»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.
00:03 yqt left 00:05 cognominal left, cognominal joined 00:17 mr_ron left, mr_ron joined 00:30 leont left 00:41 aborazmeh joined, aborazmeh left, aborazmeh joined 00:44 laouji joined 00:53 tokuhiro_ joined 00:54 pmurias left 00:57 tokuhiro_ left 01:02 mr_ron left 01:09 Zoffix is now known as CPAN2, CPAN2 is now known as Zoffix_, Zoffix_ is now known as FunBag, FunBag left
awwaiid hm. Well I was going to ask a question about some slang thing I was messing with but then I saw that whitespace-before-method-dot (or infix:<.>) is core now. amazing. 01:11
01:11 lichtkind joined 01:12 Zoffix joined 01:13 mr_ron joined 01:20 aborazmeh left
Juerd awwaiid: Before and after, if you like :) 01:20
awwaiid hops up and down and mutters about being excited for Christmas 01:24
01:36 BenGoldberg joined
Zoffix mumbles something about this still being just September 01:40
timotimo learnxinyminutes probably has some verbiage about "you have to put a \ after $foo and before the newline" 01:42
timotimo disappears
01:51 lichtkind_ joined 01:58 vvz left 02:08 makapa joined, kid51_ joined 02:10 lichtkind_ left, kid51 left 02:11 kid51_ is now known as kid51 02:19 telex left 02:20 telex joined 02:21 mr_ron left 02:24 kid51 left
dalek kudo/nom: e5a266b | TimToady++ | src/Perl6/Actions.nqp:
infix:<o> shouldn't curry WhateverCode
 02:25
02:33 AlexDaniel left
dalek ast: fc511ea | TimToady++ | S03-operators/composition.t:
test o against whatevers
 02:35
02:35 noganex joined 02:38 noganex_ left 02:45 makapa left 02:54 tokuhiro_ joined 02:58 kaare_ joined 02:59 tokuhiro_ left 03:12 pierrot left 03:13 tokuhiro_ joined, pierrot joined 03:21 lichtkind left 03:39 makapa joined 03:44 makapa left 04:15 makapa joined 04:27 makapa left 04:28 makapa joined 04:36 skids left 04:40 [Sno] left 04:43 davido_ joined 05:00 BenGoldberg left
dalek kudo/nom: 92be579 | TimToady++ | src/core/Range.pm:
Calculate elems of non-num range more directly
 05:22
05:32 makapa left 05:36 anon joined 05:37 anon is now known as Guest72063 05:44 Guest72063 left, Guest72063 joined 05:45 araujo left, Guest72063 is now known as wooden_, araujo joined 05:48 wooden_ left 05:56 darutoko joined 05:57 emilbarton joined
emilbarton hello, how can I pass a reference to a class as attribute in another class and not get "Cannot lookup attributes in a Type object" when trying to use a method of the first class within a method in the second one? 05:58
I mean a ref to a class instance 06:02
06:05 [Sno] joined 06:09 FROGGS joined
emilbarton hello, how can I pass a reference to a class instance as attribute in another class and not get "Cannot lookup attributes in a Type object" when trying to use a method of the first class within a method in the second one? 06:20
FROGGS .tell jnthn why will shaped arrays in future declared as 'my @a[3,3]' but used as '@a[1;1] = 42'? why not semicolons in the decl too? 06:27
yoleaux FROGGS: I'll pass your message to jnthn.
nine emilbarton: if you can show us some code, that would help answering your question 06:30
FROGGS .tell Skarsnik You'd declare the CStruct as 'class Foo is repr<CStruct> { has int32 $.bar }' and then 'say nativecast(Foo, $pointer)' 06:33
yoleaux FROGGS: I'll pass your message to Skarsnik.
06:37 Ven joined
TimToady they're declared as my @a[3;3] 06:37
Ven waves 06:38
06:40 bjz joined 06:41 bjz left 06:46 davido_ left
moritz particles 06:49
emilbarton hello, how can I pass a reference to a class instance as attribute in another class and not get "Cannot lookup attributes in a Type object" when trying to use a method of the first class within a method in the second one? 06:51
nine emilbarton: if you can show us some code, that would help answering your question 06:55
emilbarton nine: Class1 { method my1 { ... } } ; Class2 { has Class1 $!classref; method my2 { .. $!classref.my2() } }; my Class1 $C1 = new.(); my Class2 $C2 = new.( classref => $C1 ); $C2.my2(); # Cannot lookup attributes in a Type object... 06:57
nine m: class Class1 { method my1 { } } ; class Class2 { has Class1 $.classref; method my2 { $!classref.my1() } }; my Class1 $C1 = Class1.new(); my Class2 $C2 = Class2.new( classref => $C1 ); $C2.my2(); 06:59
camelia ( no output )
nine emilbarton: a fixed version ^^^
emilbarton yes I forgot Classx. but that's not the point, it still answers idem 07:00
nine emilbarton: the error is in the has Class1 $!classref; That creates a private attribute which the default constructor does _not_ set. You have to make it public or to add your own BUILD method to initialize it using the parameter of the constructor.
07:01 abraxxa joined
nine emilbarton: doc.perl6.org/language/classtut#Constructors 07:01
emilbarton nine: I thought there was no real public attribute in perl6 ($.classref does n't help) 07:02
nine emilbarton: that's sort of true. The attribute itself is still only accessible through the accessor methods. But those are generated for you by declaring it as has $.classref; and this also tells the constructor to initialize it. 07:03
m: class Class1 { has $.foo = "foo"; method my1 { self.foo; } } ; class Class2 { has Class1 $.classref; method my2 { $!classref.my1() } }; my Class1 $C1 = Class1.new(); my Class2 $C2 = Class2.new( classref => $C1 ); $C2.my2();
camelia ( no output )
nine emilbarton: also a reference to a class instance is more commonly referred to as "object". 07:05
emilbarton nine: ok, I see where it messed up, I already have a BUILD() in Class2 but forgot to update it when I added classref. Thanks. 07:08
07:10 eugu joined 07:11 Ven left 07:12 emilbarton left
tokuhiro_ hi, i got a issue around IO::Socket::Async. my network server written in perl6 stucks around 1000 requests. 07:13
code is here: gist.github.com/tokuhirom/0b6901f3cea1c3bf6170
07:14 bjz joined
FROGGS TimToady: okay, so just a typo in jnthn++'s blog :o) 07:16
tokuhiro_ ah, i forgot to close socket. 07:20
07:21 bjz left, bjz joined 07:24 bjz left 07:29 Ven joined
Ven I just realized D has a similar features to our phaser, but with a separate keyword (instead of uppercased ones): `scope(success) write("yay"); scope(failure) write(":(");` 07:35
(I just uppercased names are just considered "not explicit enough" in that kind of area) 07:36
07:38 bjz joined
moritz
.oO( explicit content )
 07:40
07:42 g4 joined
timotimo plasmas 07:52
masak good state of matter, timotimo
timotimo though i find you can't really write "/me liquids" or "/me gasses" without it sounding ... funny in a bad way :) 07:53
moritz better than a matter of state, for sure
masak .oO( enemy of the states of matter ) 07:54
lizmat condenses
moritz waits for somebody to (re)sublimate 07:55
Ven m: say(); say() 07:56
camelia rakudo-moar 92be57: OUTPUT«␤␤»
lizmat m: note(); note() 07:57
camelia rakudo-moar 92be57: OUTPUT«Noted␤Noted␤»
lizmat m: note; note
camelia rakudo-moar 92be57: OUTPUT«Noted␤Noted␤»
moritz duly.
Ven .oO( /dullies ) 07:58
timotimo no time for duly dallying 08:03
lizmat m: say (-> --> Int { |(1,2,3) })().WHAT # eh, shouldn't this fail ? 08:09
camelia rakudo-moar 92be57: OUTPUT«(Slip)␤»
08:11 duncan_dmg joined
masak lizmat: at compile time, yes. 08:11
lizmat well, ideally, yes
masak yes
lizmat but I'll settle for runtime
for now
masak .oO( or both ) :P
lizmat m: say (-> --> Int { fail })().WHAT
camelia ( no output )
lizmat m: say (-> --> Int { die })().WHAT 08:12
camelia rakudo-moar 92be57: OUTPUT«Died␤ in block <unit> at /tmp/i4Xc5H0FAT:1␤␤»
timotimo well, fail works with the outer Routine, so at least i understand the first thing
08:12 RabidGravy joined
lizmat RT #126232 submitted 08:20
08:28 eugu left 08:31 Khisanth left 08:32 larion joined 08:36 vvz joined 08:41 dakkar joined 08:43 telex left 08:51 Khisanth joined 08:54 laouji left 08:58 laouji joined, telex joined 09:01 ][Sno][ joined, bjz left 09:02 bjz joined 09:04 [Sno] left
timotimo it seems like i didn't push the manual change to the CALLER::CALLER:: stuff 09:04
and i also didn't start implementing the optimization in the static optimizer 09:06
another thing is that my manual optimization didn't factor out the CALLER::CALLER:: into a single lookup yet 09:07
that could also give a tiny performance boost on top of what's already there 09:08
though of course the nqp ops that access the lexpad "directly" aren't nearly as expensive as allocating a pseudostash or something like that 09:09
09:31 laouji left 09:34 laouji joined 09:38 ChoHag joined
ChoHag I have (want) a class which has a Foo attribute. A Foo can be created from a string (and a string created from a Foo) so Foo has a method new (Str $newfoo) { ... }. How can I create an attribute so that the class using it will create it from a string without needing to manually create a Foo? 09:43
yoleaux 1 Sep 2015 20:46Z <TimToady> ChoHag: irclog.perlgeek.de/perl6/2015-08-30#i_11144152 looks like a bug to me, since .= is sugar for assignment, and assignment is supposed to be eager; please file a bug report, if you haven't already
timotimo do we still have the COERCE multi sub? 09:44
ChoHag ie. how can I enable the ability to say Bar.new(foo => "Something") rather than being forced to say Bar.new(foo => Foo.new("Something"))?
