»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or rakudo:, std:, or /msg camelia p6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by masak on 12 May 2015.
00:02 firstdayonthejob joined
[Coke] konobi: no clue, sorry 00:05
00:10 firstdayonthejob left 00:12 rurban joined 00:13 tokuhirom_ joined 00:17 tokuhirom_ left 00:18 Hor|zon joined 00:19 [Sno] left 00:20 [Sno] joined 00:23 Hor|zon left 00:28 testing joined 00:29 testing is now known as Guest77431, aborazmeh joined, aborazmeh left, aborazmeh joined 00:41 bjz left, bjz joined
psch konobi: i'd look in Makefile-JS.in first 00:43
konobi: at least from how the other backends build, the gen-* usually just does stuff slightly beyond concatenation
konobi: although -js might of course do things differently and pmurias is definitely the person to ask... :) 00:44
(of course it also depends on which part of the build you want to affect...)
konobi psch: i think it's that gen script that generates the Makefile-JS.in 00:45
psch konobi: i'm pretty sure that's not the case. the gen- script generates a Makefile out of Makefile-common.in and Makefile-$BACKEND.in 00:49
konobi: github.com/perl6/nqp/blob/master/t...file-JS.in it's on the repo, so it's not generated. it *might* get altered by the gen- script, but that'd be different than the other backends
00:50 llfourn joined
psch konobi: and i'd still be confident in saying that Makefile-JS.in won't get altered on disk, if so a copy gets altered 00:50
00:50 weihan joined
psch konobi: from the looks of it, gen-js-makefile.nqp writes the Makefile used to build nqp-js 00:51
00:52 Peter_R left
psch although i don't see a reference to Makefile-JS.in 00:52
00:53 weihan left
psch konobi: Configure.pl calls fill_template_fill from tools/NQP/Configure.pl, which (indirectly) does some processing and (basically) cats the template into Makefile 00:55
00:55 weihan joined
psch s/_fill/_file/ 00:55
psch .oO( we do have a quite interesting build system... )
00:55 llfourn left 00:56 weihan left 00:58 weihan joined 00:59 weihan left, vendethiel left 01:02 weihan joined
dalek kudo/nom: 5cd7ef0 | TimToady++ | src/Perl6/ (2 files):
make with/without topicalize properly
 01:05
TimToady finally...
konobi psch: >.< 01:08
01:10 tokuhiro_ joined 01:14 tokuhiro_ left
dalek ast: b26e8b3 | TimToady++ | S04-statements/with.t:
unfudge with topicalization
 01:16
01:18 cognominal left 01:19 bjz_ joined, Hor|zon joined
psch konobi: not sure what you're trying to tell me..? :) in any case, what is that you want to change about the nqp-js build? 01:19
01:20 bjz left 01:24 Hor|zon left 01:25 Jaglor joined
konobi psch: just trying to work it out 01:30
01:36 leont left 01:45 BenGoldberg left, ilbot3 left 01:46 ilbot3 joined 01:55 araujo_ joined 01:56 araujo left
dalek osystem: cf85616 | (Fayland Lam)++ | META.list:
Add Gravatar::URL to ecosystem

See github.com/fayland/perl6-Gravatar-URL		 02:05
02:06 llfourn joined 02:14 noganex_ joined 02:17 noganex left 02:20 rurban left, Hor|zon joined 02:25 Hor|zon left 02:30 tokuhirom_ joined 02:34 tokuhirom_ left 02:36 Sgeo__ left 02:37 Sgeo joined 02:38 llfourn left
lichtkind good night 02:42
02:48 ngie joined
dalek kudo-star-daily: bfd9d62 | coke++ | log/ (9 files):
today (automated commit)
 02:52
kudo-star-daily: 5a1ca12 | coke++ | log/ (8 files):
today (automated commit)
rl6-roast-data: 6639af4 | coke++ | / (9 files):
today (automated commit)
kudo/nom: 4f2ee3d | TimToady++ | src/Perl6/ (2 files):
fix topicalization of without as well
 02:53
ast: 47f8eb0 | TimToady++ | S04-statements/with.t:
test with/without much more thoroughly
02:54 Suzeanne joined, weihan left 02:57 eliasr left
dalek ast: df29028 | TimToady++ | S04-statements/unless.t:
test that 'unless true' returns Slip
 02:59
03:00 xxx left, tokuhiro_ joined 03:01 lichtkind left, ingy left 03:03 tokuhirom_ joined, xxx joined, xxx is now known as Guest36282 03:04 tokuhiro_ left 03:09 ngie left 03:12 Suzeanne left 03:17 kid51 left 03:21 Hor|zon joined 03:26 Hor|zon left 03:29 llfourn joined 03:37 Jaglor left 03:43 tokuhirom_ left 03:45 aborazmeh left 03:54 Brynn joined 03:56 weihan joined 03:57 Brynn left 04:06 tokuhiro_ joined 04:10 tokuhiro_ left 04:22 Hor|zon joined
llfourn my perl6 REPL is no longer working nicely with ← giving me ^[[D 04:27
after rebuilding -- how2fix?
04:27 Hor|zon left
llfourn ^^ dw I fixed it :) 04:28
04:29 Suzeanne joined 04:33 khw left
konobi llfourn: might be worth looking at the npm library "blessed". It has a native implementation of tput... should work for *nix and win32 04:41
masak m: sub foo($n) { $n == 2 ?? Empty !! $n }; say (1, 2, 3).map(&foo) 04:43
camelia rakudo-moar 4f2ee3: OUTPUT«(1 3)␤»
masak m: sub foo($n) { $n == 2 ?? Slip.new !! $n }; say (1, 2, 3).map(&foo)
camelia rakudo-moar 4f2ee3: OUTPUT«(1 3)␤»
masak is there any detectable difference between Empty and Slip.new ?
llfourn konobi: I was just talking about the inbuilt Perl 6 REPL.... I just needed to reinstall Linenoise. blessed looks cool though :) 04:44
konobi yeah, but blessed it all native... also doesn't even need readline 04:45
llfourn konobi: You use it with Perl6 somehow 04:47
?
konobi nope, just worked with it in the early days
um... blessed that is
though I also worked with perl6 back in the early days too ^_^ 04:48
masak :)
llfourn I see :)
masak how early are the early days, OOC?
04:51 araujo_ left
konobi masak: good ol' pugs era 04:51
btw... how self-contained is moarvm? 04:53
04:54 baest_ joined 04:55 ollej_ joined 04:56 nyar1ath0tep joined 04:57 psch_ joined, ashleydev joined, jojotus_ joined, riatre_ joined, boegel|quassel joined, xenu_ joined 04:58 grondilu_ joined, erdic_ joined, atta_ joined
konobi jnthn: ping! 04:59
05:00 kst` joined, nyar1ath0tep left, masak_ joined 05:01 masak_ is now known as masak_grr, Hotkeys_ joined 05:02 hobbified joined, ilbot3 left, [Tux] left, gensym left, grondilu left, boegel left, Gardner left, preyalone left, revdiablo left, baest left, atta left, Hotkeys left, jojotus left, lsm-desktop left, kipd left, SmokeMachine_ left, erdic left, avuserow1 left, Debolaz left, zemmihates left, Spot__ left, xenu left, dylanwh_ left, ashleyde1 left, psch left, ugexe left, riatre left, Fluor left, ollej left, masak left, hobbs left, kst left, erdic_ is now known as erdic 05:03 gensym_ joined, gensym_ is now known as gensym, revdiablo joined
masak_grr ah, Pugs. 05:04
05:05 Gardner joined
masak_grr I wonder how path Perl 6 would have taken if not for Pugs. 05:05
konobi masak_grr: it was a lot of fun at the time... but it was parrot timelines that turned me off, for sure.
masak_grr: the current "perl6 1.0" release is moarvm based, right? 05:08
05:08 SuzieQueue joined, lsm-desktop joined, Fluor joined, araujo joined, araujo left, araujo joined, [Tux] joined 05:09 ingy joined, tokuhirom_ joined, mrsolo left, dylanwh_ joined 05:10 ugexe joined, Debolaz joined, avuserow1 joined 05:11 ilbot3 joined 05:12 Suzeanne left, kipd joined, SmokeMachine_ joined 05:14 preyalone joined 05:15 Spot__ joined, zemmihates joined
TimToady konobi: jnthn++ is off on his honeymoon 05:16
konobi TimToady: oh wow... nice! if there's someone worthy of having a nice time off, mr worthington is definitely up there. 05:17
TimToady: how's things? excited to finally release before christmas? =0)
TimToady it...please me :) 05:21
*pleases
m: if 0 {} orwith Nil { .say } else { .say } 05:23
camelia rakudo-moar 4f2ee3: OUTPUT«Too many positionals passed; expected 0 arguments but got 1␤ in block <unit> at /tmp/FM2qMOlbSk:1␤␤»
TimToady but I found a bug...
konobi TimToady: <3
05:23 Hor|zon joined
konobi TimToady: have been out of the perl world for a while... but with the announcement... definitely swinging back in to have a peek 05:24
TimToady well, we still have lots of work to do before Christmas...
but it's nice to officially be in beta state now
konobi =0)
so it's moarvm based now, right? parrot is still off in the sidelines? 05:25
TimToady yes, moarvm is the main effort, but looks like jvm is sticking close too 05:26
parrot...not so much
and there's a js backend coming along too
konobi TimToady: have a look at the vmkit stuff that's available for llvm at all?
TimToady llvm is very...ll 05:27
konobi TimToady: I'll try to help with the JS backend as much as I can... since I'm pretty tied to nodejs... for my sins.
TimToady ok, pmurias++ is the one working on it mostly
05:27 Hor|zon left
TimToady he might well be glad of some assistance 05:28
now, to figure out this bug... 05:30
konobi TimToady: rgr! old belfast.pm folk to the rescue! 05:31
05:34 masak_grr is now known as masak
masak konobi: yay, JS backend! \o/ 05:35
I still belive that one will be a game-changer in unknowable ways. 05:36
(simply because JavaScript is eating everything and things will only survive long-term to the extent they compile down JavaScript)
konobi masak: ruh-roh
masak down to*
konobi masak: oh hey... masak! how's it going! 05:37
masak :D
konobi now I clock ^_^
masak well, hey. :) yet another survivor from back in the Pugs days. :)
konobi: you may be doing nodejs for your sins -- I'm doing Perl 6 macros in Rakudo for my sins ;) 05:38
TimToady: speaking of which, I have a question for you about unquotes. 05:39
TimToady: given the code `quasi { my $x = "OH NOES"; { {{{$stmt}}}; say($x) } }`, is there any AST $stmt could be that would make that code print "OH HAI\n"? 05:40
konobi masak: yar... node has taken up a lot of my life, the last few years... but mainly because i was involved very early.
masak konobi: sounds... rather nice, actually. I mean, node seems very interesting. and v8, too.
konobi: I'm very into JS these days, though not node in particular. just, you know, the whole thing.
I definitely should do more server-side things. 05:41
and build/toolchain things
llfourn speaking of JS is there any way to tie a self to a bit of code like in JS? 05:42
konobi masak: the fact that nodejs and npm use TAP, is genuinely my fault =0)
masak huh, Perl 6 mentioned in news.ycombinator.com/item?id=10404720 -- a thread about Ravi, a typed Lua
konobi: kudos :)
llfourn: depends what you mean.
