cpan?
labster Properly on non-unix OSes, I guess I should say. It's a bunch of string operations, so it shouldn't affect jvm work at all. 21:19
arnsholt labster: Proper path handling on non-Unix is a very laudable goal, IMO 21:22
dalek p: a33f00e | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/ (6 files):
Simplify invokeInternal exception handling.

Part of preparation for moving it into the generated code.		 21:24
p: 3326be2 | jnthn++ | src/vm/jvm/ (2 files):
Generate handler currently in invokeInternal.
p: fbf65e4 | jnthn++ | src/vm/jvm/ModuleLoader.nqp:
Eliminate use of a hack.
lizmat raiph: pinto is most definitely very interesting and stratopan should be watched 21:25
labster I had a prototype working as IO::Path::More in the ecosystem. I'm not sure, but I might have broken it with the changes to IO::Path in the core. Well, it should still work with Star. 21:26
lizmat but before we can think about that, we need some perl 6 internals to fix
labster Is anyone making a Star for this month? 21:27
lizmat sleep&
raiph lizmat: thanks. it's a great video too. .oO ( crowdfund jnthn )
lizmat fwiw, I don't think funding is the pb, tuits is 21:28
raiph right
lizmat really sleep&
raiph gnite
PerlJam lizmat: jnthn would have more tuits if Perl6 were his full-time job ;) 21:29
er, more tuits for perl 6 anyway
dalek p: 2a34af7 | jnthn++ | src/vm/jvm/ (3 files):
Basic usage of invokedynamic for sub calls.