timotimo well, with a .new it's very easy
if all else fails, the constructor can do the conversion for you
and you can have a Proxy for your attribute that does conversion as well
so no tricky business needed even in that case 09:45
does that help, ChoHag? 09:47
if Str had a method "Foo" on it, you can just set the type of $.foo to be Foo(Str) 09:48
you can augment Str to have that method, but subclasses of Str won't have it (like IntStr, NumStr, RatStr, Complexstr)
ChoHag No, I want to be able to use it in locations other than another class' attribute. 09:49
ie. my SpecialString $foo = "normal-string-but-fooified". 09:50
timotimo unless you declare $foo to be coercive (like Foo(Str)) that won't work
ChoHag my SpecialString(Str) $foo? 09:51
timotimo yeah
that will call .SpecialString on anything that's Str
09:51 laouji left
ChoHag So can I have something like ReallySpecialString which stands in for SpecialString(Str)? 09:53
timotimo there used to be something that'd allow you to declare coercions as a multi sub, but that has the problem that unless you export that multi sub into other scopes, the coercion wouldn't work there
good question
m: constant CToInt = Int(Str); my CToInt $foo = "100"; say $foo.WHAT 09:54
camelia rakudo-moar 92be57: OUTPUT«5===SORRY!5=== Error while compiling /tmp/IwUEdyBXDu␤Coercion Int(Str) is insufficiently type-like to qualify a variable␤at /tmp/IwUEdyBXDu:1␤------> 3nstant CToInt = Int(Str); my CToInt $foo7⏏5 = "100"; say $foo.WHAT␤ expecting any of…»
timotimo m: my Int(Str) $foo = "100"; say $foo.WHAT
camelia rakudo-moar 92be57: OUTPUT«5===SORRY!5=== Error while compiling /tmp/WjQi1BQkqO␤Coercion Int(Str) is insufficiently type-like to qualify a variable␤at /tmp/WjQi1BQkqO:1␤------> 3my Int(Str) $foo7⏏5 = "100"; say $foo.WHAT␤ expecting any of:␤ constraint␤»
timotimo oh, i didn't know that doesn't work
that seals it, then
ChoHag Well basically what I have/want is an object which can search through a bunch of linked items. The search object needs to be passed around/stored in various places, and the original string needs to be obtainable from it. 10:00
And basically I want to be able to provide a regular string in all the places the search object is expected and have it magically turn into a search string for me. 10:01
Pretty basic.
gfldex ChoHag: did you consider to subclass Str? 10:02
timotimo doesn't really help to make the code simpler there
there's a module that'll let you change how literals are created, but that'll really only enhance your literals 10:03
and it'll enhance "all" literals
what you can do is mix in a role to improve your Str into the SearchStr 10:04
10:04 lizmat left, lizmat joined
ChoHag role Magic { ... }; class SearchStr is Str does Magic { ... }? 10:04
timotimo not even that 10:05
role Magic { ... }; my $foo = "foobar" but Magic;
also:
10:05 espadrine joined
timotimo sub searchWithMagic($inputStr is rw) { $inputStr does Magic unless $inputStr ~~ Magic; ... } 10:05
10:10 laouji joined
ChoHag So anywhere I need a magic string but can coerce to one from a regular string I need to do so explicitely? 10:11
timotimo seems so 10:12
ChoHag Sticking to constructors then, how do I enhance this: 10:13
m: role Magic { method end { "knob" } }; class Wizard { has Magic $.staff }; my $rincewind = Wizard.new(staff => "wizzard");
camelia rakudo-moar 92be57: OUTPUT«Type check failed in assignment to $!staff; expected Magic but got Str␤ in block <unit> at /tmp/OyhDQgdKqC:1␤␤»
10:14 rindolf joined
timotimo m: role Magic { method end { "knob" } }; class Wizard { has Magic $.staff; method new(Str :$staff) { self.bless(staff => $staff ~~ Magic ?? $staff !! $staff but Magic) } }; my $rincewind = Wizard.new(staff => "wizzard"); say $rincewind.staff.perl 10:15
camelia rakudo-moar 92be57: OUTPUT«"wizzard"␤»
timotimo m: role Magic { method end { "knob" } }; class Wizard { has Magic $.staff; method new(Str :$staff) { self.bless(staff => $staff ~~ Magic ?? $staff !! $staff but Magic) } }; my $rincewind = Wizard.new(staff => "wizzard"); say $rincewind.staff.end
camelia rakudo-moar 92be57: OUTPUT«knob␤»
ChoHag Of course. I should have known that. 10:17
10:19 laouji_ joined, laouji left 10:20 Ven left
ChoHag Is there any advantage to the ternary rather than a multi method which distinguishes between Str :$staff and Magic :$staff? 10:23
grondilu suspects using a multi is better a choice 10:25
because possible compile time optimizations or something 10:26
but other people surely know better
gfldex m: my Int|Str $a;
camelia rakudo-moar 92be57: OUTPUT«5===SORRY!5===␤Type 'Int' is not declared. Did you mean 'int'?␤at /tmp/kuUWtnhKeF:1␤------> 3my Int7⏏5|Str $a;␤Malformed my␤at /tmp/kuUWtnhKeF:1␤------> 3my Int7⏏5|Str $a;␤␤»
gfldex i'm pretty sure Int is declared
grondilu m: my int|Str $ 10:27
camelia rakudo-moar 92be57: OUTPUT«5===SORRY!5===␤Type 'int' is not declared. Did you mean 'Int'?␤at /tmp/3S4oURehT5:1␤------> 3my int7⏏5|Str $␤Malformed my␤at /tmp/3S4oURehT5:1␤------> 3my int7⏏5|Str $␤␤»
grondilu lol
gfldex and i believed that used to work
m: sub foo(Int|Str $a){};
camelia rakudo-moar 92be57: OUTPUT«===SORRY!===␤No compile-time value for Str␤»
10:28 zakharyas joined 10:32 kid51 joined
moritz gfldex: you can make an explicit subtype 10:33
subset IntOrStr of Cool where Int|Str
10:36 AlexDaniel joined
grondilu m: my Str|Int $ 10:38
camelia rakudo-moar 92be57: OUTPUT«5===SORRY!5===␤Type 'Str' is not declared. Did you mean 'str'?␤at /tmp/FI7BDfvGlk:1␤------> 3my Str7⏏5|Int $␤Malformed my␤at /tmp/FI7BDfvGlk:1␤------> 3my Str7⏏5|Int $␤␤»
dalek osystem: 66aa639 | (Tokuhiro Matsuno)++ | META.list:
Retire Raw-Socket
 10:50
10:58 andreoss joined 11:08 abraxxa left, rindolf left 11:13 vvz left 11:14 Ven joined, g4 left 11:19 telex left 11:20 telex joined
masak grondilu: that's definitely a LTA error message. 11:28
11:29 cognominal left
masak so is the one with the `foo(Int|Str $a){}` 11:30
Ven .tell timotimo irclog.perlgeek.de/perl6/2015-10-01#i_11301705 something like this, probably
yoleaux Ven: I'll pass your message to timotimo.
gfldex i rakudobuged that LTA as #126234
Ven m: say (*o*) 11:31
camelia rakudo-moar 92be57: OUTPUT«WhateverCode.new␤»
Ven it's great to have smileys :-)
11:31 rindolf joined
moritz m: say (*o*)(* * 2, * + 2)(5) 11:31
camelia rakudo-moar 92be57: OUTPUT«14␤»
Ven m: my \id = -> $a { $a }; (*o*)(id, id)(5).say
camelia rakudo-moar 92be57: OUTPUT«5␤»
masak .oO( PIG WITH GLITTERING EYES )
moritz still occasionally wants * as a callable to be the identity function 11:32
timotimo don't we have an identity method now, though? 11:33
yoleaux 11:30Z <Ven> timotimo: irclog.perlgeek.de/perl6/2015-10-01#i_11301705 something like this, probably
11:34 [ptc] joined, [ptc] left
lizmat moritz: it it is just for map, we could e.g. add a candidate for Whatever ? 11:34
moritz lizmat: no, never for map 11:35
usually for experimenting with other higher-order functions
lizmat where would you use it as a callable then ?