05:42 jojotus_ is now known as jojotus
konobi the big thing for node at the time and still to this day, is that it's async from day one. 05:43
llfourn tries to figure out what he means
masak m: class A { has $.x; method foo { say($.x) } }; class B { has $.x }; A.^find_method("foo")(B.new(:x("OH HAI")))
camelia rakudo-moar 4f2ee3: OUTPUT«Type check failed in binding <anon>; expected A but got B␤ in method foo at /tmp/DeTUyyeFKg:1␤ in block <unit> at /tmp/DeTUyyeFKg:1␤␤»
masak hm :)
m: class A { has $.x; method foo { say($.x) } }; class B is A { has $.x }; A.^find_method("foo")(B.new(:x("OH HAI")))
camelia rakudo-moar 4f2ee3: OUTPUT«OH HAI␤»
masak so, that works :) 05:44
it's a kind of "tie a self" thing, if you will
konobi masak: any idea how independent moarvm is at all?
llfourn masak: But can you return a code object with the self tied into it without calling it
konobi masak: ie: could it be easily bundled in a statically linked fashion? 05:45
masak llfourn: just put it in a lambda, I guess? 05:46
konobi: dunno -- maybe ask #moarvm?
konobi also... how useful is moarvm by itself o.O
masak &
konobi masak: freenode?
llfourn masak: yeah -- that's what I'm doing. I was thinking that maybe you could return the method object and somehow prebind self into it and then just $method-install.()... 05:47
I mean $method.() I don't know what "install" is there
developer.mozilla.org/en-US/docs/W...ction/bind <= that is what I meant 05:55
but I guess lambdas are much less cumbersome in p6 v JS so not super necessary 05:57
dalek kudo/nom: ef77bd7 | TimToady++ | src/Perl6/Grammar.nqp:
'else' needs $/ update to see $<sym>[-1]
 06:02
ast: e8bc45e | TimToady++ | S04-statements/with.t:
test mixed if/with/else usage
TimToady masak: I presume a statement resembling $::("OUTER::x") = "OH HAI"; would work 06:04
06:07 tokuhiro_ joined
konobi TimToady: is there any dtrace in the implementation currently? 06:10
06:12 tokuhiro_ left
TimToady wha? dtrace is more like a kernel level thing, afaik 06:12
konobi nope 06:14
it's a utility you can use from kernel the whole way up through userland
TimToady then I don't understand "in the implementation" 06:15
konobi while lots of the core syscalls, and libc calls are covered... you can add dtrace instrumentation to lots of stuff... perl5 has it, jvm has it, ruby and python have it... nodejs mostly does... 06:16
llfourn perl5 doc for dtrace: perldoc.perl.org/perldtrace.html 06:17
TimToady feel free to hack something in :)
TimToady is just the language designer...
konobi llfourn: that's a little differnt too 06:18
brb... lemme bug the dtrace guy
06:24 Hor|zon joined 06:29 bjz_ left, Hor|zon left
masak TimToady: I realize I didn't constrain my question enough -- you're not allowed to modify the existing $x :) 06:29
TimToady: what I'm after is -- is there any way for an unquote to create a new variable?
in the *current* scope 06:30
06:31 domidumont joined
masak because if there is, as soon as there's an unquoted statement or bigger, all bets are off. we can't even say "Variable '$y' is not declared", because for all we know, it was declared in the unquoted part. 06:31
06:32 boegel|quassel left
masak I take solace in the fact that (I hope) something like `my $COMPILING::y = "OH HAI";` does not make sense inside a Qtree injected into an unquote. I'm just wondering if there are other possible loopholes which might allow that. 06:33
06:33 boegel joined 06:35 domidumont left 06:37 domidumont joined 06:40 bjz joined 06:43 softmoth left 06:44 pmurias joined
pmurias konobi: hi 06:44
konobi: tools/build/gen-js-makefile.nqp is the thing that needs to be updated 06:45
06:50 llfourn left
masak what's the idiomatic way in Perl 6 to say "$x and $y are of exactly the same type" ? 07:00
$x.WHAT === $y.WHAT ?
07:02 karim joined
karim I want my class to be able to parse itself from json, something like 07:03
class C1 {
method parse-from-json($json) {
}
}
1) How can I make that method static so I'll be able to call it on the class level?
2) how to return "self" being an instance of C1 from "parse-from-json"?
masak m: class C1 { method parse-from-json($json) { say "look ma, no instance!" } }; C1.parse-from-json('{}') 07:04
camelia rakudo-moar ef77bd: OUTPUT«look ma, no instance!␤»
masak that's your first question, I guess. by default, methods can be called both on the class and on instances. 07:05
m: class C1 { has $.json; method parse-from-json($json) { self.bless(:$json) } }; C1.parse-from-json('{}').json 07:06
camelia ( no output )
masak m: class C1 { has $.json; method parse-from-json($json) { self.bless(:$json) } }; say C1.parse-from-json('{}').json
camelia rakudo-moar ef77bd: OUTPUT«{}␤»
07:06 karim left
masak and that's the answer to your second question, I guess. you call `self.bless` to construct an instance, passing as named parameters the attribute values. 07:06
ShimmerFairy masak: though S12 does say it should default to :D on invocants, so really you should need to do method foo(C1:_:) {...} to get the current behavior (I don't know if there are still plans to fix that, or keep current behavior) 07:09
masak ShimmerFairy: S12: 'If you declare an ordinary method, it can function as a "class" method when you pass it a type object such as "C<Dog>" regardless of how defined the prototype object is, as long as the method body doesn't try to access any information that is undefined in the current instance.' 07:13
ShimmerFairy: which part of S12 were you reading? :)
ShimmerFairy "In standard Perl 6, Int is generally assumed to mean Int:_, except for invocants, where the default is Int:D. (The default new method has a prototype whose invocant is :U instead, so all new methods all default to allowing type objects.)" 07:14
masak: ^^^ that one :)
masak I'd say those two parts are in opposition. 07:15
(and that Rakudo is currently implementing the one I dug up.)
ShimmerFairy For the record, the part I read is newer by about a year :P (2011 compared to 2010) 07:16
masak yes, I suspected that. 07:17
but the newer part is still conjectural, and unimplemented.
07:21 bjz left 07:23 bjz joined, ifim left 07:25 Hor|zon joined 07:30 Hor|zon left 07:32 kaare_ joined 07:35 rurban joined
masak heh. that thing where you go "why isn't this passing? this should be passing!", and then you notice that the expected value in the test is wrong :P 07:39
07:44 kanishka joined, kanishka left 07:50 llfourn joined 07:55 RabidGravy joined, CIAvash joined 08:05 firstdayonthejob joined 08:09 darutoko joined, tokuhiro_ joined 08:14 tokuhiro_ left 08:26 Hor|zon joined 08:31 Hor|zon left
FROGGS yay, I finally seem to have a proper fix for subst-in-a-loop 08:34
it took three evenings :S
08:38 margeas joined 08:43 s_kilk joined
moritz that's the nature of some bug fixes 08:44
dalek osystem: 7c1cb44 | (Shane Kilkelly)++ | META.list:
Add Config::Clever to ecosystem

See github.com/ShaneKilkelly/perl6-config-clever		 08:48
osystem: d32e392 | RabidGravy++ | META.list:
Merge pull request #80 from ShaneKilkelly/master

Add Config::Clever to ecosystem
08:49 Ven joined 08:50 vendethiel joined 08:51 Ven left 09:02 kjs_ joined
[Tux] test 50000 37.913 37.801 09:05
test-t 50000 39.782 39.669
09:21 szabgab joined 09:22 Zoffix joined
Zoffix .botsnack 09:22
yoleaux :D
synbot6 om nom nom
yoleaux 14 Oct 2015 18:15Z <azawawi> Zoffix: it was the HTTP code 303 after all that caused the hanging. Now it is working :)
14 Oct 2015 21:48Z <ZoffixMobile> Zoffix: check out github.com/tony-o/perl6-hiker when you're sober
15 Oct 2015 15:53Z <tony-o_> Zoffix: the other thing to check out, one step more removed from hiker is HTTP::Server::Router - github.com/tony-o/perl6-http-server-router
09:23 s_kilk left
szabgab p6: say 1+int rand*6 09:23
camelia rakudo-moar ef77bd: OUTPUT«5===SORRY!5=== Error while compiling /tmp/tmpfile␤Two terms in a row␤at /tmp/tmpfile:1␤------> 3say 1+int7⏏5 rand*6␤ expecting any of:␤ infix␤ infix stopper␤ postfix␤ statement end␤ statemen…»
szabgab p6: say 1+int(rand*6)
camelia rakudo-moar ef77bd: OUTPUT«Cannot invoke this object␤ in block <unit> at /tmp/tmpfile:1␤␤»
szabgab p6: rand*6
camelia rakudo-moar ef77bd: OUTPUT«WARNINGS:␤Useless use of "*" in expression "rand*6" in sink context (line 1)␤»
szabgab p6: 6*rand
camelia rakudo-moar ef77bd: OUTPUT«WARNINGS:␤Useless use of "*" in expression "6*rand" in sink context (line 1)␤»
szabgab p6: say 6*rand
camelia rakudo-moar ef77bd: OUTPUT«0.687734836974753␤»
09:23 bjz left
szabgab p6: say int(6*rand) 09:24
camelia rakudo-moar ef77bd: OUTPUT«Cannot invoke this object␤ in block <unit> at /tmp/tmpfile:1␤␤»
moritz m: say (6*rand).Int 09:25
camelia rakudo-moar ef77bd: OUTPUT«4␤»
moritz m: say (^6).pick
camelia rakudo-moar ef77bd: OUTPUT«5␤»
RabidGravy I was just going to suggest that
moritz szabgab: .pick on a Range is the idiomatic way to generate a random integer in p6 09:26
szabgab p6: (1..6).pick
camelia ( no output )
szabgab p6: say (1..6).pick
camelia rakudo-moar ef77bd: OUTPUT«6␤»
moritz m: .say for (1..6).roll(10)
camelia rakudo-moar ef77bd: OUTPUT«1␤6␤4␤4␤4␤5␤4␤2␤1␤1␤»
moritz or .roll, for that matter
szabgab moritz: thanks, it would be nice to mention this on doc.perl6.org/routine/rand 09:27
p6: say (1..6).roll
camelia rakudo-moar ef77bd: OUTPUT«1␤»
szabgab p6: say (1..6).roll
camelia rakudo-moar ef77bd: OUTPUT«5␤»
09:27 Hor|zon joined
RabidGravy It's mentioned somewhere in the docs, but can't remember where 09:28
szabgab I mean a link from rand to pick and roll 09:29
CurtisOvidPoe I now have a second box on which I’m getting this panda failure. I can’t build any Perl 6 modules as a result. gist.github.com/Ovid/8d3085d466110d78aeda
Running “rakudobrew build moar” works just fine, however. 09:30
RabidGravy that's curious 09:31
09:31 ^elyse^ joined 09:32 Hor|zon left
CurtisOvidPoe Running it through the debugger now. Last time I tried I got stuck and didn’t have the time to finish. 09:32
Zoffix Weird: "Perl v6.0.0 required--this is only v5.20.2" seems it's using perl instead of perl6 09:33
RabidGravy CurtisOvidPoe, is this a completely fresh install or was there a rakudo on there before? 09:34
CurtisOvidPoe I was wondering if somehow perlbrew and rakudobrew were incompatible.
RabidGravy I seem to recall something like this from ages ago
CurtisOvidPoe RabidGravy: on my brand, spanking new MacBook, it’s a fresh install.
But I’ve coped over my .bash* files and I suspect there’s something in there which might cause an issue. 09:35
RabidGravy could be a weird interaction with perlbrew
CurtisOvidPoe Getting closer. I can manually replicate this with: ‘cd /Users/curtispoe/.rakudobrew/moar-nom/panda; /Users/curtispoe/.rakudobrew/bin/../moar-nom/install/bin/perl6 rebootstrap.pl’ 09:38
moritz PSA: I've just tweaked permissions of the perl6 organization on github. Every member can now create a new repo in that organization. 09:39
Zoffix \o/ 09:40
moritz along with some more permissions regarding teams in that org
masak got it. thanks. 09:41
szabgab could you please review: perl6maven.com/number-guessing-game ? 09:48
Zoffix CurtisOvidPoe, is your PROVE_COMMAND env var set to anything? 09:49
CurtisOvidPoe Nope.
Still digging in :)
Zoffix :)
09:49 FROGGS_ joined 09:50 bjz joined 09:51 FROGGS left 09:57 rindolf joined
Zoffix I've reached a dead end here: github.com/tadzik/panda/blob/maste...ter.pm#L24 :) Seems it's not picking up the "-e /Users/curtispoe/.rakudobrew/bin/../moar-nom/install/bin/perl6" bit and runs p5 instead. But I don't have an OSX to test on :P ¯\_(ツ)_/¯ 10:00
10:01 ggoebel left
CurtisOvidPoe Zoffix: at that point, $*EXECUTABLE is set to /Users/curtispoe/.rakudobrew/bin/../moar-nom/install/bin/perl6, and $prove-command is set to ‘prove’. 10:04
But I’m guessing you know that :) 10:05
Zoffix I suspected so :) I wonder if that '/../' bit tripping something up in p5 land. 10:08
masak we never did reverse our decision to make .bless only reach dot-sigil attributes. is that a thing we might want to get in before Sixmas?