Plenty left to improve yet, but this gets the basics in place.		 21:30
masak PerlJam: I'm pretty sure jnthn wouldn't make Perl 6 his full-time job, given the chance, though. 21:30
PerlJam: thing is, with our job at edument.se, that it's pretty much the perfect combination of teaching, consulting, and 20% time over for Perl 6. 21:31
PerlJam whatever keeps it fun :)
masak yeah.
this mix seems to be more or less what the combustion engine needs to work.
jnthn I could see myself wanting to shuffle the percentage more Perl 6-wards in the future, but I don't think I'd want to leave my teaching work completely behind. :)
masak oh, agree. 21:32
jnthn There's a chance I'll need to refactor various aspects of my life in a year or so anyways, though... :)
masak and if someone were to offer me a 100% Perl 6 job, I'd certainly be willing to discuss terms... :)
jnthn.phase++
jnthn The teaching is enjoyable, but also physically helpful: it gives me a form of work that doesn't involve being crouched over a keyboard. :) 21:35
I'm pretty sure if I did that all day every day it'd start hurting.
PerlJam needs something more physical to do in his job/life 21:36
Really I just need to walk in the woods every once in a while. the problem is that where I live there aren't really woods to walk in.
[Coke] masak: hey, when's t3 coming (oh wait, I don't care I didn't submit one to those. ;) 21:37
masak PerlJam: we're hiring. we have woods. 21:38
[Coke]: :P
PerlJam masak: you're also on another continent. I doubt I'll get my wife and kids to go for that :)
masak I have six t3 solutions here.
masak .oO( no, *you're* on another continent... ) :P 21:39
PerlJam :-P
labster Huh, I could use a job. It might be nice to live in a country that doesn't invade other countries every few years. 21:40
BenGoldberg I'm confused. How do you know which of you is on this continent, and which of you is on another one? :) 21:41
BenGoldberg For various definitions of "this" continent ;) 21:41
avuserow rn: say 1_000
PerlJam BenGoldberg: we use continental junctions. 21:42
camelia rakudo 37c995, niecza v24-37-gf9c8fc2: OUTPUT«1000␤»
avuserow \o/
jnthn
.oO( this was so much easier back in the good old Pangaea days... )
labster I don't think any of us live on yonder continent.
masak avuserow: hey, long time no see.
avuserow hi masak! 21:42
[Coke] git st 21:43
avuserow I've been busy with $dayjob but occasionally keep up on stuff here
avuserow hopes everyone is doing well 21:44
[Coke] er, ww.
kurahaupo labster: I do. :-) 21:46
BenGoldberg So, I've been thinking about &is-prime since last night... and wondering, how difficult would it be to change it so that instead of returning True and False, it would only return True if the number is known (absolutely) to be prime, false if the number is known to be composite, or lastly, a Num, $p, between 0 and 1 (exclusive) indicating that the number is prime with probability at least $p. 21:47
masak BenGoldberg: "how difficult" != "how desirable" 21:49
PerlJam BenGoldberg: sounds weird.
BenGoldberg: if you were going to go through that trouble, why not "True but $probability"
?
masak simple protocols have their clear advantages.
PerlJam BenGoldberg: i.e. *always* give the probability 21:50
BenGoldberg Oh, I didn't think of that :) 21:51
Ok, second is-prime related question: How desirable would it be to change the spec so that it requires that multiple calls to is-prime on any Int do multiple independent tests? 21:54
masak isn't the probability simply 1/2 ** $tries ? 21:55
BenGoldberg Normally, yes
But libtommath uses an internal lookup table
masak why would you need that back from the function? when it's you who passes $tries, I mean?
BenGoldberg Well, in the case of the library being used, if $tries is bigger than 256, it simply returns an error code, instead of doing the tests 21:56
Secondly, if do $x.is-prime(4) twice, the second call will perform the exact same computations as the first call, due to that lookup table. 21:57
PerlJam BenGoldberg: why do you want to change is-prime in these ways? What's the motivation? 21:58
Are you trying to minimize surprises when someone passes a low value for $tries? 21:59
(or a too-high value)
BenGoldberg rn: 5.is-prime(10000)
camelia rakudo 37c995, niecza v24-37-gf9c8fc2: ( no output )
BenGoldberg rn: say 5.is-prime(10000)
camelia rakudo 37c995: OUTPUT«False␤»
..niecza v24-37-gf9c8fc2: OUTPUT«True␤»
PerlJam Seems to me that one would be better served by a constraint on $tries 22:00
If nothing else, you could get a meaningful message when you do something screwy :) 22:01
BenGoldberg rn: say 5.is-prime(257)
camelia rakudo 37c995: OUTPUT«False␤»
..niecza v24-37-gf9c8fc2: OUTPUT«True␤»
BenGoldberg If separate calls to is-prime were specified to use independent, random numbers for the tests, then $x.is-prime(1000) would be semantically the same thing as doing [&&] map *.is-prime, 1..4 22:06
err, make that [&&] map *.is-prime(250), 1..4 22:07
err, make that [&&] map $x.is-prime(250), 1..4
But do to libtommath (used by nqp, used by rakudo) using the same "random" numbers each and every time, the results cannot be combined that way. 22:09
japhb_ masak: blog typo: "rectangula haiku" 22:11
masak japhb_: thanks, fixing. 22:13
masak BenGoldberg: that does sound like a serious issue. 22:15
cognominal I suppose that if I want to avoid infinite recursion on some of my rules in my bast grammar, I must express them using <EXPR> and precedence? 22:21
jnthn cognominal: Main thing is be sure to avoid left-recursion. 22:22
cognominal ya, that's exactly what I mean 22:23
japhb_ I understand the desire for independent tests in is-prime, but couldn't the immediate solution be to just cap the test count so that Rakudo does the right thing on insanely high inputs? After all, I suspect 2**-256 is a WAY smaller probability than the chance of bit error, even with ECC.
japhb_ *insanely high test counts 22:23
cognominal also I will need to hook the top of my grammar in nqp <EXPR> 22:24
masak 'night, #perl6 22:32
FROGGS gnight masak 22:35
dalek p: ba958d2 | jnthn++ | src/vm/jvm/ (2 files):
Support calling invokedynamic BS with extra args.
 22:37
p: b1d4669 | jnthn++ | src/vm/jvm/ (2 files):
Pass name as extra string arg.