I usually use something like *.Str
Ven timotimo: I want to update the learnx big time for xmas 11:39
timotimo it gets mentioned from time to time on twitter still
mostly in the "look at this crazy shit, a hundred pages to learn the basics of perl6" 11:40
11:40 ptc joined, ptc is now known as Guest69182, Guest69182 left
Ven timotimo: I'd be interested to see such tweets 11:42
"the author got a bit carried away :)" well, as a learnx contributor, I can say without a single doubt that the perl6 tutorial goes far, far, far beyond what the others provide 11:44
timotimo right 11:45
Ulti .tell tony-o just sent you a little pull request for Slang::SQL for use nqp; there is another issue around the /is rw/ ish changes 11:46
yoleaux Ulti: I'll pass your message to tony-o.
11:52 ptc_ joined, ptc_ is now known as [ptc] 11:53 FROGGS[tab] left 11:54 [ptc] left 11:55 [ptc] joined, laouji_ left 11:57 lizmat left 12:01 rindolf left 12:02 FROGGS[tab] joined 12:03 [ptc] left, [ptc] joined
Ven timotimo: not sure that's a bad, or good thing, per se 12:12
12:16 sufrostico joined
timotimo it makes a little bit of a bad impression; learnxinyminutes isn't the right place for such a big thing, it seems 12:19
i'm not saying "you did bad" or "the tutorial isn't good"; it's just the unfortunate view of others :\
Ven timotimo: that's why I stopped putting stuff in it, fwiw 12:20
Ven certainly has tons of ideas on how to make ruby's tutorial 10 times bigger
ShimmerFairy Just giving it a quick glance, it seems like it needs a serious reduction, if my understanding of xinyminutes as an introduction to get you up and running is correct. 12:21
12:21 rindolf joined 12:23 AlexDaniel left 12:24 abraxxa joined 12:35 ZoffixWork left 12:42 zakharyas left 12:47 zakharyas joined
dalek osystem: 9a888ba | (Shoichi Kaji)++ | META.list:
add HTTP::Tinyish
 12:48
osystem: f16bd64 | (Zoffix Znet)++ | META.list:
Merge pull request #62 from shoichikaji/http-tinyish

add HTTP::Tinyish
masak maybe the learnx page should link on to pages of the perl6 docs, for example. 12:56
"look here to learn more!", like
Ulti or clone the structure from other languages and fill out the Perl 6 stuff doc.rust-lang.org/stable/book/ 12:58
I semi started doing that via a git books site 12:59
Ven I should probably start writing my blog...
13:03 pippo joined
pippo m: say lines(); 13:03
camelia rakudo-moar 92be57: OUTPUT«(Céad slán ag sléibhte maorga Chontae Dhún na nGall Agus dhá chéad slán ag an Eireagal ard ina stua os cionn caor is coll; Nuair a ghluais mise thart le Loch Dhún Lúich’ go ciúin sa ghleann ina luí I mo dhiaidh bhí gleanntáin ghlas’ Ghaoth…»
Ulti m: lines().elems.say 13:04
camelia rakudo-moar 92be57: OUTPUT«20␤»
pippo m: say grep {/^C/}, lines();
camelia rakudo-moar 92be57: OUTPUT«Method 'match' not found for invocant of class 'Any'␤ in block <unit> at /tmp/yY3Dm2vEXU:1␤␤»
13:04 aborazmeh joined, aborazmeh left, aborazmeh joined
pippo ^^ that used to work. 13:05
moritz m: say grep /^C/, lines()
camelia rakudo-moar 92be57: OUTPUT«(Céad slán ag sléibhte maorga Chontae Dhún na nGall)␤»
Ulti m: lines().map( *.words.elems ).Bag.perl.say
camelia rakudo-moar 92be57: OUTPUT«(5=>1,9=>1,12=>1,13=>4,11=>1,8=>1,16=>2,15=>2,14=>3,0=>3,17=>1).Bag␤»
moritz workaround
pippo moritz: thank you. Is is a bug?
ShimmerFairy pippo: I seem to recall jnthn calling that an accident that it ever did work, but I'm not sure.
pippo ShimmerFairy: Hah. OK then. I'll correct my script. 13:06
Thank you all!
moritz m: given 'abc' { say !/x/ } 13:07
camelia rakudo-moar 92be57: OUTPUT«Method 'match' not found for invocant of class 'Any'␤ in block <unit> at /tmp/xMRcmv9l9D:1␤␤»
moritz I'm sure *that's* a bug
and somehow it smells like the same (or a similar) issue
ShimmerFairy moritz: perhaps, though to be honest I've never seen someone try to negate a regex (and visually, it just doesn't make sense to me) 13:08
at least, not until people started pointing out that bug :)
13:08 duncan_dmg left
moritz ShimmerFairy: I've seen it quite often 13:08
ShimmerFairy P5 must be weird then :P 13:09
moritz my @sane = grep { !/\s/ }, @list_of_potential_usernames
ShimmerFairy: it's not negating the regex, it's negating the result of the regex match
ShimmerFairy moritz: but it reads like negating a regex. I'd argue that you shouldn't expect it to work in the first place.
(and fwiw I still think not using ~~ to match against a regex is kinda weird :P) 13:10
13:10 duncan_dmg joined
moritz ShimmerFairy: but why not? 13:11
ShimmerFairy: /foo/ works in a boolean context
ShimmerFairy: and a negation provides a boolean context
13:11 Ven left
moritz programming languages are all about composing bigger pieces from smaller ones, some composability is really key 13:11
if you think something *shouldn't* compose, you need a really good argument to convince me. 13:12
ShimmerFairy moritz: I honestly think it's a mistake that m// and friends magically match against $_ . And I still say it reads like you're negating the actual regex
pippo my @a = <a b c>; my regex letters = { @a }; say grep /^<letters>/, lines();
m: my @a = <a b c>; my regex letters = { @a }; say grep /^<letters>/, lines();
camelia rakudo-moar 92be57: OUTPUT«5===SORRY!5=== Error while compiling /tmp/Zai3tUiH0z␤Malformed regex␤at /tmp/Zai3tUiH0z:1␤------> 3my @a = <a b c>; my regex letters 7⏏5= { @a }; say grep /^<letters>/, lines()␤»
moritz ShimmerFairy: well, but // does match against $_, even if you don't like it 13:13
ShimmerFairy: and "it reads like" doesn't convince me. To me, it doesn't read like it.
pippo: just "my regexe letters { ... }"
m: my @a = <a b c>; my regex letters { @a }; say grep /^<letters>/, lines;
camelia rakudo-moar 92be57: OUTPUT«()␤»
13:13 rindolf left
moritz m: my @a = <A B C>; my regex letters { @a }; say grep /^<letters>/, lines; 13:13
camelia rakudo-moar 92be57: OUTPUT«(Céad slán ag sléibhte maorga Chontae Dhún na nGall Agus dhá chéad slán ag an Eireagal ard ina stua os cionn caor is coll; Ag taisteal dom amach trí chnoic Ghleann Domhain ’s an Mucais ar mo chúl Ach trom lámh Gall, le cluain 's le feall, a thi…»
ShimmerFairy moritz: regardless, it looks to me like you're negating the regex. It'd be like claiming that !"str" didn't negate the string, but rather negated the result of doing $_ eq "str" 13:14
moritz ShimmerFairy: ! simply negates the boolean value of whatever comes after it
ShimmerFairy: and the boolean value of a regex just happens to be the match against $_
pippo moritz: TY
m: my @a = <a b c>; my regex letters { @a }; say grep /^<letters>/, lines();
camelia rakudo-moar 92be57: OUTPUT«()␤»
moritz ShimmerFairy: so, by current rules, it should work. That it does not is a bug 13:15
pippo m: my @a = <a b C>; my regex letters { @a }; say grep /^<letters>/, lines();
camelia rakudo-moar 92be57: OUTPUT«(Céad slán ag sléibhte maorga Chontae Dhún na nGall)␤»
eiro hello #p6 folks
pippo eiro: o/
timotimo yeah, /foobar/ isn't the regex, it's an application of the regex already
otherwise you'd write rx/.../ or regex { ... }
ShimmerFairy moritz: that's far too magical a boolean value for my tastes.
m: $_ = "abc"; say ?rx/def/ # then this shouldn't fail, if rx// is meant to be an unapplied regex, no? 13:16
camelia rakudo-moar 92be57: OUTPUT«Method 'match' not found for invocant of class 'Any'␤ in block <unit> at /tmp/g50UwRos6q:1␤␤»
moritz ShimmerFairy: I understand that. You mentioned that multilple times. But that doesn't preclude composability
eiro i remember seeing somthing like the haskell all@(x:xs) in perl6 and currently trying to find the relevant documentation about it 13:18
13:18 skids joined
eiro any clue ? 13:18
ShimmerFairy moritz: You can't convince me otherwise that !/.../ looks so, so wrong. I'm totally fine with the current behavior.