10:09 ggoebel joined
masak dot-sigil right now is "that arbitrary thing you have to do in order to have the default BUILD recognize your attribute" 10:10
CurtisOvidPoe Bah. I run the command manually (prove -e /Users/curtispoe/.rakudobrew/bin/../moar-nom/install/bin/perl6 -Ilib t/builder.t) and I get “Could not find Panda::Common in any of: …”
10:10 tokuhiro_ joined
CurtisOvidPoe Well, will you look at that. When I set PERL6LIB correctly from the command line, it all passes ( PERL6LIB=/Users/curtispoe/.rakudobrew/moar-nom/install/share/perl6/site/lib,ext/File__Find/lib,ext/Shell__Command/lib,ext/JSON__Fast/lib,lib prove -e /Users/curtispoe/.rakudobrew/bin/../moar-nom/install/bin/perl6 -Ilib t/builder.t ) 10:12
ShimmerFairy masak: from the perspective of "$.foo is public, $!foo is private", it makes sense. From the perspective of "$!foo is an attribute, $.foo will generate a getter method for it too", not so much :) 10:13
masak ShimmerFairy: I remember how it came about. it was in a p6c thread I started, I believe.
ShimmerFairy: the intent was noble: prevent excessive knowledge of truly private things 10:14
10:14 szabgab left
masak the eventual fallout was twofold: disallow .new/.bless from installing values in private attrs, and disallow .perl from showing them 10:15
m: class C { has $!x = 42 }; say C.new.perl
camelia rakudo-moar ef77bd: OUTPUT«C.new␤»
10:15 tokuhiro_ left
masak :( 10:15
I'm not so sure those decisions were correct anymore.
.perl is a debugging tool, a way to say "OK, just show me what's inside"
ShimmerFairy masak: I think the problem is that "public" and "private" don't mean anything truly special in class attributes. Now that I think about it, I find it hard to believe that the distinction exists for P6 at all.
masak it's not necessarily a tool that needs to play by encapsulation rules
ShimmerFairy: "private" does mean something. Perl 6 objects really are opaque. the question is more what role you are in when you do .new and .perl, and whether normal concerns of encapsulation apply there. 10:16
dalek kudo/nom: 32d7b08 | FROGGS++ | src/ (2 files):
fix usage of match variable in replacement part of substitution

  $_ ~~ s[(.)] = "$0" compiled to $_.subst{-mutate} before. This was
problematic in so far that the $0 in the string were interpolated before the match (.) happened. The we compile to a method call to .match and APPLY-MATCHES separately, which gives us the opportunity to set $/ before interpolating the match result in the right hand side.
ShimmerFairy masak: In other words, the $.foo form feels less like "this is an attribute of the public kind", and more like "this is sugar for a $!foo attr and writing both a 'method foo'" 10:17
masak well, yes 10:18
but those things are for "normal" use of a class
ShimmerFairy Aside from the artificial-seeming limitation of the default BUILD, I don't think there's a real, technical difference. I think P6's version of "public" turned out to be implemented as just "methods returning a [private] attribute"
masak: here's the problem, though: I can define a method foo { $!foo } , and suddenly it's public. I can define my own method foo { die } , and now the 'foo' attribute is private. 10:19
CurtisOvidPoe Zoffix: By the time we get to run-and-gather-output(*@command), everything seems fine. If I skip the Proc::Async code (to avoid trying to debug that!) and let it run via ‘my $p = shell("$cmd 2>&1", :out);’, it fails for the same reason. If I run the $cmd (setting PERL6LIB) from the command line, it works just fine.
So are there any known bugs regarding the shell() command on OS X? 10:20
RabidGravy .tell szabgab you might find something like gist.github.com/jonathanstowe/c569...f13cb43abd is even more idiomatic
yoleaux RabidGravy: I'll pass your message to szabgab.
RabidGravy .tell szabgab but otherwise yeah great :) 10:21
yoleaux RabidGravy: I'll pass your message to szabgab.
masak m: class C { has $!x; method !x() is rw { return-rw $!x }; method new { my $o = self.bless; $o!x = 42; $o } }; say .^attributes[0].get_value($_) given C.new 10:22
camelia rakudo-moar ef77bd: OUTPUT«42␤»
masak ShimmerFairy: I don't see that as a problem, as long as the default accessor method does the right thing.
ShimmerFairy masak: unless there's a real difference between public and private beyond just "there is[n't] a method for it" that I'm unaware of, I think you could actually make the argument that Perl 6 doesn't technically have a concept of public/private attributes, they're all "private" :) 10:23
10:25 bjz_ joined 10:27 bjz left
CurtisOvidPoe OK, I don’t think it’s the shell() command because I have a small test case which runs just fine with that :/ 10:28
10:28 Hor|zon joined
CurtisOvidPoe How come when I run a shell command and print the results with “for $result.out.lines { .chars && .say; }”, it works fine but ends with the warning “Invocant requires an instance of type IO::Pipe, but a type object was passed. Did you forget a .new?”? 10:29
masak ShimmerFairy: yes, that argument can indeed be made. which only strengthens the case for re-allowing .new/.bless/.perl to reach them 10:31
(because the distinction "has an auto-generated accessor" isn't a very meaningful one)
ShimmerFairy masak: that's why, now that I'm thinking about it, I can't really see much of a point for making .new/.bless/.perl ignore "private" attributes :) 10:32
(and now I also wonder what the distinction ends up being like in other languages :P)
masak that's what I'm saying.
it was a noble goal, but it doesn't really make sense to exclude accessor-less attributes. 10:33
10:33 Hor|zon left
CurtisOvidPoe I asume we can’t actually run the shell() command with p6? I’d try, but it seems rude :) 10:35
moritz CurtisOvidPoe: you may try; just don't do something fatal with it, in case it works 10:37
m: shell("echo foo")
camelia rakudo-moar ef77bd: OUTPUT«shell is disallowed in restricted setting␤ in sub restricted at src/RESTRICTED.setting:1␤ in sub shell at src/RESTRICTED.setting:15␤ in block <unit> at /tmp/GCmiBtbPDx:1␤␤»
llfourn learns that there is a a way to restrict use of some things in rakudo 10:39
CurtisOvidPoe Here’s the exact error I’m getting: gist.github.com/Ovid/fe4fb45360d32369cf67
(Along with the code to invoke it. I can’t see what I did wrong) 10:40
Hmm, should have been .chars, but that doesn’t impact it. 10:41
RabidGravy CurtisOvidPoe, need an :out on the shell() 10:42
llfourn perl6 -e 'shell("ls -l /")' by itself prints stuff out o.O 10:44
10:44 ^elyse^ left
RabidGravy yeah, but without the :out adverb it doesn't capture it hence the .out is just a type object 10:45
CurtisOvidPoe Cheers!
llfourn so it is simply outputting to stdout by default?
RabidGravy it's not capturing so the output of the command is just going to where it would go
llfourn orite 10:46
CurtisOvidPoe I feel like such a complete n00b :) 10:47
RabidGravy :) but we love you anyway 10:48
CurtisOvidPoe And I can verify that “my $result = shell("prove -e $*EXECUTABLE -Ilib test.p6"):out;“ works just fine in a small test case. It’s in building panda that something goes horribly wrong.
Does anyone here have an OS X box on which they have successfully done ‘rakudobrew install panda’? 10:58
RabidGravy sorry
10:58 CIAvash left 10:59 CIAvash joined 11:00 Nickmob joined 11:01 Nickmob left
llfourn CurtisOvidPoe: yes, but months ago 11:01
for me it's rakudobrew build-panda 11:02
11:03 Kogurr left 11:07 ^elyse^ joined 11:09 virtualsue joined
CurtisOvidPoe It’s something in my environment. I can now build panda with this: env -i PATH=/usr/bin:/Users/curtispoe/.rakudobrew/bin rakudobrew build-panda 11:12
RabidGravy weird 11:15
CurtisOvidPoe I’ll figure it out now, but I have a couple of other things I need to resolve first.
In short: this is great news :) 11:16
RabidGravy :)
in a completely un-related note what's the best way to capture control when the accessor is being added for a "public attribute" in order to apply a role to the created method? 11:18
11:18 bjz joined
moritz m: say Num.^can('sqrt').^name 11:19
camelia rakudo-moar 32d7b0: OUTPUT«List␤»
11:19 bjz_ left
moritz m: say Num.^can('sqrt')[0] does role { method flurb() { } }; 11:19
camelia rakudo-moar 32d7b0: OUTPUT«sqrt␤»
moritz RabidGravy: something like this? 11:20
RabidGravy well sort of - I want to do "has $.foo is something" and for the application of something to cause a role to be applied to the accessor's method 11:21
so adding the role is fine and I grok that, I just want to get at the accessor method as early as possible 11:22
moritz ah, you want to write a trait
RabidGravy yeah, I'm cool with writing the trait, just want to know (.e.g) if I can apply a role to the attribute object via the trait which in turn can influence the accessor method 11:23
moritz what difference does early versus late make? 11:24
llfourn RabidGravy: Maybe custom metaclass and override: docs.perl6.org/type/Metamodel::Attr..._attribute 11:25
11:26 kjs_ left
RabidGravy moritz, largely because what I actually have in mind is a form of "access control" for object members and in that case I figure that earliest is bestest 11:27
moritz m: role important { }; multi sub trait_mod:<is>(Attribute $a, important) { say "applying trait"; $a does important }; class Foo { has $.x is important }; 11:29
camelia rakudo-moar 32d7b0: OUTPUT«applying trait␤»
RabidGravy with a normal method it's fairly simple - trait on the method that dicks with the CALL-ME or something to provide the appropriate control
11:29 Hor|zon joined
moritz RabidGravy: fwiw if you supply your own accessor, none will be generated 11:29
11:29 CIAvash left
RabidGravy yeah, but doesn't make such a nice interface 11:30
dalek atures: 0286bce | (Zoffix Znet)++ | template.html:
Fix markup error
Zoffix Oh god >_< 11:31
moritz RabidGravy: I don't think we're quite on the same wavelength here
Zoffix So much for --patch when you still commit with your 'add --all' alias
moritz Zoffix: get rid of that alias, I'd say :-) 11:32
Zoffix :)
11:33 Hor|zon left
RabidGravy moritz, actually I got it I think, need to wrap Attribute.compose, callsame the original then get the method and apply the role to it 11:37
11:37 diana_olhovik_ joined 11:45 kjs_ joined 11:48 kjs_ left
nine r: sub s_s(*%n) { %n>>.say }; s_s(|{:assoc<list>},:assoc<left>); 11:49
yoleaux 17 Oct 2015 17:45Z <jnthn> nine: IIRC, it was .clone not being deep enough, but I don't remember exactly either :(
17 Oct 2015 17:45Z <jnthn> nine: Maybe not creating fresh Scalar containers or so
camelia rakudo-moar 32d7b0: OUTPUT«list␤»
..rakudo-jvm 32d7b0: OUTPUT«left␤»
11:52 BenGoldberg joined, Ven joined 11:53 Ven left, Ven_ joined 11:54 virtualsue left, xfix joined 11:55 spider-mario joined 12:02 karim joined
CurtisOvidPoe I found the problem with my ‘rakudobrew build-panda’. As a convenience, I’ve set my PERL5LIB to ‘lib’. If I run ‘PERL5LIB= rakudobrew build-panda’, it works just find. Why does setting my PERL5LIB to ‘lib’ break rakudobrew build-panda? 12:03
12:04 ^elyse^ left 12:06 ngie joined 12:07 ^elyse^ joined 12:08 kid51 joined, diana_olhovik_ left 12:09 telex left 12:10 azawawi joined
azawawi hi #perl6 12:10
12:10 telex joined
moritz CurtisOvidPoe: it might try to load a Perl 6 module from Perl 5 somewhere down the rabbit hole 12:11
llfourn CurtisOvidPoe: same for me! -- looks like putting 'lib' must goof one of the perl commands because it adds a lib and perl5 tries to run a perl6 thing.
azawawi github.com/azawawi/perl6-selenium-...es/ex01.p6 # Automated UI test: Searching using google for "Perl 6\n" keyword :) 12:12
Zoffix azawawi, neat. What User-Agent does it report to Google? 12:17
('cause I know google blocks Perl 5's LWP, for example) 12:18
dalek ast: f665268 | FROGGS++ | S04-statements/for.t:
unfudge/add tests for capture in subst in loop
azawawi Zoffix: well im not setting a user agent atm
Zoffix azawawi, that "sleep 1;" is iffy though. What if it takes more than a second? Is there no way to check whether the request has completed? 12:19
azawawi Zoffix: going to implements waits :)
Zoffix \o/
12:21 vendethiel left
karim is there any way to return an instance on the current class from a method without having to specify the name of the class? I'd like something like this: 12:21
class C1 { method m1() { ....; return self(1, 2, 3); } } 12:22
llfourn karim: self.new?