This avoids running into issus where we use chars that are not allowed in JVM method names, fixing the 11-sub.t regression.
p: 55e954c | jnthn++ | src/vm/jvm/ (2 files):
Pass call site index as static data, not per call.
cognominal g'night 22:42
dalek : 6b5eea2 | (Tobias Leich)++ | lib/Perl6/P5World.pm:
use nqp::getenvhash
 22:51
: 31ca7ab | (Tobias Leich)++ | lib/Perl5.pm:
move slang setup to EXPORT sub, and fiddle with CUSTOM_LIB<Perl5>, which is our @INC
: ff9ed03 | (Tobias Leich)++ | lib/Perl6/P5Actions.pm:
add implicit :from<Perl5> for require and dispatch @INC to %CUSTOM_LIB<Perl5>
: e88c222 | (Tobias Leich)++ | lib/Perl6/P5Grammar.pm:
add implicit :from<Perl5> for use
: 2e5ba5e | (Tobias Leich)++ | t/test_summary:
sort the tests by pass and plan
 22:52
timotimo masak++ # t2 review 23:01
"because it bleads through into the whole" - probably typo'd "bleeds" here? 23:02
japhb_ The problem with being around punsters all the time is that your sense of spelling goes to hell. 23:07
BenGoldberg rn: my @a = 0; push @a, map 1-*, @a for 1..6; say @a 23:20
camelia rakudo 37c995, niecza v24-37-gf9c8fc2: OUTPUT«0 1 1 0 1 0 0 1 1 0 0 1 0 1 1 0 1 0 0 1 0 1 1 0 0 1 1 0 1 0 0 1 1 0 0 1 0 1 1 0 0 1 1 0 1 0 0 1 0 1 1 0 1 0 0 1 1 0 0 1 0 1 1 0␤»
BenGoldberg Just for fun, how would I make a lazy infinite list similar to that one? 23:21
(That's the Thue-Morse, in case anyone's curious) 23:22
(That's the Thue-Morse sequence, in case anyone's curious)
timotimo you can always gather/take, can't you? 23:25
japhb_ camelia is responding quite slowly; is everyone else trying stuff out? :-) 23:28
r: my @a = 0, -> *@b {map 1-*, @b} ... *; say @a[^50]; 23:30
camelia rakudo 37c995: OUTPUT«0 1 1 0 1 0 0 1 1 0 0 1 0 1 1 0 1 0 0 1 0 1 1 0 0 1 1 0 1 0 0 1 1 0 0 1 0 1 1 0 0 1 1 0 1 0 0 1 0 1␤»
BenGoldberg :)
[Coke] japhb_++ 23:31
japhb_ I find it interesting that the operational part of the infinite stream answer (the part before say) is actually shorter than the finite list code 23:33
timotimo do i see correctly that that sequence just kind of repeats after a pretty shord period? 23:36
BenGoldberg It never repeats exactly 23:38
japhb_ timotimo, I don't see it -- it is tricky on the eye, but so far I'm seeing almost-the-same for 8 or 10 bits and then a missing or flipped bit 23:39
timotimo ah, ok
i could just look up the wikipedia article ;)
japhb_ Wow, that is one cool sequence, for how simply it is defined. 23:43
timotimo it reminds me a bit of how you generate a whatsitcalled code, where going +1 or -1 will always be one bit flipped 23:45
japhb_ The third paragraph in the WP article under "Some properties" reminds me a lot of Penrose tilings ... while non-periodic, any layout of radius R is guaranteed to appear again within a distance of nR (and I think for the classic dart/kite, n == 2?)
timotimo but that one's made rather recursive
japhb_ Grey code? 23:48
Sorry, Gray Code 23:49
I always forget which way his name was spelled
timotimo i think that's the one, yes 23:50
you take the sequence so far, prepend it with a 0 and a 1 and flip one of them around i believe
BenGoldberg Here's another way of generating it 23:59
r: my @a = 0; @a := map {$_,1-$_}, @a for 1..6; say @a
camelia rakudo 37c995: OUTPUT«0 1 1 0 1 0 0 1 1 0 0 1 0 1 1 0 1 0 0 1 0 1 1 0 0 1 1 0 1 0 0 1 1 0 0 1 0 1 1 0 0 1 1 0 1 0 0 1 0 1 1 0 1 0 0 1 1 0 0 1 0 1 1 0␤»

Powered by Cro and the Raku® Programming Language.

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

Thank you!