moritz m: sub f(*@all [ $first, *@rest] ) { say @rest }; f <a b c>; # for eiro 13:19
camelia rakudo-moar 92be57: OUTPUT«[b c]␤»
moritz subsignatures
not sure where or if they are documented
eiro thanks moritz 13:21
13:22 Ven joined
moritz eiro: doc.perl6.org/type/Signature#Destru...Parameters 13:26
masak moritz++ 13:28
13:34 rindolf joined
jdv79 ShimmerFairy: Perl is all about terse magic stuff like that. 13:35
pippo o/ #perl6 13:37
13:37 pippo left
masak ShimmerFairy: as another data point, I have no problem with reading !/\s/ in the intended way, as "match $_ against /\s/ and negate the result" 13:39
ShimmerFairy: to be more precise, there are many situations in Perl 6 (and Perl 5) where a regex stands in for its match result
ShimmerFairy: I agree that it's one thing we could aggressively wash that out of the language (like we are slowly doing with implicit flattening everywhere). but I just don't see the reason. 13:40
eiro thanks moritz... i realize i have to make those terms appear in my slides 13:41
jdv79 wash out topical? 13:42
eiro since [+] ^5; why not [&say] ^5 ?
masak jdv79: wash out auto-matching regexes. 13:44
Ven eiro: because &say could be an array value :)
m: [[&say]] &5; # eiro: hence
camelia ( no output )
jdv79 but thst basically kills one liners
Ven m: [[say]] &5; # eiro: hence 13:45
camelia rakudo-moar 92be57: OUTPUT«5===SORRY!5=== Error while compiling /tmp/IkIr9DhIIb␤Unsupported use of bare "say"; in Perl 6 please use .say if you meant $_, or use an explicit invocant or argument, or use &say to refer to the function as a noun␤at /tmp/IkIr9DhIIb:1␤------> 3…»
ShimmerFairy jdv79: no it doesn't, you can always do $_ ~~ /.../ (I myself have never done without using ~~, as far as I can recall) 13:46
eiro fair enought ... so if [[&say]] ^5, why not [[-> $x , $y { $x + $y }]] ?
jdv79 that alone would push most of minr to a file 13:47
terseness is serious biz in one liners 13:48
13:49 zakharyas left
eiro Ven, ok ... i just realized what [[&say actually is ... forget the question :) 13:51
Ven m: [[&say]] ^5; # eiro: hence
camelia rakudo-moar 92be57: OUTPUT«01␤True2␤True3␤True4␤»
Ven heh.
eiro i'm so excited about [] i hopped we could use a function there 13:52
masak eiro: the parser needs to decide whether the thing it's seeing is a reduce metaop, or an array. 13:55
eiro: [42] is an array. so is [[[[1]]]]. so is [&say]
eiro: only if it sees an infix op in there does it go the metaop path.
13:57 TEttinger left
eiro masak, thanks for those lines 13:58
masak eiro: in general, if you need to reduce on a function, I'd recommend using the .reduce method on lists/arrays. 14:00
m: sub foo($l, $r) { 5 * $l + $r }; say [1, 2, 3, 4].reduce(&foo) 14:01
camelia rakudo-moar 92be57: OUTPUT«194␤»
eiro that's what i'm trying to do right now.
masak eiro: the reason what you were trying is kind of hard is that there *should* be a barrier between the op and function namespaces 14:02
or rather, between their usage syntax
eiro i got it ... not the deep way (i need years for that) but a "pragmatic" way. 14:03
the thing is: whenever there is a operator, i prefer it over the method call 14:04
masak remember that locally, in your lexical scope, you can always create all the functions and operators that you need 14:05
that means you get your wish inside that scope, without screwing up the environment for everyone outside of that scope
more and more I do this, and I really like that Perl 6 supports this in a big way by making a lot of things lexical
eiro i already love it too :) 14:08
PerlJam It's funny to me how "lexical" plays an important role in the 5 -> 6 transition just like it did in the 4 -> 5 transition. 14:09
s/funny/interesting/ (I mean, it doesn't make me laugh or anything :) 14:10
14:11 pmurias joined 14:12 FROGGS[tab]_ joined 14:13 Ven left, FROGGS[tab] left 14:16 tokuhiro_ left
masak PerlJam: and it's even the same idea/concept. in Perl 6 we just turn it up even more. 14:16
14:16 aborazmeh left
pink_mist turn it up to eleve^Wsix! 14:19
14:23 Ven joined 14:30 vvz joined 14:36 pmurias left 14:38 pmurias joined, kaare_ left
jdv79 lex all the things 14:38
lexx wasnt too terrible of a show 14:39
14:39 kaare_ joined
jdv79 what else. lex luther. all i got. 14:39
masak the Lex building, Lex the Iraq war dog, Lex Vehicle Leasing, USS Lexington, and Lex, West Virginia. 14:42
PerlJam
.oO( google++ )
 14:43
masak there's also the old Roman law by that name. 14:44
mspo elx club 14:46
14:48 Ven left 14:52 Ven joined
dalek ecs: 9bf4b95 | (Nova Patch)++ | S32-setting-library/Containers.pod:
remove loitering space
 14:52
skids Can't have loiterers conglomerating in the specs. 14:54
Ven
.oO( Tune down for loit )
masak .oO( will loit for food ) 14:56
nebuchadnezzar just read 6guts article 14:57
jnthn: I read the single argument rule in S07, as a non native english (maybe), I needed to think twice to understand the comment in “@a.push: @b, # same, trailing comma doesn't make > 1 argument” 15:00
[Tux] panda still fails on rakudobrew checkout 15:01
no time for games or trial-and-error, so no timings today 15:02
_itz [Tux]: I had to "nuke" before it worked 15:04
[Tux] so have a fresh start?
_itz yes
[Tux] $ make again
nebuchadnezzar in the “for” case, “@a” and “@a,” are different, but in “push”, “@b” and “@b,” are the same 15:05
masak nebuchadnezzar: I believe there's been some very recent (as in yesterday) change there.
nebuchadnezzar: .push has split up into .push and .append, IIUC
nebuchadnezzar erf, should read in the git repo 15:06
masak nebuchadnezzar: to reduce the number of special cases of the kind you're confused about
nebuchadnezzar masak: thanks
I thought I understood but in fact not, I see the consistency between “for (@a,)” and “[@a,]”, but not with the “@a.push: @b,” 15:12
:-D
hmm, MoarVM september is out, need to package it 15:13
skids nebuchadnezzar: it was decided it was best if pop always got the same things that were pushed, so if you want the consistency, use append. 15:14
(or I guess you have to choose between one of two different consistencies. :-) 15:15
15:16 tokuhiro_ joined 15:20 Ven left 15:21 tokuhiro_ left 15:22 fling left 15:26 lizmat joined
[Coke] m: sub postfix:<thousand>($n) { $n*1000 }; say 1 thousand; 15:28
camelia rakudo-moar 92be57: OUTPUT«5===SORRY!5=== Error while compiling /tmp/zpDjL6c4m6␤Two terms in a row␤at /tmp/zpDjL6c4m6:1␤------> 3ostfix:<thousand>($n) { $n*1000 }; say 17⏏5 thousand;␤ expecting any of:␤ infix␤ infix stopper␤ postfix␤…»
[Coke] m: sub postfix:<thousand>($n) { $n*1000 }; say 1thousand; 15:29
camelia rakudo-moar 92be57: OUTPUT«1000␤»
15:29 FROGGS left
[Coke] I have a clue on the the profile issue with exit. 15:32
I updated MVM_profile_end to unconditionally throw an exception. --profile -e "" says "===SORRY!===\nSURPRISE" . --profile -e "exit" says "SURPRISE\n\nUnhandled exception: SURPRISE\n at..." 15:33
15:38 cognominal joined 15:39 Ven joined
cognominal I forget the syntax to use a function as an operator. Help 15:40
[Tux] Type check failed in assignment to $chunk; expected Str but got Seq
I hoped p6 would have stabalized on that by now 15:41
loop (my int $i = 0; $i < @ch.elems; $i = $i + 1) {
my Str $chunk = @ch[$i]; <==== that line
could that be the .push vs .append change? 15:42
so $buffer.defined and @ch.push: chunks ($buffer, $chx);
lizmat do you have the most recent panda ?
[Tux] will now do something different?
[Coke] m: sub b($a,$b) { $a * 2 + $b } ; 3 [&b] 4 #cognomial 15:43
camelia ( no output )
[Coke] m: sub b($a,$b) { $a * 2 + $b } ; say 3 [&b] 4 #cognomial
camelia rakudo-moar 92be57: OUTPUT«10␤»
[Tux] lizmat, was that q to me or to cognominal ?
lizmat [Tux]: to you
[Tux] I rebuilt from scratch
rm -rf moar*
lizmat ok, weird
rm -rf install is my usual way of nuking things 15:44
cognominal [Coke] thx.
[Tux] $buffer.defined and @ch.push: |chunks ($buffer, $chx); # note the |
fixes things
[Coke] rants about not having gdb available on os x so that ./perl6-gdb-m fails.