karim thx
azawawi Zoffix: then create a good documentation website like this one selenium-python.readthedocs.org/en/...arted.html
llfourn karim: my pleasure :)
12:23 karim left, leont joined
Zoffix azawawi, nice. That'll be a useful module. 12:23
azawawi yup
12:24 AlexDaniel left
azawawi Zoffix: your github ticket is really useful. We need more documentation. That's what drew me into Perl in the first place. 12:25
Zoffix: doc.perl6.org is useful but lacks organization to a newbie imho 12:26
Zoffix It lacks a more powerful search too.
azawawi Zoffix: we need also to cater to our users e.g. new to Perl, old Perl 5 user and existing Perl 6 user 12:27
saying infix, postfix to someone new to Perl is like saying to him/her go away :) 12:28
i.e. doc.perl6.org/routine.html 12:29
12:29 Ven_ left 12:30 karim joined, Hor|zon joined
azawawi comparing selenium-python.readthedocs.org/en/...arted.html to doc.perl6.org => no introduction or getting started whatsoever 12:31
karim I can't find the information about the "return", is it optional? in the documentation it's not specified, however it seems it is optional.
dalek atures: fd38b65 | (Zoffix Znet)++ | template.html:
Fix incorrect overflow on mobile (Closes #4). Fix missing footnotes (Closes #3)

pos:abs; on footnotes without a near parent causes them to invisibly overflow past .table-responsive
FROGGS_ azawawi: please file a doc issue 12:32
azawawi karim: return is optional
karim and how about ";" ?
llfourn azawawi: We do need a more general tutorial and to make tutorials more visually seperate.
azawawi llfourn: true 12:33
Zoffix karim, it's usually not optional. There are cases where it is, but even in those it's recommended that you place it anyway 12:34
azawawi m: sub foo { }; say foo.perl;
camelia rakudo-moar 32d7b0: OUTPUT«Nil␤»
azawawi m: sub foo { return }; say foo.perl;
camelia rakudo-moar 32d7b0: OUTPUT«Nil␤»
llfourn karim: return isn't documented properly anywhere afaik sorry!
azawawi m: sub foo { 1 }; say foo.perl;
camelia rakudo-moar 32d7b0: OUTPUT«1␤»
azawawi m: sub foo { ; }; say foo.perl; 12:35
camelia rakudo-moar 32d7b0: OUTPUT«Nil␤»
Zoffix FWIW: I was responding to "<karim> and how about ";" ?" :)
12:35 Hor|zon left
llfourn Zoffix: ah. Yeah I guess wtf is ';' is a question a lot of newbies will have. 12:36
this is where a general intro tutorial would be good
karim sorry, I don't understand. Is ";" optional or does it depend whether it's being used in the last sentence or not?
12:37 kjs_ joined
Zoffix karim, depends on whether it's last. if it's not last, then it's not optional. 12:37
12:37 TEttinger left
karim ok 12:38
12:38 karim left
dalek atures: 4cc9f61 | (Zoffix Znet)++ | template.html:
Remove debugging markup
 12:39
azawawi how can i embed github markdown in pod6 blocks? 12:47
rationale: markdown is way easier to work with than pod6 12:49
pink_mist disagrees with that rationale, but don't let that get in the way of you trying =) 12:50
12:50 vendethiel joined
moritz azawawi: indent it 4 spaces 12:52
azawawi tries it 12:53
vendethiel FROGGS_++ 12:54
12:55 spider-mario left 12:56 xenu_ is now known as xenu
azawawi moritz: but why 4 spaces btw? 12:57
12:57 cognominal joined
pink_mist isn't that the standard markdown indenation style? 12:58
moritz azawawi: that way it doesn't get falsly parsed as pod
the beauty is that you can use *any* markup in pod6 if you indent it 12:59
Zoffix 0.o
ShimmerFairy moritz: um... that's not right. Indenting just makes it an implicit =code block, and only in certain blocks. If you had, say, a line starting with an = in there, it'll be parsed as trying to begin a block 13:00
moritz ShimmerFairy: yes, and you can put markdown in there all you like 13:02
ShimmerFairy moritz: sure, but it'll just be a section of code. Suggesting code blocks as anything but that is just wrong.
RabidGravy couldn't you just do =begin markdown ... =end markdown ? 13:03
ShimmerFairy The real way to do it would be to put your markdown in some kind of =Markdown block, and then have a module or renderer that knew how to handle those blocks.
RabidGravy: yes, but not spelled 'markdown', that's a reserved block name :)
moritz RabidGravy: sure, but no renderer currently does anything with that block
RabidGravy well, surely if people want to use it then they should consider making a renderer ;-) 13:04
pmurias masak: re hiding private attributes from perl, how can you get them for debugging? 13:08
moritz pmurias: through the MOP
doc.perl6.org/type/Attribute#method_get_value 13:09
13:09 adhoc left
pmurias seems really inconvenient 13:09
what was the reason for hidding the from .perl?
moritz it's *private* 13:10
azawawi m: constant NUMPAD_0 = "\x01A";
camelia ( no output )
moritz private attributes aren't inistialized by the default constructor
azawawi m: constant NUMPAD-0 = "\x01A";
camelia rakudo-moar 32d7b0: OUTPUT«5===SORRY!5=== Error while compiling /tmp/85iGiBeUAD␤Missing initializer on constant declaration␤at /tmp/85iGiBeUAD:1␤------> 3constant NUMPAD7⏏5-0 = "\x01A";␤»
moritz so including it in .perl output doesn't make sense, because you can't use it recreate the value
azawawi NUMPAD-0 is not a valid kebab case constant identifier? 13:11
moritz azawawi: no
because 0 isn't an identifier
azawawi i see
dalek c: 3f80b9a | moritz++ | doc/Type/Attribute.pod:
Fix a usage example
ShimmerFairy moritz: as mentioned earlier, the problem at least to me is that there isn't really a difference between private and public attributes in Perl 6; the "no private attr by default" thing for .new and such feels arbitrary to me. 13:12
moritz ShimmerFairy: .new is part of the public API. If it set a private attribute, that attribute wouldn't be private
dalek c: 774d9bd | moritz++ | doc/Type/Attribute.pod:
fix another Usage example
 13:13
ShimmerFairy moritz: but what is the difference between public and private attributes? Just sugar for an accessor method.
13:13 aborazmeh joined, aborazmeh left, aborazmeh joined, zakharyas joined
azawawi moritz: so kebab case in Perl 6 has its edge cases to handle when converting legacy code 13:13
moritz ShimmerFairy: no; there's meta data that identifies the attribute as public
azawawi: Perl 6 identifiers follow rules, just like identifiers in any sane language. Of course this has implications. 13:14
ShimmerFairy moritz: but what for? AFAICT it's only for limiting what .new and .bless do by default, and what you do and don't see in .perl . There's no real difference between them. (Unlike in, say, C++, where public and protected attrs are directly usable by a subclass, and privates aren't) 13:15
Zoffix m: class Foo { has $!bar = 42; has $.ber = 43; method moo { $!bar = 54; } }; my $x = Foo.new; $x.moo; say $x.get_value('bar') 13:16
camelia rakudo-moar 32d7b0: OUTPUT«Method 'get_value' not found for invocant of class 'Foo'␤ in block <unit> at /tmp/PUtL2yxMNb:1␤␤»
moritz ShimmerFairy: what for? for avoiding private attributes leaking out
Zoffix How do I fix this error?
ShimmerFairy moritz: my argument is that Perl 6 doesn't really have the notion of public vs. private attributes other than the seemingly arbitrary limits on a couple methods, such as .new
RabidGravy azawawi, from Perl 5 things you have watch for is e.g. print time-rand
moritz Zoffix: get_value is a method of the Attribute class
ShimmerFairy moritz: I can write method foo { $!foo } and suddenly it's like I typed has $.foo . I don't see how that's a useful distinction for limiting what .new and .bless accept by default
moritz Zoffix: which you can get via introspection 13:17
ShimmerFairy: it has the distinction, and it's not arbitrary
Zoffix alright.
moritz ShimmerFairy: it's designed so that as somebody extern to the class, you never interact with private attributes in any way, except if you chose to go cheating (via the MOP)
masak m: sub foo { Empty }; say foo().perl 13:18
camelia rakudo-moar 32d7b0: OUTPUT«()␤»
moritz ShimmerFairy: which is the very definition of "private"
ShimmerFairy moritz: why is it not arbitrary, though? As far as I can tell, P6's "public" attribute is just a private attribute that creates an accessor method for you; you still interact with those attributes in-class via $!foo , for example
moritz ShimmerFairy: well, that quote isn't entirely accurate
ShimmerFairy: and it's not arbitrary, because those restrictions cause an outsider to never see a private attribute 13:19
ShimmerFairy: which is what makes them *private*
ShimmerFairy: what would you consider private?
"the amount of money I have on the bank is private, but if you call .perl on the bank account, you see it anway"?
ShimmerFairy moritz: I don't think you're understanding me. I'm arguing that because of how Perl 6 decided to implement public vs. private attributes, there's no real distinction. All attributes are private, and the has $.foo declaration just creates a getter method for you 13:20
moritz ShimmerFairy: but that's not true
ShimmerFairy: which I'm trying to tell you
ShimmerFairy: a public attribute is a private attribute + accessor + META DATA THAT MARKS IT AS PUBLIC
and it's that meta data that makes .perl and .new respond to them
pmurias moritz: the real distinction is that the private attributes are declare as has $.foo the private ones as $!foo 13:21
ShimmerFairy moritz: what I'm arguing is that the "marks it as public" that lets default .new and so on use it isn't necessary. To me, the distinction should be far stronger than "the compiler says so"
pmurias s/.foo/.foo and the/
ShimmerFairy: it's not necessary
moritz ShimmerFairy: it is necesary
ShimmerFairy: otherwise, privates would leak out 13:22
you'd have to overwrite your own .perl and .new just to prevent privates from leaking out
which wouldn't make them very private
is that really that hard to understand?
pmurias moritz: .perl is just for debugging
moritz pmurias: no
ShimmerFairy moritz: every time you say "privates would leak out", I feel like you don't get me. I'm saying that attributes are all private, what makes them public is just the getter method.
moritz ShimmerFairy: and I'm telling you again and again that it's not true 13:23
pmurias ShimmerFairy: what makes them public is the getter method and the magic "public" bit
moritz ShimmerFairy: and you haven't yet presented a coherent design where that could be true, and privates wouldn't accidentally leak out
pmurias ShimmerFairy: we could only have private attributes but it would be bloody annoying to write the getters
13:24 eliasr joined
moritz pmurias: .perl is specified to round-trip under EVAL, not be useful for debugging 13:24
ShimmerFairy moritz: to me, "leak out" means "an external user can see it". With my "all attrs are really private" idea, that doesn't happen unless you specifically write a method that gives it. 13:25
pmurias what uses besides debugging does it have?
moritz pmurias: serialization
ShimmerFairy pmurias: to be clear, I'm not advocating abolishing has $.foo , just the part where that magically makes it ok for default .new, .bless, and .perl to see it :)
moritz ShimmerFairy: ok, please formulate your "all attrs are really private" proposal in a gist or so; including answers to questions like "how does the default constructor work?" 13:26
pmurias ShimmerFairy: that part is just useful
13:26 BenGoldberg left
pmurias moritz: is there a way to safely unpack .perl output? 13:27
ShimmerFairy moritz: the default constructor works like it does in NQP, or in other words takes all attributes by default. I don't know why you're so insistent on not understanding my point (ooc, have you seen the backlog where masak brought it up?)
moritz ShimmerFairy: if it respects all attributes, then those attributes aren't really private
ShimmerFairy: which *directly* contradicts your "all attributes are private" approach 13:28
pmurias: no; but there are many use cases where you don't have to 13:29
azawawi moritz: i see. thanks :)
ShimmerFairy moritz: when I talk about private attributes, I'm not talking about the strict, language-agnostic term, but I'm referring to specifically $!foo . In other words, "All $.foo are really $!foo with a method foo created for you"
moritz ShimmerFairy: ok, so then how would you make language-agnostic private attributes with your proposal?
ShimmerFairy: attributes that the outside never sees?
azawawi moritz: solved it by moving it into a hash github.com/azawawi/perl6-selenium-...r/Keys.pm6 :) 13:30
moritz azawawi: looks like a sensible choice :-) 13:31
13:31 _28_ria left
ShimmerFairy moritz: All I know is that it's more often annoying than helpful for me that .new won't accept private attributes. To me, "public" means "the user of the object can mess with these variables", where "private" is more "the user of the object isn't trusted to safely mess with these directly" 13:31
13:31 Hor|zon joined, ngie left
llfourn moritz: do you coulnt setting the value of a private attribute at object contruction time to be seeing it? 13:31
13:31 rindolf left
llfourn count* 13:32
moritz ShimmerFairy: if the behavior of private attributes isn't useful to you, just use public attributes
ShimmerFairy So for .new, there are cases where I want the user to set an initial value, but otherwise I need to control retrieval and/or modification from the class.
moritz llfourn: of course
ShimmerFairy: there are solutions to that in current Perl 6 too
llfourn mortiz: ok just checking as I think this is the crux of your disagreement
13:33 _28_ria joined
RabidGravy ShimmerFairy, in that case I'd consider making a BUILD that pretended to do that 13:33
ShimmerFairy moritz: And I'd disagree with that. To me, public/private is more about usage, and construction is a different process that happens before usage can begin.
moritz llfourn: if the outside can interact with it by default, it's not private. It's quite easy, IMHO :-)
ShimmerFairy: well, if you have your own private definition of "private", of course it won't align with the default of how private attributes work
llfourn mortiz: being able to set it at object contruction time and then never set or get it again.... some might consider that private?
ShimmerFairy RabidGravy: the issue here is that having to write a new BUILD just because one attribute of yours doesn't fit neatly into either 'public' or 'private' as descriptors is LTA :)
llfourn mortiz: but I see your point :) 13:34
colomon or to put it another way, you should be able to change the name of a private attribute without affecting any code outside of your class.
moritz colomon: thanks, that's a nice way to put it
RabidGravy :)
13:35 Alina-malina left
ShimmerFairy RabidGravy: actually, the worst is when you write a .new that does positional stuff (thus you don't use attribute names as a user), and you still have to write your own BUILD, even though "you can change the name of a private attribute without affecting outside code" with that kind of .new :) 13:35
13:36 Hor|zon left
RabidGravy in that case, if this was going to happen more than once in some code I intended to write I would make a role that provided a BUILD that would provide for the initialisation of otherwise private attribuates that are marked with some trait 13:37
:)
Zoffix In Perl 5's Moo, this type of stuff is done with public attributes that are "ro" (read only). Can't modify them after initialization. Maybe Perl 6 could have that, if it does not already. <ShimmerFairy> where "private" is more "the user of the object isn't trusted to safely mess with these directly"
Or rather 'rwp' would be more appropriate: read only to the outside / read-write inside. 13:38
13:38 ecocode joined
ShimmerFairy Zoffix: that's how public attributes are by default (read-only). I suppose my issue is just that I don't call them public, and being forced to do so feels like a mistake 13:38
colomon what RabidGravy seems to be asking for is public attributes without accessors.
which is pretty trivial to do by hand for each such attribute.
and I suspect could easily be written up as a trait. 13:39
Zoffix ShimmerFairy, what do you call something that is invisible from the outside?
llfourn ShimmerFairy: so what you want is an attribute that is available to bless and default new but has no accessors?
RabidGravy *I'm* not asking for anything, I'm suggesting how one would achieve that without changing Perl
ShimmerFairy Basically, being forced to write your own BUILD just because you want to organize things lightly differently feels a bit too Boost-like for my tastes.
colomon RabidGravy: sorry, meant ShimmerFairy
RabidGravy :)
ShimmerFairy Keep in mind that masak is the one that brought it up, I'm generally fine with how things are now. I'm just arguing that the distinction doesn't seem significant enough, and thus the limits on default .new and .bless feel arbitrary 13:40
13:40 abaugher joined
moritz ShimmerFairy: can you name a few use cases of attributes that you want to be set from the outside during construction, but that shouldn't be readable once constructed? 13:40
pmurias ShimmerFairy: it's not arbitary, it's the only way that makes sense
ShimmerFairy (by "too Boost-like", I mean that the Boost library has too much of that "Want it lightly different? Now you have to write *ALL* the code!") 13:41
pmurias ShimmerFairy: just import some module which give you the no-access trait
RabidGravy yeah which is what I am suggesting 13:42
ShimmerFairy SecureRandomNumbers.new(:seed(42)) -- let the user set the seed, but I'm guessing you probably don't want that freely available in a more secure environment :)
moritz ShimmerFairy: but Perl 6 isn't a more secure environment
ShimmerFairy: so that feels like a very constructed argument 13:43
ShimmerFairy moritz: they why are you carping about the need for private variables to not leak in the first place?
moritz private attributes can *never* be a security feature within the same process
ShimmerFairy: because I want to avoid things becoming API that aren't meant to be API 13:44
ShimmerFairy: it's the difference between avoiding leaks to avoid environmental pollution, and avoiding leaks as a security feature
ShimmerFairy moritz: I realize they're different areas, but I don't think it's very fair for you to go on about security in one respect (keeping API stable), and then criticize an example referencing security in another respect 13:46
RabidGravy yeah, data hygiene is different to data security
moritz ShimmerFairy: I've never today talked about security before
Zoffix ShimmerFairy, but how is that example secure? The user has the number ("42"), so restricting their reading it via an accessor is not much of a feature. What am I missing? 13:47
moritz ShimmerFairy: more about safety, really. Mentally mixing those two isn't a good idea
RabidGravy you do all realise in the time that this discussion has taken to get nowhere some proof of concept as I outlined above could have been written? ;-) 13:48
Zoffix :)
moritz RabidGravy: I asked for an example use case because I thought the discussion was going nowhere
RabidGravy: and so far got only one constructed, unapplicable example
13:49 azawawi left
llfourn enjoyed the discussion and understood both sides 13:49
moritz TBH in the past I've sometimes wished for this feature too, but upon closer inspection, none of my use cases has held up 13:50
so now I'm looking for other people's use cases
13:50 SqrtNegInf joined
llfourn I also feel that having to write special methods to use constructor arguments to influence or set default values for $! attributes is a bit annoying 13:54
which is why I hardly ever use them
13:56 kjs_ left
llfourn something like class A { has $!a is contructor-param; } so $!a could be set by bless would be nice. 13:56
RabidGravy which could be done with a user supplied role and a trait 13:57
llfourn RabidGravy: yep probs best to test out the idea in that respect at least 13:58
RabidGravy not everything has to be in the language 14:00
llfourn on a similar subject is there a way to set default values for attributes from other attributes like: class A { has $.a = 'default'; has $.b = $.a; }
like in Moose with lazy => 1
moritz m: class A { has $.a = 'default'; has $.b = $.a; }; say A.new.b 14:02
camelia rakudo-moar 32d7b0: OUTPUT«5===SORRY!5=== Error while compiling /tmp/jAIrABouZT␤Virtual call $.a may not be used on partially constructed objects␤at /tmp/jAIrABouZT:1␤------> 3s A { has $.a = 'default'; has $.b = $.a7⏏5; }; say A.new.b␤ expecting any of:␤ …»
moritz m: class A { has $.a = 'default'; has $.b = self.a; }; say A.new.b
camelia rakudo-moar 32d7b0: OUTPUT«default␤»
14:02 aborazmeh left
moritz m: class A { has $.a = 'default'; has $.b = self.a; }; say A.new(a => 'blurb').b 14:02
camelia rakudo-moar 32d7b0: OUTPUT«blurb␤»
llfourn mortiz: awesome :)
llfourn goes to document this feature
RabidGravy has that always worked? I thought I tried that before and it complained about not being able to use self there 14:03
moritz llfourn: btw if you tab-complete my nick name after the first two or three characters, you're unlikely to misspell it every time :-)
RabidGravy: it has worked for years, though it used to work with $.a syntax instead of self.a 14:04
14:04 ecocode left
RabidGravy Hmm weird, must be a false memory or I was trying to do something more complicated 14:05
llfourn moritz: thanks!
14:07 kjs_ joined
dalek ast: da8c0ee | usev6++ | S11-modules/require.t:
Add "use lib '.';" for test that needs it
 14:08
llfourn moritz: is it meant to work wtih $.a? 14:11
dalek kudo/nom: 23ccdc2 | lizmat++ | src/core/Supply.pm:
Fix an omission in emission tracking
pink_mist lizmat++ I bet you enjoyed writing that commit message :P 14:12
lizmat :-) yes I did :-)
RabidGravy the VW commit
lizmat been binge watching Bones to get in the mood
RabidGravy: yes, and that :-)
afk again& 14:13
dalek ast: 85edc25 | usev6++ | S32-num/power.t:
(Re-)fudge test for big exponent on rakudo.moar

test was unfudged with b0f743e59e but is failing for a while now		 14:18
grondilu_ m: say &[>+<] 14:20
camelia rakudo-moar 32d7b0: OUTPUT«5===SORRY!5=== Error while compiling /tmp/Jy9OiphBYb␤Unable to parse expression in infix noun; couldn't find final ']' ␤at /tmp/Jy9OiphBYb:1␤------> 3say &[>7⏏5+<]␤»
grondilu_ was wondering what is this equiv('>+<') mentioned above 14:21
14:22 grondilu_ is now known as grondilu, Akagi201 joined 14:23 grondilu left, grondilu joined, b2gills left 14:24 Sgeo left, Sgeo joined
dalek c: 9793f64 | (Lloyd Fournier)++ | doc/Language/objects.pod:
mixed in a self into the attribute default example
 14:31
14:32 kjs_ left, Hor|zon joined, pmurias_ joined 14:34 Ven_ joined
masak m: say Seq.^roles 14:34
camelia rakudo-moar 32d7b0: OUTPUT«Method 'iterator' must be implemented by PositionalBindFailover because it is required by a role␤ in any compose_method_table at gen/moar/m-Metamodel.nqp:2661␤ in any apply at gen/moar/m-Metamodel.nqp:2671␤ in any compose at gen/moar/m-Metamodel.n…»
masak m: say Array.^roles
camelia rakudo-moar 32d7b0: OUTPUT«Method 'iterator' must be implemented by Iterable because it is required by a role␤ in any compose_method_table at gen/moar/m-Metamodel.nqp:2661␤ in any apply at gen/moar/m-Metamodel.nqp:2671␤ in any compose at gen/moar/m-Metamodel.nqp:2807␤ in…»
14:34 Ven_ left
masak is that... intentional? 14:34
14:35 Ven_ joined, pmurias left 14:36 Alina-malina joined, vendethiel left, Hor|zon left 14:38 khw joined
nine m: say Iterable 14:40
camelia rakudo-moar 32d7b0: OUTPUT«Method 'iterator' must be implemented by Iterable because it is required by a role␤ in any compose_method_table at gen/moar/m-Metamodel.nqp:2661␤ in any apply at gen/moar/m-Metamodel.nqp:2671␤ in any compose at gen/moar/m-Metamodel.nqp:2807␤ in…»
nine masak: I guess it's the same for all abstract roles
m: say role Foo { method foo { ... } }
camelia rakudo-moar 32d7b0: OUTPUT«Method 'foo' must be implemented by Foo because it is required by a role␤ in any compose_method_table at gen/moar/m-Metamodel.nqp:2661␤ in any apply at gen/moar/m-Metamodel.nqp:2671␤ in any compose at gen/moar/m-Metamodel.nqp:2807␤ in any make_…»
masak nine: something feels wrong about that, though. 14:41
nine: I wasn't asking for an implementation of the role, just for its name.
nine: what's the idiomatic way to ask for the roles implemented by Array? :)
psch_ it feels consistent but useless, yes 14:42
14:42 psch_ is now known as psch
nine masak: you sure? roles returns the type objects for the roles, not only their names 14:42
psch the type object puns, which is the consistent part 14:43
masak ok, then I would like to be able to hold a type object without it exploding.
nine So yes, consistent, but certainly useless :)
m: my $iterable = Iterable; say $iterable.^name;
camelia rakudo-moar 32d7b0: OUTPUT«Iterable␤»
nine masak: you can ;) 14:44
psch m: say Array.^roles>>.^name
camelia rakudo-moar 32d7b0: OUTPUT«List␤»
masak meh.
nine So it's just gist that could be improved
masak yeah
m: Iterable.gist
camelia rakudo-moar 32d7b0: OUTPUT«Method 'iterator' must be implemented by Iterable because it is required by a role␤ in any compose_method_table at gen/moar/m-Metamodel.nqp:2661␤ in any apply at gen/moar/m-Metamodel.nqp:2671␤ in any compose at gen/moar/m-Metamodel.nqp:2807␤ in…»
psch ehh, can't .^ down the list... 14:45
nine Though I would not know how, since gist can only work on the punned type object
masak I think #perl6 has had this discussion before :)
and I don't have a good answer either, I just thinks it's not optimal the way it is
awwaiid In Rakudo, EVAL can take 'context', but I don't see that in the design docs. I'm exploring what all 'context' gives me (like... I can see lexical vars but callframe() isn't affected by the passed context). Is EVAL "$fo", context => ... even a thing I should be doing? 14:46
masak 'night, #perl6
Zoffix \o
colomon masak: where are you?
nine Good night masak
colomon: on .cn time
colomon yow. 14:47
awwaiid gist.github.com/awwaiid/f537ccf5d25cf4ad7cf7 -- EVAL w/ context not affecting callframe() example 14:48
14:50 Ven_ left
CurtisOvidPoe This concurrent Perl 6 code appears to be blocking. Can someone help me understand what I did wrong? gist.github.com/Ovid/06ac6e9e0187e84a9589 14:51
timotimo CurtisOvidPoe: you're waiting for both the timeout *and* the get requests to complete 14:53
"allof" seems to be the wrong combiner in the last line
CurtisOvidPoe Rerunning now with ‘anyof’ 14:54
timotimo you probably have to anyof(timeout, allof(results))
14:54 Ven_ joined, larion joined
RabidGravy yeah 14:55
though possibly allof() the requests then anyof that and the timeout 14:56
CurtisOvidPoe That seems like I can’t discard individual URLs which exceed the timeout. I’m guessing I need to construct individividual promise objects and have a Promise.in($timeout) for each?
timotimo RabidGravy: isn't that exactly what i suggested? 14:57
CurtisOvidPoe Also, changing that last line to “await @promises” blocks and prints nothing.