[Tux] almost 15:45
15:45 jasanj`` joined
lizmat [Tux]: what you're seeing are some skirmishes in the aftermath of the GLR 15:46
:-(
_itz there is lldb on mac now 15:47
15:49 jasanj` left, lichtkind joined
lichtkind will ther be eval in 6.0? 15:50
lizmat m: EVAL "say q/yes'"
camelia rakudo-moar 92be57: OUTPUT«5===SORRY!5=== Error while compiling EVAL_0␤Couldn't find terminator / (corresponding / was at line 1)␤at EVAL_0:1␤------> 3say q/yes'7⏏5<EOL>␤ expecting any of:␤ /␤»
lizmat m: EVAL "say q/yes'/"
camelia rakudo-moar 92be57: OUTPUT«yes'␤»
[Coke] _itz: aye. 15:51
lizmat eval is spelled EVAL nowadays (well, since july 2014)
[Tux] github.com/Tux/CSV/commit/a7c3fc3f...b7e985243e 15:52
All tests successful.
Files=25, Tests=19732, 70 wallclock secs ( 3.39 usr 0.17 sys + 227.77 cusr 2.61 csys = 233.94 CPU)
Result: PASS
lizmat yeah: so nuking install did the trick ? 15:53
[Tux] no, it did the trick for *panda*
lizmat ah, changes needed for CSV ?
grondilu m: constant fact = 1, * * ++$ ... *; say fact[4];
[Tux] see that commit to view what I had to change in CSV
camelia rakudo-moar 92be57: OUTPUT«24␤»
lizmat will do
[Tux] yes :(
15:54 Ven left
grondilu wonders if C<1, * * ++$ ... *;> is better than C<1, |[*] 1 ..*> 15:55
lichtkind lizmat, ah its upper ase now 15:56
thanks
[Tux] NOT happy
lichtkind but why?
[Tux] test 50000 38.704 38.580
test-t 50000 41.004 40.880
lizmat [Tux]: why did you change the first push to append and the third and last one to using a slip ?
[Tux]: when was your last timing 15:57
?
[Tux] yesterday-morning
on the .append I did not know a better notation 15:58
feel free to test and time
lizmat yeah, so I guess 4f7065226c4f28f88 is really to blame for the performance regression
[Tux] 14% slowdown
15:59 Ven joined
[Tux] dinner and then meeting. will be late till I return today (if at all) 15:59
enjoy
lichtkind lizmat++
lizmat lichtkind: because using EVAL means you're doing something potentially naughty, which means it should be more clear from the code 16:04
lichtkind ah thee weird things should look weird doctrine
lizmat yup
lichtkind alright and my second question was if i should position my perl 6 tut (code of first two chapter written) should be in or collective p6 group 16:05
and if yes could someone please create 16:06
16:06 lichtkind_ joined
moritz lichtkind: if you're willing to adhere to some style guidelines, you could place them in perl6/doc 16:06
lichtkind: if not, you must come up with a repo name; nobody knows the topic better than you 16:07
lichtkind no its a tutorial paralleling my already existing p5 tut written for freies magazin
moritz is it meant to be a community project?
lichtkind yes
as long as nobody disagrees with me :9
cognominal m: my token infix:sym<--\>> { "$^a -> $^b\n" }; [~] 'a' X[-->] <b c>
camelia rakudo-moar 92be57: OUTPUT«5===SORRY!5=== Error while compiling /tmp/XLcJTzbXjy␤Prefix -- requires an argument, but no valid term found␤at /tmp/XLcJTzbXjy:1␤------> 3m<--\>> { "$^a -> $^b\n" }; [~] 'a' X[--7⏏5>] <b c>␤ expecting any of:␤ prefix␤»
moritz aka "no" :-)
lichtkind it was a joke
moritz here too 16:08
lichtkind from exprience i know most people will submit typos fixes anyway
cognominal I thought LTM would apply here :(
lichtkind i think about a title like perl6 for problem solver
cognominal what do you think? 16:09
16:09 pmurias left
cognominal m: sub arrow { "$^a -> $^b\n" }; [~] 'a' X[&arrow] <b c> 16:09
camelia ( no output )
cognominal m: sub arrow { "$^a -> $^b\n" }; say [~] 'a' X[&arrow] <b c>
camelia rakudo-moar 92be57: OUTPUT«a -> b␤a -> c␤␤»
lichtkind moritz: something like problem-solver-tutorial ? so we have still ame spaces for other tuts 16:10
16:10 lichtkind_ left
lizmat m: my token infix:sym«-->» { "$^a -> $^b\n" }; [~] 'a' X[-->] <b c> 16:11
camelia rakudo-moar 92be57: OUTPUT«5===SORRY!5=== Error while compiling /tmp/geTg5LpEb8␤Prefix -- requires an argument, but no valid term found␤at /tmp/geTg5LpEb8:1␤------> 3ym«-->» { "$^a -> $^b\n" }; [~] 'a' X[--7⏏5>] <b c>␤ expecting any of:␤ prefix␤»
lizmat cognominal: feels like a rakudobuggable thing to me 16:12
cognominal ok
16:18 tokuhiro_ joined, larion left, rindolf left
cognominal bug sumitted 16:22
16:22 rindolf joined, tokuhiro_ left
Ven
.oO( the bug summit )
 16:23
lizmat fwiw, re 4f7065226c4f28f88 I wonder why .push couldn't keep the (old) semantics, and have .append do the non-flattening way 16:26
tadzik hey, you may find this an easy way to get a nice tee-shirt :) www.digitalocean.com/company/blog/...t-is-back/
TimToady push needs to work well with pop
lizmat is testing a single value variant of .push 16:32
keeping in mind the comments above that :-)
TimToady it also needs to handle slip
lizmat yeah :-)
taking care of that (I hope)
16:37 Ven left 16:38 MARTIMM joined
TimToady m: (*o*)(*.self, *.self)(5).say 16:41
camelia rakudo-moar 92be57: OUTPUT«5␤»
TimToady moritz: ^^
PerlJam If pop is the inverse of push, what's the inverse of append? 16:42
lizmat prepend
TimToady splice
lizmat ? 16:43
skids append is a lossy operation; there is no actual inverse
PerlJam skids: what do you mean that append is lossy?
skids You losethe segmentation between lists
(You can recoup it from the .elems and such, but...) 16:44
PerlJam TimToady: Since splice works on both ends and there's a real low demand for "end aware" removal of multiple elements that make sense I guess
skids: oh, I was just thinkind of "add many" and "remove many" 16:45
skids Well, you can push many.
PerlJam sure, but now it's "natural" to use append for that :)
jnthn lizmat: If push needs to handle slip as TimToady mentioned, note (if you didn't already :)) that the cheapest way will be a multi candidate for Slip :) 16:46
yoleaux 06:27Z <FROGGS> jnthn: why will shaped arrays in future declared as 'my @a[3,3]' but used as '@a[1;1] = 42'? why not semicolons in the decl too?
lizmat jnthn: that's exactly the approach I'm taking :-)
jnthn \o/ 16:47
lizmat++
lizmat jnthn: gist.github.com/lizmat/b4b6d8494bbe295c411a 16:48
feels there is something wrong with the Slip candidate though...
jnthn .tell FROGGS because I thunko; fixed :) 16:49
yoleaux jnthn: I'll pass your message to FROGGS.
jnthn Wait, isn't the push one wrong?
Uh, the \value one
It shouldn't be checking for iscont or Iterable, just unconditionally treaitng it as an item? 16:50
I thought that was the point of the change
lizmat ah... mmm...
jnthn The Slip one actually looks correct
Well, if append-list does the right thing... :)
lizmat method !append-list(@values) 16:51
lizmat tests again 16:53
jnthn: gist.github.com/lizmat/b4b6d8494bb.../revisions
jnthn Seems better, hopefully :) 16:54
lizmat yeah, looks better so far 16:55
PerlJam does append-list return self?