RabidGravy timotimo, oh so it was, distracted
timotimo CurtisOvidPoe: my initial idea would be to litter the code with "say" statements to see where it ends up not doing anything 14:58
like, is the LWP::Simple.get call taking forever for some reason?
CurtisOvidPoe Well, it works in serial. Those calls take about a minute and a half if run one after another. 14:59
timotimo or is it just that the poor HTML parser is woefully insufficient to handle a gigantic site like yahoo.com?
definitely weird. 15:00
CurtisOvidPoe It appears that not a single one is returning after the .get
timotimo thenperhaps LWP::Simple has a problem with parallelism
can you try HTTP::UserAgent instead?
i think it also has a submodule that you can use that'll give you the exact same API as LWP::Simple 15:01
15:02 Akagi201 left
CurtisOvidPoe Reducing this down to fetching a single URL (and thus creating only one promise), also fails. 15:02
15:03 kjs_ joined, tokuhirom_ left
timotimo that'? kind of bad 15:03
15:03 larion left, larion joined 15:04 larion left, larion joined
nine timotimo: have you had a look at p6for already? Is it useful to you? 15:04
CurtisOvidPoe Just running this hangs on my box: perl6 -e 'use LWP::Simple;await start { LWP::Simple.get("www.rakudo.org/") }' 15:05
timotimo i've had a brief look and it looks helpful indeed :)
nine timotimo: yeah!
timotimo it also brought up the ol' problem of "what to optimize in what order"
15:06 Ven_ left 15:09 b2gills joined 15:10 Ven joined
CurtisOvidPoe This also hangs, but it works great in the debugger :( perl6 -MHTTP::UserAgent -e 'my $ua = HTTP::UserAgent.new; await start { say $ua.get("www.rakudo.org/").content }' 15:13
timotimo well, god damn it. 15:14
CurtisOvidPoe I know.
timotimo hm
15:14 bjz left
timotimo we could try setting a scheduler up that ignores multi-threading 15:14
as in:
m: my $*SCHEDULER = CurrentThreadScheduler.new(); await start { say "hi" };
camelia rakudo-moar 23ccdc: OUTPUT«hi␤»
CurtisOvidPoe Nope. Now it’s hanging in the debugger. 15:16
timotimo o_O
why would it do that >:(
CurtisOvidPoe is confused
timotimo what about without debugger?
RabidGravy too, most of the time
15:16 BenGoldberg joined 15:18 bjz joined
CurtisOvidPoe I meant the REPL, sorry. 15:18
timotimo oh
CurtisOvidPoe And now it *works* in the REPL. WTF???
timotimo wellllll.... 15:19
it could very well be because it tries to print the value of the last statement?
15:19 pmurias_ is now known as pmurias
timotimo if not, the REPL has had weirdness issues in the past, for example before the commit that made the := operator work 15:20
15:21 s_kilk joined
CurtisOvidPoe Hmm, this might be reproducible. Hold on. 15:22
Got it, but it’s a weird bug … gist.github.com/Ovid/4cb42aefd126fde16000 15:24
You have to call $ua.get() at least once before the await start {} and it works
timotimo :o
CurtisOvidPoe Can someone else reproduce the error on their box? 15:25
15:25 diana_olhovik_ joined
timotimo can't try right now ;( 15:25
llfourn CurtisOvidPoe: I actually tried to install HTTP::UserAgent just now but it hangs in one of the tests :(
CurtisOvidPoe So how do I properly submit a rakudobug?
timotimo just a mail to [email@hidden.address] 15:26
CurtisOvidPoe Thanks for trying, llfourn
llfourn You also put [BUG] in the title to [email@hidden.address] I think
RabidGravy CurtisOvidPoe, is that the same for both LWP::Simple *and* HTTP::UserAgent 15:27
CurtisOvidPoe Yes.
Wondering if it’s socket related?
RabidGravy yeah possibly some evil interation between the event loop and libuv's socket thingies 15:28
ugexe but the socket isn't created until the `get` is called 15:29
15:30 bjz_ joined, Peter_R joined
CurtisOvidPoe rakudobug filed. 15:30
RabidGravy yeah, but if there is some thread context thing in the socket implementation
15:31 bjz left
CurtisOvidPoe Crap. I should have mentioned it’s on OS X. Forgot to include my version number, too. 15:31
RabidGravy It does it on Linux too FWIW
15:33 Hor|zon joined
CurtisOvidPoe Cheers, RabidGravy. 15:33
rt.perl.org/Ticket/Display.html?id...d8068e1539
ugexe never leaves the Planned state
15:33 gtodd left 15:34 ^elyse^ left 15:36 Guest77431 left
Zoffix CurtisOvidPoe, if you're just looking for some UA, HTTP::Tinyish seems to work: github.com/shoichikaji/perl6-HTTP-Tinyish/ 15:36
2.6s (though that's mostly startup and not get): fpaste.scsys.co.uk/500400
RabidGravy I guess they are all similarly afflicted
ooh 15:37
Zoffix Oh. I think this one just shells out to `curl' 15:38
15:38 Hor|zon left
ugexe yep 15:38
CurtisOvidPoe On the plus side, every time I update a Perl 6 talk, I get fewer errors in Perl 6 :)
timotimo i'm getting kind of nervous about our lack for advanced sockety stuff and being so close to christmas 15:39
udp sockets, for example
RabidGravy and/or local sockets 15:40
CurtisOvidPoe Just drop it from the spec ;)
timotimo just let people build that with NativeCall? :)
RabidGravy I've toyed with that idea
CurtisOvidPoe waits for all the NativeCall concurrency bugs ...
timotimo someone already did that thing, RabidGravy
blog.64p.org/entry/2015/09/15/192721 15:41
btyler sockets (and async sockets) can use any spare love that's available, for sure
RabidGravy the libshout native binding works fine in a concurrent kind of thing
15:41 xfix left 15:43 Ven left 15:44 Ven joined
timotimo i don't think nativecall has big problems with concurrency 15:45
sadly, libuv isn't very friendly towards how we do multiple threads 15:46
RabidGravy yeah it'll mostly be libraries themselves that cause the problem
timotimo but ideally, our synchronous I/O would bypass libuv completely
15:47 vendethiel joined
timotimo that'd also remove the "you can't read data on a socket when your code runs on a different thread from the one that created it" 15:49
which is just ugh
RabidGravy is IO::Socket::Async thus aflicted? 15:50
timotimo i don't think so, as it does all its IO things on the extra thread anyway
as in, moar builds a thread dedicated to doing all its async dispatching when the first async i/o thing appears 15:51
BenGoldberg m: my $sock = Raw::Socket::INET.new( :host<localhost>, :port<9000> ); my $b = Buf.new(8); say await start { say $sock.recv($b, 8, 0); 42 }; 15:58
camelia rakudo-moar 23ccdc: OUTPUT«Could not find symbol '&INET'␤ in block <unit> at /tmp/lNxgIKSirP:1␤␤Actually thrown at:␤ in block <unit> at /tmp/lNxgIKSirP:1␤␤»
15:58 rindolf joined
BenGoldberg Odd error message, that. 15:58
15:59 ^elyse^ joined
psch m: say A::B::C 15:59
camelia rakudo-moar 23ccdc: OUTPUT«Could not find symbol '&C'␤ in block <unit> at /tmp/SaBVCqy9xE:1␤␤Actually thrown at:␤ in block <unit> at /tmp/SaBVCqy9xE:1␤␤»
psch not sure if that should complain about every not-found level 16:00
or if it could, even
16:02 s_kilk left, Akagi201 joined 16:03 riatre_ is now known as riatre, tokuhirom_ joined 16:05 Ven left 16:08 tokuhirom_ left, tokuhiro_ joined 16:13 Ven joined, tokuhiro_ left 16:15 Akagi201 left 16:16 rurban left 16:18 rurban joined 16:21 Ven left 16:24 weihan left, Zoffix left 16:25 ^elyse^ left 16:26 Ven joined
moritz the error message would be better if it complained about the full name, IMHO 16:30
cxreg2 I've been thinking about blogs.perl.org/users/pawel_bbkr_pab...read.html. Maybe you've already moved on from this issue, but might it not be possible to simply stop parallelizing and do work in serial when the threadpool is (nearly) exhausted? 16:31
seems like maybe not something every programmer should need to be aware of; at least then you just get slow instead of crashing 16:32
16:34 Hor|zon joined, cxreg2 is now known as cxreg 16:35 Ven left 16:36 ptarmigan999 joined 16:37 gtodd joined
cxreg I'm referring to the scheduling work performed by start/await, not something the program would do explicitly 16:37
16:37 Ven joined 16:38 ^elyse^ joined 16:39 Hor|zon left
ptarmigan999 Hi - I was just trying out the XML example from RosettaCode and it failed, I think because XML has added text nodes to @nodes. I was trying to restrict the nodes returned to just the XML::Element nodes. 16:39
The existing code used: say .<Name> for $xml.nodes.grep(*.name eq 'Student') 16:40
I've tried varlations like say .<Name> for $xml.nodes.grep((*.WHAT.perl eq 'XML::Element') and (*.name eq 'Student')) and
$xml.nodes.grep(*.WHAT.perl eq 'XML::Element').grep(.name eq 'Student') - with and without the '*' - what's the correct syntax to use? 16:41
colomon ptarmigan999: If you have two *s in whatever code, you’re declaring a block that takes two arguments. 16:42
you could do it with two grep statements: .grep(*.WHAT.perl eq 'XML::Element’).grep(*.name eq 'Student’)) 16:43
or you could rewrite it with brackets: .grep({ $_.WHAT.perl eq ‘XML::Element’ && $_.name eq ‘Student’)
whoops, forgot closing bracket 16:44
or you could rewrite it with brackets: .grep({ $_.WHAT.perl eq ‘XML::Element’ && $_.name eq ‘Student’})
ptarmigan999 Ah ok - I tried it with '*' instead and was getting 'Cannot use Bool as Matcher with '.grep'. Did you mean to use $_ inside a block?' :D 16:45
geekosaur would something like $^a work there also?
colomon hmmm, seems like a less than awesome error message.
yes $^a would work, but it would also need { } 16:46
so there’s no real advantage to using $^a instead of $_ if you only want one arg
16:46 gtodd left
psch m: say ^10 .grep: (* %% 2 && * > 4) 16:47
camelia rakudo-moar 23ccdc: OUTPUT«(5 6 7 8 9)␤»
colomon (unless you also need $_ for something else in the block!)
psch m: say ^10 .grep: { $^a %% 2 && $a > 4}
camelia rakudo-moar 23ccdc: OUTPUT«(6 8)␤»
psch curious how it uses the second Whatever 16:48
colomon psch: think of the simpler * + *
psch m: say (* + *)(2, -5)
camelia rakudo-moar 23ccdc: OUTPUT«-3␤»
psch vOv
stupid transitive property :)
colomon that said, I have no idea how (* %% 2 && * > 4) is working up there.
psch wait, that's not transitive, is it..? 16:49
no, it's cumutative..
colomon m: (^10).grep(*.say && *.say)
camelia rakudo-moar 23ccdc: OUTPUT«5===SORRY!5=== Error while compiling /tmp/mYAs7o_UzQ␤Unable to parse expression in argument list; couldn't find final ')' ␤at /tmp/mYAs7o_UzQ:1␤------> 3(^10).grep(*.say && *.say7⏏5)␤ expecting any of:␤ infix␤ infix …»
timotimo m: say ^10.grep: (*.say && *.say)
camelia rakudo-moar 23ccdc: OUTPUT«Potential difficulties:␤ Precedence of ^ is looser than method call; please parenthesize␤ at /tmp/e_AGy8uUAX:1␤ ------> 3say ^107⏏5.grep: (*.say && *.say)␤10␤0..^1␤»
timotimo m: say ^10 .grep: (*.say && *.say)
camelia rakudo-moar 23ccdc: OUTPUT«0␤1␤2␤3␤4␤5␤6␤7␤8␤9␤(0 1 2 3 4 5 6 7 8 9)␤»
psch m: say (* / *)(0, 5) 16:50
timotimo m: ^10 .grep: (*.say && *.say)
camelia rakudo-moar 23ccdc: OUTPUT«0␤»
rakudo-moar 23ccdc: OUTPUT«0␤1␤2␤3␤4␤5␤6␤7␤8␤9␤»
colomon wait a minute...
m: say (^10).grep(* + * > 5) 16:51
camelia rakudo-moar 23ccdc: OUTPUT«((4 5) (6 7) (8 9))␤»
colomon m: say (^10).grep(* + * > 9)
camelia rakudo-moar 23ccdc: OUTPUT«((6 7) (8 9))␤»
ptarmigan999 I tried: say .<Name> for $xml.nodes.grep(*.WHAT.perl eq 'XML::Element’).grep(*.name eq 'Student’) and got:Missing infix inside S at /disk/home/david/development/perl6/./xml_example.p6:17 ------> erl eq 'XML::Element’).grep(*.name eq 'S⏏tudent’) expecting any of: infix infix stopper
psch ptarmigan999: that looks like a bad quote character
*.WHAT.perl eq 'XML::Element’ 16:52
ptarmigan999 sorry - cut'n paste error
timotimo m: say 'Student’
camelia rakudo-moar 23ccdc: OUTPUT«5===SORRY!5=== Error while compiling /tmp/p2U4MGUrsD␤Unable to parse expression in single quotes; couldn't find final "'" ␤at /tmp/p2U4MGUrsD:1␤------> 3say 'Student’7⏏5<EOL>␤ expecting any of:␤ argument list␤ singl…»
timotimo m: say 'Student'
camelia rakudo-moar 23ccdc: OUTPUT«Student␤»
timotimo m: say ‘Student’
camelia rakudo-moar 23ccdc: OUTPUT«Student␤»
timotimo m: say ‚Student‘
camelia rakudo-moar 23ccdc: OUTPUT«Student␤»
timotimo the smartest of quotes on these students
16:53 pmurias_ joined
colomon ptarmigan999: cutting and pasting my code on here is dangerous, my IRC client loves to substitute the characters it thinks I wanted instead of what I actually typed. 16:53
timotimo that string started at 'XML::Element and went all the way to .name eq '
psch m: say uniname("‚")
camelia rakudo-moar 23ccdc: OUTPUT«SINGLE LOW-9 QUOTATION MARK␤»
psch m: say uniname(",")
camelia rakudo-moar 23ccdc: OUTPUT«COMMA␤»
psch they look the same here vOv
psch might need a better font...
timotimo :D
ptarmigan999 corrected quotes and back to 'Bool as Matcher' error
timotimo oh! 16:55
*.WHAT doesn't work
you really want to just use "XML::Element" as the matcher
m: say *.WHAT
camelia rakudo-moar 23ccdc: OUTPUT«(Whatever)␤»
timotimo m: say (*.WHAT eq 'String')
camelia rakudo-moar 23ccdc: OUTPUT«Use of uninitialized value of type Whatever in string context␤Any of .^name, .perl, .gist, or .say can stringify undefined things, if needed. in block <unit> at /tmp/C5Qw4JHLbO:1␤False␤»
colomon oh, sorry, I should have caught that
ptarmigan999 let me retype from scratch and double check
timotimo type checking is best done with ~~
psch m: say (5, "5", 10, "10").grep(Int).grep(* > 5)
camelia rakudo-moar 23ccdc: OUTPUT«(10)␤»
timotimo and grep does ~~ "under the hood" for you
colomon Think you want * ~~ XML::Element
16:56 pmurias left
timotimo colomon: that's the exact same as just XML::Element in this case 16:56
m: say (* ~~ Bool)
camelia rakudo-moar 23ccdc: OUTPUT«WhateverCode.new␤»
colomon timotimo: good point. :)
timotimo m: say (True, False, "FileNotFound").grep(* ~~ Bool).
camelia rakudo-moar 23ccdc: OUTPUT«5===SORRY!5=== Error while compiling /tmp/_RhMv21qge␤Missing required term after infix␤at /tmp/_RhMv21qge:1␤------> 3 False, "FileNotFound").grep(* ~~ Bool).7⏏5<EOL>␤ expecting any of:␤ dotty method or postfix␤»
timotimo m: say (True, False, "FileNotFound").grep(* ~~ Bool)
camelia rakudo-moar 23ccdc: OUTPUT«(True False)␤»
timotimo m: say (True, False, "FileNotFound").grep(Bool)
camelia rakudo-moar 23ccdc: OUTPUT«(True False)␤»
psch hrm 16:57
so i added debug output to Ops.setobjsc to print the typeName for non-concrete objects during setting compilation 16:58
and apparently that gets called three times for Attribute
and 11 times for Attribute:D 16:59
but hey, it also gets called lots for e.g. Int
but it shouldn't get called for non-concrete objects at all, from what i remember jnthn++ saying on this
which i had some kind of actual, high-level understanding what SCs do and how they work, instead of just poking at things... :P 17:00
well, and low-level too of course 17:01
timotimo hm, could that be about "compile time evaluate" and WVals perhaps?
psch timotimo: maybe..? setobjsc gets called from HLL::World.add_object, which does get called in quite a few compile-time value-ish things in Perl6::World... 17:03
hm, maybe the isconcrete check doesn't actually do enough in setobjsc..? 17:04
i mean, Perl6::World.scalar_wrap adds the object to the SC and then binds a value
does that already make the Scalar concrete? 17:05
ptarmigan999 colomon/timotimo: thanks. ~~ XML::Element works :)
17:06 cognominal left, softmoth joined
leont I'm dealing with an exception from a promise that isn't pointing to a useful line 17:07
timotimo perhaps you want to decont before you isconcrete?
leont How do I get the information that I actually want?
17:07 llfourn left
leont As in, actual line number and more information 17:07
timotimo leont: sometimes running perl6 with --ll-exception gives you a bit more 17:08
17:08 ptarmigan999 left
leont Tried that already, it wan't helpful. It gave the $promise.result location instead of the actual source 17:08
timotimo oh, that means we're losing the original exception there? 17:10
that place would ideally have only a rethrow-like thing in there
leont I think so, I've seen this happen before 17:11
17:12 Akagi201 joined
leont Right now all I have to go on is "Too many positionals passed; expected 2 arguments but got 3", which is definitely not from that line 17:12
timotimo aye
as a work-around, you can put a CATCH { .say } into the start block 17:13
but i realize that sucks
17:18 Akagi201 left
leont Yeah that does help 17:19
timotimo oops. rebuilding nqp now even though the only changes were in js-related parts 17:21
psch timotimo: adding a decont gives me the same numbers 17:24
17:24 firstdayonthejob left
timotimo hum. 17:25
17:35 Hor|zon joined 17:37 Ven left 17:38 Ven joined, softmoth left 17:39 Hor|zon left, zakharyas left 17:40 tokuhirom_ joined 17:44 zakharyas joined, tokuhirom_ left 17:49 kaare_ left
pmurias_ psch: what do you want to know about SCs? 17:50
17:50 pmurias_ is now known as pmurias, Ven left
pmurias psch: I reimplemented a lot of the serialization stuff for the js backend so I might be able to help 17:51
psch pmurias: that's a hard question...
pmurias: the background is that we have a problem on jvm, with repossession apparently
pmurias: for some reason, contrary to what moar does, the type object for Attribute on jvm gets repossessed to CORE.setting, from BOOTSTRAP.nqp 17:52
pmurias: so while trying to figure out why and how that happens i'm trying to learn about what an SC actually does and how it does that 17:53
pmurias: what i get from the jvm impl, any SC is mostly just a data holder for "where can i find this object" 17:54
17:54 Ven joined
pmurias once serialized the SC holds the state of a bunch of objects at the point of serialization 17:55
repossession allows changing the object and then storing it somewhere else
(I haven't yet reimplemented repossesion so I'm that familiar with the tiny details of that) 17:56
psch: repossesion happens when the object is change (the objects are transfered to the currently compiled sc) 18:01
psch: this can be disabled (which is done when doing stuff like setting temporary caches at stuff)
colomon CurtisOvidPoe: I just built panda without issue on OS X. 18:02
CurtisOvidPoe colomon: I’ve found that if fails if it’s 18:03
18:03 margeas left
CurtisOvidPoe PERL5LIB=lib rakudobrew build-panda 18:03
pmurias psch: you could try hooking into repossesObject and printing when a type object for Attribute gets repossesed (and from which sc to which)
psch: in SerializationContext.java 18:04
colomon CurtisOvidPoe: ah. I don’t think I have PERL5LIB set to anything
psch pmurias: but to check if i have an Attribute or not i need the TC, don't i?
CurtisOvidPoe Yeah, it’s a handy shortcut I use, but it bit me here.
psch (well, at least Ops.typeName needs the TC, and i don't know how to identify a specific type otherwise...) 18:05
18:05 mrsolo joined 18:06 Ven left, kaare_ joined 18:07 eliasr left, Ven joined
pmurias psch: hook into scwbObject 18:08
it's the only place that calls repossesObject and there is a thread context there
psch pmurias: yeah, i did that a few days ago. it doesn't show any calls for Attribute where isconcrete isn't 1
DrForr seen BooK
psch pmurias: which means there's no actual repossession of the Attribute type object happening, afaiu 18:09
pmurias the setobjsc op is called on the Attribute type object? 18:10
psch pmurias: yeah, that happens, but not only for Attribute but also for e.g. Int
18:10 tokuhiro_ joined
psch well, it happens for pretty much everything, from the looks of it... 18:11
18:11 diana_olhovik_ left
psch i'll add printing the description there, that probably adds some information... :) 18:12
pmurias psch: it seems to be that if this is done on an object that is already part of an sc bad things will happen 18:13
18:14 Akagi201 joined 18:15 rbk_ joined
pmurias psch: and there is an add_object_if_no_sc in places where we aren't sure 18:15
(in Perl6::World)
18:15 tokuhiro_ left, rbk_ left 18:18 Ven left 18:19 Akagi201 left 18:22 diana_olhovik joined
psch pmurias: fwiw, the first time setobjsc gets called on Attribute it sets the SC to BOOTSTRAP, but there's two more calls that set it to CORE 18:25
18:25 gtodd joined
psch but as mentioned, similar things happen for pretty much every other type object 18:25
18:27 diana_olhovik left
colomon Could not download module metadata: No such method 'host' for invocant of type 'HTTP::Request' 18:31
in method update at /home/colomon/.rakudobrew/moar-nom/install/share/perl6/site/lib/Panda/Ecosystem.pm:110
:
18:31 gtodd left
colomon :/ 18:31
pmurias psch: how does the problem manifest (with the typeobject being "repossessed" manifest) itself when running Perl 6 code? 18:32
18:36 Hor|zon joined
pmurias .tell konobi I'll replace the build system for nqp-js with a saner one tommorrow 18:38
yoleaux pmurias: I'll pass your message to konobi.