lizmat if it doesn't, it should :-)
PerlJam: yes it does
jnthn m: sub infix:sym<--\>> { "$^a -> $^b\n" }; say [~] 'a' X[-->] <b c> 16:57
camelia rakudo-moar 92be57: OUTPUT«5===SORRY!5=== Error while compiling /tmp/Hptg5mkzZU␤Prefix -- requires an argument, but no valid term found␤at /tmp/Hptg5mkzZU:1␤------> 3\>> { "$^a -> $^b\n" }; say [~] 'a' X[--7⏏5>] <b c>␤ expecting any of:␤ prefix␤»
jnthn m: sub infix:sym<--\>> { "$^a -> $^b\n" }; say 'b' --> 'c'
camelia rakudo-moar 92be57: OUTPUT«5===SORRY!5=== Error while compiling /tmp/yuia2iO3k_␤Confused␤at /tmp/yuia2iO3k_:1␤------> 3ix:sym<--\>> { "$^a -> $^b\n" }; say 'b'7⏏5 --> 'c'␤ expecting any of:␤ postfix␤ statement end␤ statement modifier…»
jnthn The problem isn't the meta-op, I htink tht's that \ isn't an escape there :)
m: sub infix:sym<< --> >> { "$^a -> $^b\n" }; say 'b' --> 'c'
camelia rakudo-moar 92be57: OUTPUT«5===SORRY!5=== Error while compiling /tmp/UaKxYtoKSU␤Confused␤at /tmp/UaKxYtoKSU:1␤------> 3sym<< --> >> { "$^a -> $^b\n" }; say 'b'7⏏5 --> 'c'␤ expecting any of:␤ postfix␤ statement end␤ statement modifier…»
lizmat m: my token infix:sym«-->» { "$^a -> $^b\n" }; [~] 'a' X[-->] <b c> # jnthn: tried this also 16:58
camelia rakudo-moar 92be57: OUTPUT«5===SORRY!5=== Error while compiling /tmp/3nqdOLWn0S␤Prefix -- requires an argument, but no valid term found␤at /tmp/3nqdOLWn0S:1␤------> 3ym«-->» { "$^a -> $^b\n" }; [~] 'a' X[--7⏏5>] <b c> # jnthn: tried this also␤ expecting any …»
16:58 espadrine left
jnthn Uh, token won't do anything useful :) 16:58
It's just wrong
Should be sub :)
I wonder if it's 'cus --> is a terminator though
lizmat m: sub infix:sym«-->» { "$^a -> $^b\n" }; [~] 'a' X[-->] <b c> # jnthn: tried this also
camelia rakudo-moar 92be57: OUTPUT«5===SORRY!5=== Error while compiling /tmp/gGzejOlnSq␤Prefix -- requires an argument, but no valid term found␤at /tmp/gGzejOlnSq:1␤------> 3ym«-->» { "$^a -> $^b\n" }; [~] 'a' X[--7⏏5>] <b c> # jnthn: tried this also␤ expecting any …»
jnthn It's just wrongm: sub infix:sym«~~>» { "$^a -> $^b\n" }; [~] 'a' X[~~>] <b c> 16:59
m: sub infix:sym«~~>» { "$^a -> $^b\n" }; [~] 'a' X[~~>] <b c>
camelia rakudo-moar 92be57: OUTPUT«Type check failed in binding op; expected Any but got Mu␤ in block <unit> at /tmp/2zzZT39x5t:1␤␤»
lizmat ok, push is spectest clean now, fixing some error reporting fallout
jnthn m: sub infix:sym«~~>» { "$^a -> $^b\n" }; say X[~~>] <b c>
camelia rakudo-moar 92be57: OUTPUT«5===SORRY!5=== Error while compiling /tmp/bQGhqWKTdy␤Expected a term, but found either infix ~~ or redundant prefix ~␤ (to suppress this message, please use a space like ~ ~)␤at /tmp/bQGhqWKTdy:1␤------> 3ix:sym«~~>» { "$^a -> $^b\n" }; say…»
jnthn m: sub infix:sym«~~>» { "$^a -> $^b\n" }; say 'b' ~~> 'c'
camelia rakudo-moar 92be57: OUTPUT«5===SORRY!5=== Error while compiling /tmp/F5ahM81TG4␤Variable '&infix:<~~>>' is not declared␤at /tmp/F5ahM81TG4:1␤------> 3ix:sym«~~>» { "$^a -> $^b\n" }; say 'b' 7⏏5~~> 'c'␤»
jnthn wow, there it parses but screw up the symbol installation? 17:00
but yeah, I think --> is "special"
yeah
token terminator:sym<arrow> { '-->' }
So we never even look for an infix, I guess
TimToady the push patch looks to me like it will wrongly flatten .push: [1,2,3] 17:02
jnthn TimToady: After the chnages I asked lizmat++ to do?
TimToady I think I was looking at the last one
lizmat TimToady: if so, then there is no test for it
jnthn TimToady: Now it's a straight nqp::push, which is always single item :) 17:03
TimToady okay, I'll wait and see
17:03 duncan_dmg left
jnthn TimToady: Her first version had the bug you mentioned, so hopefully... :) 17:03
TimToady coulda been looking at the wrong one
TimToady is still backlogging and only paying half a mind
PerlJam does that infix:sym<< >> syntax actually work? I thought it was just infix:<< >> 17:04
dalek kudo/nom: 692aaec | lizmat++ | src/core/Array.pm:
Add a single value Array.push variant
jnthn Oh, that too :)
TimToady infix:sym is only for parse rule names, not operator names 17:05
PerlJam right
jnthn m: sub infix:«~~>» { "$^a -> $^b\n" }; say 'b' ~~> 'c'
lizmat yeah... I'm looking at the self-is-lazy fails messages, making sure that they say "append" when appending, and "pus" when pushing
camelia rakudo-moar 92be57: OUTPUT«b -> c␤␤»
jnthn yes, works fine
lizmat *push
jnthn m: sub infix:«-->» { "$^a -> $^b\n" }; say 'b' --> 'c'
camelia rakudo-moar 92be57: OUTPUT«5===SORRY!5=== Error while compiling /tmp/lkpo9zIN_t␤Confused␤at /tmp/lkpo9zIN_t:1␤------> 3 infix:«-->» { "$^a -> $^b\n" }; say 'b'7⏏5 --> 'c'␤ expecting any of:␤ postfix␤ statement end␤ statement modifier…»
jnthn But that one won't :)
m: sub infix:«~~>» { "$^a -> $^b\n" }; say [~~>] <b c>
camelia rakudo-moar 92be57: OUTPUT«b -> c␤␤»
jnthn There we go, works fine if you use the right syntax
jnthn volunteers somebody to write that on the RT (and I guess rejecting it for using the wrong syntax) while he goes for dinner 17:06
bbl &
TimToady m: sub infix:«~~>» { "$^a -> $^b\n" }; say &[~~>].name
camelia rakudo-moar 92be57: OUTPUT«infix:<~~>>␤»
TimToady we've really gotta fix that to be infix:<~~\>> or some other round-trippable name 17:07
17:11 Ven joined 17:16 Ven left
dalek kudo/nom: a10ec97 | lizmat++ | src/core/Array.pm:
Make sure we mention correct push/append if lazy
 17:16
17:17 dakkar left
Juerd The "Malformed UTF-8" error is surprisingly LTA 17:21
It would help to know where the malformed UTF-8 was encountered. 17:22
TimToady a lot of low-level errors are still rather LTA 17:23
dalek kudo/nom: 469d463 | lizmat++ | src/core/Array.pm:
Add single value candidates for Array.unshift
 17:25
cognominal m: my %a; say %a.^roles 17:30
camelia rakudo-moar 692aae: 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:2666␤ in any apply at gen/moar/m-Metamodel.nqp:2676␤ in any compose at gen/moar/m-Metamodel.nqp:2812␤ in…»
lizmat .tell [Tux] hope I've been able to recoup some of the lost performance
yoleaux lizmat: I'll pass your message to [Tux].
cognominal Again, the metamodel fails as a tutorial device
lizmat goes off to get some Martian air
17:33 Peter_R joined 17:38 baest left 17:40 perlatex joined 17:41 baest joined
perlatex hi 17:41
17:42 FROGGS joined 17:44 kid51 left 17:45 perlatex left 17:47 AlexDaniel joined
FROGGS o/ 17:48
yoleaux 16:49Z <jnthn> FROGGS: because I thunko; fixed :)
FROGGS jnthn: for a moment I thought this msg was about my serialization bug :o)
dalek c: b5886ca | (Zoffix Znet)++ | lib/Language/grammars.pod:
Fix typo
 17:51
17:52 abraxxa left 17:53 Alina-malina left 17:58 Alina-malina joined
tony-o 18:00
yoleaux 11:46Z <Ulti> tony-o: just sent you a little pull request for Slang::SQL for use nqp; there is another issue around the /is rw/ ish changes
tony-o Ulti: merged it in, thanks for the PR
18:01 lichtkind_ joined 18:06 lichtkind_ left, AlexDaniel left 18:11 xfix joined 18:16 larion joined 18:18 yqt joined 18:19 tokuhiro_ joined 18:24 tokuhiro_ left 18:39 ZoffixWork joined
dalek kudo/nom: f52f0db | FROGGS++ | src/Perl6/Grammar.nqp:
call quoting language $~Quote instead of $~Q, RT #126128
 18:40
18:40 [ptc] left
ZoffixWork Is it known where NativeCall loads .dlls from on Windows? I've got AutoItX3.dll I want to load, but regardless of where I tried to put it, this script is telling me it can't find it: fpaste.scsys.co.uk/499720 18:40
18:41 [ptc] joined
FROGGS Zoffix: it looks in CWD and PATH 18:41
ZoffixWork: ^^ 18:42
ZoffixWork Weird. Not finding it. I did try putting it in C:/Windows/System32 and in the same folder as rakudo or my script are at
FROGGS well, then the error is not about not finding it
ZoffixWork Ah
FROGGS is that windows 8? 18:43
ZoffixWork Win7... Actually, I managed to load AutoItX3_x64.dll (or rather, at least I'm not getting any errors). Maybe it's just a 64vs32 issue. 18:44
ZoffixWork tries to call something from AutoItX3_x64
FROGGS yeah, perhaps... 18:45
18:45 andreoss left 18:54 leont joined 19:00 kid51 joined
ZoffixWork Or maybe it's a C++ lib and only C works? I've no idea what Perl6 type to use for LPCWSTR. Str doesn't work 19:00
ZoffixWork doesn't know much about C
Well, Str works, but produces trash, because LPCWSTR is a wide char string.
moritz isn't there some way to specify the encoding with nativecall? 19:02
is NativeCallEncoded['UTF-16'] 19:03
or something like that
19:05 ][Sno][ left
FROGGS the docs say: sub message_box(Str is encoded('utf8')) is native('libgui') { * } 19:05
19:06 vendethiel joined
ZoffixWork Oh, I totally missed that bit. Str is encoded('utf16') works. FROGGS++ moritz++ 19:07
And Perl6++, because I can now use AutoIt without having to deal with AutoIt :P 19:08
FROGGS \o/ 19:11
FROGGS tries to understand S05-capture/array-capture.t and fails 19:12
moritz FROGGS: I don't even have such a test file 19:19
FROGGS: array-alias.t?