18:38 gtodd joined
psch pmurias: github.com/perl6/nqp/blob/master/s...e.java#L17 we have to compare STables here, instead of classHandle identity 18:39
pmurias: that line should just read "if (rd.classHandles[i] == classHandle) {", and that's what moar does
pmurias: but on jvm we get CORE on the LHS and BOOTSTRAP on the RHS when trying to instantiate an Attribute 18:40
for the SC that is
18:41 Hor|zon left
psch the SC mismatch goes away when "BEGIN Attribute.^compose" gets removed from Any-interable-methods.pm, but that's not a fix because we don't get the augment for Attribute (or something like that, i didn't quite follow jnthns explanation on that...) 18:41
pmurias psch: how can I reproduce the error? 18:43
psch pmurias: just remove ".st" in the linked line in nqp, build nqp, then build rakudo 18:44
pmurias: doing so will die with 'No such attribute "$!name" ...' during NativeCall.pm
18:46 gtodd left 18:47 gtodd joined 18:48 gtodd left 18:52 vendethiel left 18:53 xfix joined 18:55 cognominal joined
dalek kudo/nom: ebd36d9 | lizmat++ | src/core/Supply.pm:
Emit status at end if status supply given
 18:56
18:58 Ven joined 19:04 llfourn joined
ugexe can you apply Actions to an existing Match? 19:05
19:06 Ven left 19:09 llfourn left 19:13 virtualsue joined, gonz_ left 19:15 Akagi201 joined 19:17 Ven joined 19:19 yuval_dsfdsgdf joined 19:21 Ven left 19:22 Akagi201 left, domidumont left 19:26 Ven joined 19:34 virtualsue left
moritz ugexe: you can call action methods manually 19:35
$obj.method($/)
and that method can call $/.made or $/.ast
but there is no mechanism for retroactively calling all action methods 19:36
19:36 Ven left 19:37 Hor|zon joined
lizmat m: await start -> $i { say "foo" } # probably another error going into oblivion :-( 19:37
camelia ( no output )
moritz though something like for $/.hash.keys -> $method { $actions."$method"($/) if $actions.^can($method) }; might do the trick
lizmat m: await start { say "foo" }
camelia rakudo-moar ebd36d: OUTPUT«foo␤»
19:41 Hor|zon left 19:42 virtualsue joined
dalek c: 963bb01 | moritz++ | doc/Type/DateTime.pod:
Fix typo. Closes #168
 19:42
19:43 kjs_ left
dalek kudo/nom: 77e76ae | lizmat++ | src/core/Promise.pm:
Pass any extra parameters with Promise.start
 19:45
19:45 Ven joined 19:51 Ven left 19:54 vendethiel joined
dalek pan style="color: #395be5">perl6-examples: 1b1e29e | moritz++ | categories/games/wizard.pl:
Remove wizard.pl

it does not work. And probably never did. Closes #33, though probably not as intended		 19:55
c: 5e82274 | (Steve Mynott)++ | doc/Language/faq.pod:
fix some typos
 20:02
20:04 llfourn joined, BenGoldberg left 20:06 adhoc joined 20:07 zakharyas left 20:08 yuval_dsfdsgdf left 20:09 llfourn left
moritz wow, I still had .pir files in my checkout of DBIish 20:11
tadzik :) 20:12
20:12 tokuhiro_ joined
tadzik I like it when I have an old, ufo-generated Makefile and when I run it it says something like "pir target is not supported" 20:12
20:16 tokuhiro_ left 20:18 Akagi201 joined 20:20 darutoko left
moritz ok, API design question 20:20
I've been thinking a bit about DBIish 20:21
and the archaic way we still do a while my @row = $sth.fetchrow { ... }
so I thought instead it should look like
for $sth.rows -> @row { } 20:22
and possibly
for $sth.rows(:hash) -> %row { }
lizmat moritz: did you consider a $sth.supply ?
moritz lizmat: no; I'm try to come up with a sane synchronous API first :-) 20:23
*trying
lizmat afaik, db engines such as MySQL are capable of streaming result sets per record, which would match suppies :-)
ok
moritz so, if one only executes a query once, an even nicer API would be for $dbh.select($sql, |@placeholders) -> $row { } 20:24
the only thing that's stopping me right now is that statement handles already have a method "rows", which returns the number of affected rows (like, after an UPDATE) 20:25
nine Maybe "all" instead of "rows"?
20:25 Akagi201 left
lizmat perhaps metacpan.org/pod/DBIx::Simple can be of inspiration ? 20:26
Juerd Whoaaa
moritz nine: worth considering
Juerd I still have a highlight on that :)
nine In any case the only thing from crying out "great idea, just do it!" is that as lizmat++'s comment shows, there may be bigger steps to make
awwaiid thelackthereof.org/TLT/2015.10.18/M...l6_Grammar -- perl6 blog post 20:27
flussence realises postgresql's NOTIFY statement is a perfect fit for p6 supplies... but that's jumping the gun a bit :) 20:30
moritz flussence: I just did a quick overview of the postgresql, mysql and sqlite3 C client APIs, and it seems that postgres' is the only one that does async 20:33
flussence: ... though the docs on LISTEN and NOTIFY seem to be rather sparse :( 20:34
it seems we'd have to pull for results in a thread 20:35
threads and C libraries... I forsee a lot of fun
moritz stays with the synchronous API 20:36
20:38 Hor|zon joined 20:40 firstdayonthejob joined
pink_mist moritz: I know that in the p5 world, Mojo::Pg uses postgresql asyncly ... I thought Mojo::mysql did too even ... and perl5 doesn't need threads to do it 20:42
btyler event loop based, yeah 20:43
flussence
.oO( the gist I get from postgres' docs is that it's a threadsafe-if-you-really-really-need-it library, but it's saner to just open multiple DB connections for concurrency. I think that logic makes sense for most other databases too. )
20:43 xfix left, Hor|zon left 20:44 FireFly left 20:45 bjz_ left 20:46 boggard left
leont thinks lazy lists would make more sense to most people than supplies most of the time 20:47
20:47 boggard joined
leont (re databases) 20:47
20:51 kaare_ left, kaare_ joined
flussence I might get crucified for saying this but... PHP PDO was a pretty nice API. It worked pretty similar to moritz' suggestions above (though waaaaay more verbose) 20:52
(though I don't miss the horrible fetch-directly-into-object "feature") 20:55
20:56 kaare_ left 21:05 bjz joined 21:07 AlexDaniel joined 21:09 pippo joined
pippo o/ #perl6 21:11
21:11 bjz left
pippo m: my @a = ^3; say @a[0..1,2].perl; @a[0,1,2].perl; 21:11
camelia rakudo-moar 77e76a: OUTPUT«((0, 1), 2)␤»
pippo m: my @a = ^3; say @a[0..1,2].perl; say @a[0,1,2].perl;
camelia rakudo-moar 77e76a: OUTPUT«((0, 1), 2)␤(0, 1, 2)␤»
pippo ^^Is it correct to have them different? 21:12
21:13 bjz joined
pink_mist m: my @a = ^3; say @a[(0..1,2).flat].perl; 21:14
camelia rakudo-moar 77e76a: OUTPUT«(0, 1, 2)␤»
pink_mist if you want the flattened list, you need to be explicit
pippo pink_mist: is it something intended to be like that? 21:15
pink_mist afaik yes
pippo pink_mist: OK. thank you. 21:16
21:16 bjz left, bjz joined
pippo m: my @a = ^3; say @a[|0..1,2].perl; say @a[0,1,2].perl; 21:17
camelia rakudo-moar 77e76a: OUTPUT«((1,), 2)␤(0, 1, 2)␤»
pippo m: my @a = ^3; say @a[|(0..1),2].perl; say @a[0,1,2].perl;
camelia rakudo-moar 77e76a: OUTPUT«(0, 1, 2)␤(0, 1, 2)␤»
pippo m: my @a = ^3; say @a[|^2,2].perl; say @a[0,1,2].perl;
camelia rakudo-moar 77e76a: OUTPUT«(0, 1, 2)␤(0, 1, 2)␤»
pippo :-| 21:18
o/
21:18 pippo left 21:21 Akagi201 joined 21:26 aindilis left 21:28 Akagi201 left, bjz left 21:34 gonz_ joined 21:36 bjz joined 21:37 bjz left
dalek kudo/nom: fc1ef69 | lizmat++ | src/core/Supply.pm:
Implement Supply.throttle for Callables

If the second positional parameter to Supply.throttle is a Callable, it is assumed to be a block to be executed asynchronously restrained to the limit given. This basically reduces the asynchronous image resizer example in jnthn.net/papers/2014-yapceu-async.pdf to:
   @worklist.Supply.throttle(PARALLEL_RESIZE,&resizer-one).wait		 21:37
21:39 Hor|zon joined
lizmat am I the only one who thinks coercing a Supply to a Promise, would yield a Promise that is kept when the Supply is done? 21:40
instead of a Promise that is kept for the next emitted values ?
*value
21:41 tokuhirom_ joined 21:44 Hor|zon left 21:46 tokuhirom_ left
leont Wait, what? 21:50
That doesn't sounds logical at all
lizmat
.oO( I was already waiting for 10 mins)
what doesn't? the current behaviour, or the proposed behaviour ? 21:51
21:51 vendethiel left 21:52 _28_ria left 21:55 _28_ria joined
geekosaur I can see uses for what you describe as the current case, but it sounds rather niche. Is there an existing way to get your proposed behavior? 21:56
lizmat Supply.wait
the thing is: that waits and *then* returns the result of an internal Promise 21:57
I was just thinking it would be nice to use "await" for Supplies as well
await $promise, $supply.Promise, etc...
and/or even have await() auto-coerce to Promise 21:58
22:00 pmurias left 22:05 llfourn joined 22:06 ^elyse^ left, virtualsue left
RabidGravy m: my $a = True; given "boo" { when "bar" { say "bar"; }; if !$a { when "boo" { say "boo" } } }; # am I relying on something I shouldn't here? 22:07
camelia ( no output )
RabidGravy or will the if under no circumstance change the topic for the inner whens? 22:09
22:10 rindolf left 22:11 llfourn left 22:14 tokuhiro_ joined 22:15 vike left 22:17 vike joined 22:18 tokuhiro_ left 22:21 Grrrr left 22:24 Akagi201 joined
hartenfels RabidGravy if will only change your topic if you tell it to do so explicitly 22:24
See my $a = True; given "boo" { when "bar" { say "bar"; }; if !$a { when "boo" { say "boo" } } };
22:24 rurban left
hartenfels Oops 22:24
doc.perl6.org/language/control#if 22:25
22:26 Grrrr joined
hartenfels Disregard the code, that was the wrong paste buffer. 22:26
22:27 RabidGravy left 22:31 Akagi201 left 22:37 atweiden joined, atweiden left, atweiden joined 22:40 Hor|zon joined
atweiden with NativeCall, can a ulonglong return a negative number? 22:40
be*
i ask because it's happening here github.com/atweiden/digest-xxhash/.../xxHash.pm 22:42
pink_mist iirc I heard something about unsigned nativecall stuff not being implemented yet
but I don't quite recall when or in what context 22:43
22:45 TEttinger joined, Hor|zon left
pink_mist a search through the last few weeks of the irc log would probably find it 22:46
23:02 AndyDee joined 23:03 firstdayonthejob left 23:12 tokuhirom_ joined 23:14 tokuhiro_ joined
dalek kudo/nom: 3ac9491 | lizmat++ | docs/ChangeLog:
Some more ChangeLog additions
 23:14
kudo/nom: 2819ea2 | lizmat++ | src/core/Supply.pm:
Implement Supply.await

Like Supply.wait, but instead returns the Promise, instead of blocking for the Promise to be kept/broken and then return the result.
kudo/nom: fae01fb | lizmat++ | src/core/asyncops.pm:
await $supply works, can mix with Promise/Channel

Because it will now call $supply.await. This may need some further discussions. To *me*, this feels like the natural thing, since Supplies have their handling normally somewhere else anyway. If this is consensed to be a bad idea, then maybe it should not call await, but the Promise method instead.
Heuristic branch merge: pushed 55 commits to rakudo/curli by lizmat 23:15
lizmat and on that note,I wish #perl6 a good night! 23:16
23:16 tokuhirom_ left 23:19 tokuhiro_ left 23:27 Akagi201 joined 23:34 Akagi201 left
timotimo huh, "await" as the method name seems strange to me 23:34
i'd rather expect .Promise or so? 23:35
masak morning, #perl6 23:39
masak grins 23:40
23:41 Hor|zon joined 23:46 Hor|zon left 23:47 BenGoldberg joined 23:48 captain-adequate left 23:53 [Sno] left 23:59 uucpp joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!