FROGGS moritz: err yes, this one 19:21
moritz ok(%$/.keys == 1, 'No extra captures');
bogus precedence?
FROGGS moritz: I am going to push a rewrite
19:23 sufrostico left
ZoffixWork I'm not following the signatures.... shouldn't sub send(Str, Int $mode? ) {} work for send("foo", 42) as well as send("foo")? I'm being told it'll never work: fpaste.scsys.co.uk/499729 19:29
19:30 sufrostico joined
jnthn ...wtf 19:30
m: sub send(Str, Int $?) { }; send('x', 42)
camelia ( no output )
jnthn m: use NativeCall; sub send(Str, Int $?) is native('libfoo') { }; send('x', 42) 19:31
camelia rakudo-moar f52f0d: OUTPUT«Cannot locate native library 'libfoo.so': libfoo.so: cannot open shared object file: No such file or directory␤ in method setup at lib/NativeCall.pm:225␤ in method CALL-ME at lib/NativeCall.pm:236␤ in block <unit> at /tmp/lZjIi72bdU:1␤␤»
jnthn m: use NativeCall; sub send(Str is encoded('utf16'), Int $?) is native('libfoo') { }; send('x', 42)
camelia rakudo-moar f52f0d: OUTPUT«Cannot locate native library 'libfoo.so': libfoo.so: cannot open shared object file: No such file or directory␤ in method setup at lib/NativeCall.pm:225␤ in method CALL-ME at lib/NativeCall.pm:236␤ in block <unit> at /tmp/8DRVKjCltM:1␤␤»
ZoffixWork Is it because I'm using Windows rakudo that's likely outdated? rakudo-star-2015.06-x86_64 (JIT).msi
jnthn Maybe, but it's still a bizzare thing to happen... 19:32
FROGGS star-m: sub send(Str, Int $?) { }; send('x', 42)
camelia ( no output )
jnthn Oh wait, it's a runtime error?
The "will never work" is usually compile time. That's very confusing.
FROGGS ummm 19:33
jnthn epic fail 19:34
FROGGS ZoffixWork: you cannot use optional params anyway with native subs atm
jnthn if nqp::elems($args) != $r.signature.arity {
ZoffixWork FROGGS, ah :(
FROGGS ZoffixWork: create two multies, one with the second param and one without
ZoffixWork OK. I'll try
jnthn It's doing that then throwing... 19:35
Oh, I guess we don't do varargs yet either
So we get away with that check...for now
[Tux] builds ...
yoleaux 17:30Z <lizmat> [Tux]: hope I've been able to recoup some of the lost performance
FROGGS jnthn: aye, though it would the wrong thing if it got passed that check
would do*
jnthn *nod*
Better would be to detect the optional/named at the point we apply the trait or do the setup 19:36
FROGGS aye
19:38 tokuhiro_ joined
ZoffixWork Doesn't look like you can do multies with NativeCall either. Getting "cannot invoke object": fpaste.scsys.co.uk/499731 19:42
FROGGS it should work though, I already did that 19:43
perhaps add a: proto send(|) { }
19:44 tokuhiro_ left
[Tux] test 50000 38.978 38.856 19:45
test-t 50000 38.680 38.558
19:48 loveJesus joined
ZoffixWork Not exactly sure what that does, but adding it before multies runs the code with no errors, but the DLL call doesn't seem to be made: fpaste.scsys.co.uk/499732 19:48
If I use just one send with just Str sig, then the call does work
19:48 _dolmen_ joined
FROGGS Zoffix: I think you signature should be: sub send(Str is encoded('utf16'), Int $a = 0 ) ... 19:50
without the other candidate and without the proto
ZoffixWork No, I already tried that earlier. I get "Cannot unbox a type object in method CALL-ME at lib/NativeCall.pm:248 in block <unit> at nci.p6:13" 19:51
Same error as with sub send(Str is encoded('utf16'), Int $a? ) 19:52
FROGGS strange
19:53 Possum left
FROGGS m: say "abcxyd" ~~ m/@<foo>=(.)/ # a brainfart of mine 19:53
camelia rakudo-moar f52f0d: OUTPUT«｢abcxyd｣␤ foo => ｢a｣␤ foo => ｢b｣␤ foo => ｢c｣␤ foo => ｢x｣␤ foo => ｢y｣␤ foo => ｢d｣␤»
moritz FROGGS: you implied a quantifier where there was none? 19:55
PerlJam implied a quantifier?
FROGGS I meant to trick it into create an array... but I think I can fix that 19:56
19:57 [ptc] left
FROGGS yeah, easy fix is easy 19:58
dalek ast: 8c5b0dc | usev6++ | S03-metaops/cross.t:
Fudge test which passed for the wrong reason
 19:59
p: 8402ce0 | FROGGS++ | src/QRegex/P6Regex/Actions.nqp:
fix fake quantifier for array captures
 20:00
20:04 travis-ci joined
travis-ci NQP build failed. Tobias Leich 'fix fake quantifier for array captures' 20:04
travis-ci.org/perl6/nqp/builds/83186346 github.com/perl6/nqp/compare/eaa6e...02ce098192
20:04 travis-ci left 20:06 xfix_ joined
FROGGS ENOTMYFAULT 20:06
20:07 darutoko left
ZoffixWork :D 20:07
20:08 xfix left 20:23 [Sno] joined 20:28 xfix_ left
dalek kudo/nom: 11c84a4 | FROGGS++ | tools/build/NQP_REVISION:
bump nqp for array capture fix
 20:28
kudo/nom: 77e58e4 | FROGGS++ | t/spectest.data:
run S05-capture/array-alias.t
ast: eec7994 | FROGGS++ | S05-capture/array-alias.t:
reqrite S05-capture/array-alias.t

I hope the changed semantics meet out current expectations. Please review carefully.		 20:29
FROGGS reqrite
FROGGS sighs 20:30
our*
such typo
20:31 sufrostico left 20:32 kid51 left, _dolmen_ left 20:33 vendethiel left
FROGGS gnight 20:34
20:34 FROGGS left 20:38 sufrostico joined
PerlJam FROGGS++ 20:38
20:41 vendethiel joined
lichtkind jnthn++ 20:42
20:42 colomon left 20:46 colomon joined 20:49 mtj_ left 20:51 sunnavy joined
lizmat [Tux]: :-( 20:53
jnthn: could it be we lost some optimizations due to addition of append/prepend ? 20:54
20:54 kaare_ left 20:57 sunnavy left 20:58 sunnavy joined, mtj_ joined
lichtkind jnthn thanks of the fix 20:59
21:01 petercommand joined
dalek ast: 927d805 | usev6++ | S32-hash/delete.t:
Fudge test that passed for the wrong reason
 21:01
ast: 3cbb371 | usev6++ | S (7 files):
Replace remaining eval-dies-ok with throws-like
 21:02
ast: 3e1099a | TimToady++ | S07-slip/slip.t:
slip tests
jnthn
.oO( rewuiem for FROGGS typing skills... )
dalek kudo/nom: 8bc8467 | TimToady++ | src/core/Slip.pm:
delete incorrect, duplicate code

  (If we really need to dup code for performance, it should do one-arg
lazy semantics correctly as the new slip.t tests now test for.)
kudo/nom: 0f62d9f | TimToady++ | t/spectest.data:
add S07-slip/slip.t
jnthn lizmat: Dunno, but TimToady's (useful) semantic cleanup work isn't always very performance-aware ;-) 21:03
Being fast and wrong isn't very useful though, so souldn't complain too much :) 21:04
Anyway, no point wasting time guessing what's slowed down. Just --profile it
And look at the two side by side
jnthn used the profiler to find many improvements to Rakudo to improve Text::CSV so far :) 21:05
lizmat my @a; @a.push(42) for (^100) # 65% of time spent in sink-all 21:07
TimToady m: my \c = 0, slip [\+] 1..*; say c[100000]
camelia rakudo-moar 77e58e: OUTPUT«Nil␤»
TimToady that should fix itself now
lizmat I was surprised to learn that for internally maps to map, and that to sequential-map
(and then sink-all in there) 21:08
leont Had a weird issue where an array appeared to be empty but true, still figuring it out :-s
21:08 MARTIMM left
lizmat if we're running a for of which we *know* we want the body just executed for its side effects, why do we need to run it through map with sink-all ? 21:08
21:09 telex left
leont Ah, apparently it contained a [[],] 21:09
jnthn lizmat: I think I made the map iterator special-case sink-all already to not care for the results
lizmat well, it's doing a repeated $source.pull-one instead of a push-all ? 21:10
21:10 telex joined
jnthn Indeed. github.com/rakudo/rakudo/blob/nom/...ds.pm#L158 21:10
Uh
How could it push-all?
21:10 skids left
jnthn That makes no sense. 21:10
lizmat I meant sink-all 21:11
I mean, if it can do a pull-one, it can do a sink-all, no?
jnthn ?
sink means "I don't want the values"
in for @foo -> $n { } we certainly want the values :)
(to pass for $n) 21:12
lizmat ah, yes: duh :-)
jnthn In the particular case of 'for ^100' though, I'm a bit surprised the "re-write to a while loop" opt doesn't kick in.
So for that particular range case, something is off
Maybe we don't nail it for the statement modifier form somehow 21:13
lizmat for ^100 { } gives very similar results 21:14
looks like it's still doing a map there 21:15
a manual while loop gives *very* different results 21:16
21:28 ZoffixWork left, espadrine joined
lizmat is looking at the range -> while optimizer 21:29
timotimo jnthn: the "rewrite for to while loop" opt is currently bust 21:31
because the qast to build for loops has changed quite a bit
and i didn't yet bother to come up with an elegant solution to keep the object around that we iterate over
because it's now a sibling to the for loop rather than a child :\
lizmat timotimo: you're saying you know what the problem is but don't know how to fix it ? 21:32
21:32 larion left
jnthn timotimo: ah, ok 21:32
sleep time....'night, #perl6 21:33
timotimo i know how to fix it, just not how to fix it well
lizmat gnight, jnthn
timotimo i can explain easier when we both bring up the qast dump of 'for ^100 { }'
21:34 espadrine left
timotimo then you'll see there's a bind op with a local called "for-list_1" 21:34
that's then used in map & sink
hey, what kind of operator would we use ſ for?
21:36 dha joined, dha left 21:37 dha joined
lizmat timotimo: looking at it 21:37
21:39 larion joined
lizmat timotimo: the first condition that fails, is @($op[0]) == 2 21:40
timotimo i suppose it'd be enough to store a list of the things for-list_* is bound to
the thing is, when you notice you're looking at a sink(map(...)), you can't just go to the previous sibling, because there's no reference to the parent
dalek ast: 33182cf | usev6++ | S02-types/ (2 files):
Start to replace dies-ok with throws-like (S02-types)

Thereby unfudge one passing test
timotimo unless we do something terrible like nqp::ctxlexpad(nqp::callerctx(nqp::ctx))('$node') or whatever 21:41
21:41 tokuhiro_ joined
lizmat timotimo: I'm afraid I lost you already :-( 21:42
21:45 tokuhiro_ left
timotimo i suppose you're looking at the existing code to do the for-to-while optimization, yeah? 21:47
lizmat yeah
I assume the "optimize_for_range" method will need adjustment as well, then 21:48
timotimo yes 21:51
lizmat ok, I guess I'm definitely out of my league then 21:52
timotimo i was thinking when we see a bind with a local that matches for-list_(\d+) we take that number as an index into a list that stores the upper and lower bounds we've extracted from the range (if it has a range at all)
lizmat I wonder if the initial code-gen couldn't more easily leave a marker 21:54
that would either be removed during optimization, or replace the optimization ?
masak 'night, #perl6 22:09
lizmat gnight, masak 22:12
colomon \o 22:15
22:17 cognominal left 22:24 dha left
lizmat hmmm.. I just realize we lost 15% in bare startup time as well :-( 22:26
around 115 msecs, was below 100 :-( 22:27
22:28 larion left
timotimo nine explained to me why the bind was important in the code-gen, but i forgot 22:42
perhaps there really is a way to remove the bind if we know we'll be using the .. operator (or its friends) from the core setting
22:42 larion joined
lizmat FWIW, I'm optimizing the Range iterator itself 22:51
timotimo that's good for every case where we don't have access to the range in question at compile time
which is every benchmark that takes the number of runs to do from ARGS, for example 22:52
lizmat a bare start contains at least two Range.push-all 's 22:53
timotimo oh, i didn't realize 22:54
that's good
22:54 lichtkind left
lizmat I wonder where they are... 22:54
timotimo lifts a rock
nope, not here
lizmat :-)
those pesky ^n 's
timotimo yeah 22:55
dalek kudo/nom: e7ec205 | lizmat++ | src/core/Range.pm:
Optimize the finite integer Range iterator

my @a = ^100000 is now ~ 30% faster, (^100000).map: *.Str about 15%		 23:03
ast: 870186b | lizmat++ | S05-metasyntax/angle-brackets.t:
Remove ) straggler
 23:04
23:05 Peter_R left
lizmat calls it a day 23:05
23:06 rindolf left
lizmat so good night, #perl6! 23:06
23:10 lichtkind joined 23:11 TEttinger joined 23:12 bjz left, BenGoldberg joined
BenGoldberg m: sub f { $^x + 1 }; sub g { $^x * 2 }; (f o g).say; 23:13
camelia rakudo-moar 0f62d9: OUTPUT«5===SORRY!5=== Error while compiling /tmp/9k2vLglGK_␤Undeclared routine:␤ o used at line 1␤␤»
BenGoldberg m: sub f { $^x + 1 }; sub g { $^x * 2 }; (&f o &g).say;
camelia rakudo-moar 0f62d9: OUTPUT«-> |args is raw { #`(Block|73959776) ... }␤»
BenGoldberg m: sub f { $^x + 1 }; sub g { $^x * 2 }; (&f o &g)(3).say; 23:14
camelia rakudo-moar 0f62d9: OUTPUT«7␤»
23:15 zakharyas joined 23:25 mr_ron joined 23:30 kid51 joined
mr_ron m: kill($*PID, "INT") 23:34
camelia rakudo-moar e7ec20: OUTPUT«5===SORRY!5=== Error while compiling /tmp/J2YnneE5vV␤Undeclared routine:␤ kill used at line 1␤␤»
mr_ron anyone know how to signal/kill your own process with Perl 6? 23:35
23:36 zakharyas left
timotimo .tell lizmat i wonder if push-exactly should check for "will we reach the end?" at the beginning and then perhaps use a tiny bit better code? 23:36
yoleaux timotimo: I'll pass your message to lizmat.
mr_ron It looks like a usable kill function is included in S29 but I didn't notice a test for that kill function int the S29 area of roast ... 23:37
23:42 tokuhiro_ joined
TimToady well, run("kill", $pid) will work on Unix 23:45
but we should do better
run("kill", $*PID) to kill your own process 23:46
23:47 tokuhiro_ left
TimToady m: use NativeCall; sub kill(int $proc, int $sig) is native("libc") {*}; kill($*PID,1); sleep 1 23:49
camelia rakudo-moar e7ec20: OUTPUT«Cannot locate native library 'libc.so': /usr/lib64/libc.so: invalid ELF header␤ in method setup at lib/NativeCall.pm:225␤ in method CALL-ME at lib/NativeCall.pm:236␤ in block <unit> at /tmp/lyzt5zrvBb:1␤␤»
TimToady hmm
BenGoldberg m: use NativeCall; sub kill(int $proc, int $sig) is native("") {*}; kill($*PID,1); sleep 1 23:51
camelia rakudo-moar e7ec20: OUTPUT«Cannot locate native library '.so': .so: cannot open shared object file: No such file or directory␤ in method setup at lib/NativeCall.pm:225␤ in method CALL-ME at lib/NativeCall.pm:236␤ in block <unit> at /tmp/BQdHzHUjPV:1␤␤»
23:51 vendethiel left
BenGoldberg m: use NativeCall; sub kill(int $proc, int $sig) is native(Str) {*}; kill($*PID,1); sleep 1 23:53
camelia rakudo-moar e7ec20: OUTPUT«(signal HUP)»
BenGoldberg That's not an INT!
23:53 RabidGravy left
mr_ron It's not that important ... I'll find a way around it. It looks like only a small part of S29/OS is tested in roast. It looks obvious enough not to need an RT but if someone likes I'll write one up ... 23:54
TimToady I think it's important to get it on the by-Christmas list
m: use NativeCall; sub kill(Int $proc, Int $sig) is native {*}; kill($*PID,2); sleep 1 23:55
camelia rakudo-moar e7ec20: OUTPUT«(signal INT)»
TimToady I guess just leave off the argument entirely 23:56
j: use NativeCall; sub kill(Int $proc, Int $sig) is native {*}; kill($*PID,2); sleep 1 23:57
camelia ( no output )

Powered by Cro and the Raku® Programming Language.

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

Thank you!