»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg p6eval perl6: ... | irclog: irc.perl6.org/ | UTF-8 is our friend!
Set by sorear on 4 February 2011.
00:08 awwaiid left 00:09 awwaiid joined
dalek ecza: eeed873 | sorear++ | docs/announce.v14:
Finish writing up announce.v14
 00:09
01:04 Chillance left 01:28 trippeh_ left
dalek ast: 330bd62 | (Solomon Foster)++ | S02-magicals/env.t:
Tweaks to avoid reams of "Use of uninitialized value in string context" errors.
 01:40
01:41 whiteknight joined 01:44 aaaswwefff joined 01:51 aaaswwefff left 01:56 kfo_ left 01:57 kfo joined
sorear I try to reserve the word 'error' for situations where the computation stops 02:00
phenny: tell pmurias If you see this message before Monday, I'd like it if you could give a better description of your changes for announce.v14 02:01
phenny sorear: I'll pass that on when pmurias is around.
sorear anyone else who wants to look at announce.v14, please go ahead 02:03
diakopter sorear: Parameters like @foo and %foo now insist that their arguments to Positional and Associative respectively. missing a verb? 02:04
02:13 PZt left
Yary Hi all... I was hoping to initialize a "lazy" attribute using Rakudo but I seem to be doing it wrong- 02:16
rakudo: class B {has @.x; method new(Int $y) {self.bless(*);@.x <== (0..$y)}}
p6eval rakudo 9719f7: OUTPUT«===SORRY!===␤Sorry, do not know how to handle this case of a feed operator yet. at line 1, near " (0..$y)}}"␤»
Yary is there a way around that?
dalek ecza: 830a9ed | sorear++ | docs/announce.v14:
Fix verb (diakopter)
 02:19
sorear Yary: keep it simple
Yary: <== is a parallel processing operator, it is not "simple" by any stretch
Yary was just looking at the example in Iterators on initializing a range 02:20
s07
sorear S07 is completely obsolete
Yary well my code will work with eager semantics, easily 02:21
so I can use
sorear <== has nothing to do with laziness anymore 02:22
Yary rakudo: class B {has @.x; method new(Int $y) {self.bless(*, x=> (0..$y)}}
p6eval rakudo 9719f7: OUTPUT«===SORRY!===␤Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 2␤»
Yary typo
sorear <== is the opposite of lazy, it will start extra threads to fill your array
Yary rakudo: class B {has @.x; method new(Int $y) {self.bless(*, x=> (0..$y))}}
p6eval rakudo 9719f7: ( no output )
Yary any pointers for up-to-date lazy/eager docs?
rakudo: class B {has @.x; method new(Int $y) {self.bless(*, x=> (0..$y))}}; my $b=B.new(3);say $b.x 02:23
p6eval rakudo 9719f7: OUTPUT«0 1 2 3␤»
02:25 ponbiki joined 02:27 birdwindupbird joined
[Coke] # 01/29/2012 - niecza at 99.52% 02:31
b: say 19548 - 19456
p6eval b 1b7dd1: OUTPUT«92␤»
02:32 birdwindupbird left
sorear [Coke]: do you have a breakdown by synopsis for that? is rakudo gaining or niecza losing? 02:37
Yary: no
[Coke] I don't keep the synopsis breakdown.
I have the latest numbers and can plug them into the google spreadsheet. lemme do that.
Updated: docs.google.com/spreadsheet/ccc?ke...URFM1A0cHc 02:42
rakudo's biggest lead is in S12.
(Objects) 02:43
colomon it's all the metaprogramming and introspection
[Coke] niecza is leading in S32, even without Temporal.
colomon that's just the FatRat tests. :)
Yary I like s07, 0 vs 0, I'll have to remember to take that one with a grain of salt for now 02:47
TimToady sorear: it still sez v13 on the first line 02:48
[Coke] TimToady: good catch. ;) 02:51
02:53 ponbiki left
dalek ecza: ff52ee4 | sorear++ | docs/announce.v14:
s/13/14/ TimToady++
 02:56
02:56 ponbiki joined 03:00 whiteknight left 03:06 skids left
colomon Dang, that is an utterly insane list of niecza progress in the last month. 03:08
03:19 orafu left, orafu joined
TimToady hmm, not a criticism of the document, but infix operators returning Bool should default to iffy, and ones where the returned type is different from the left argument should default to diffy 03:19
03:32 kmwallio joined 03:44 wolfman2000 joined
wolfman2000 masak: you awake? 03:47
03:48 kmwallio left
Yary Hello again, world, basic question- how do I override my class' stringification & numification? 03:49
or more accurately, change how objects in that class stringify/numify? 03:53
TimToady wolfman2000: doubtful, he's in Sweden 03:54
wolfman2000 right, I forgot that.
I was hoping to see what updates were made to his web framework that I recall he was working on. 03:55
TimToady Yary: typically we've just defined Str or Num methods, though we might generalize that at some point 04:02
Yary OK I'll try that, thanks
TimToady they should probably be submethods, as should .perl and .gist unless you're willing to speak for subclasses as well
04:05 lutok joined 04:09 Radvendii left 04:11 lutok left 04:36 Trashlord left 04:37 birdwindupbird joined
moritz \o 04:38
offby1 _/o
sorear \o 04:41
04:51 Trashlord joined
eviltwin_b /o\ 05:03
offby1 pours rubbish on eviltwin_b, since he's already taken cover 05:07
05:11 ponbiki left 05:57 alvis left 06:00 shinobicl_ left 06:12 alvis joined 06:14 kaleem joined 06:22 lestrrat left 06:23 lestrrat joined 06:27 PZt joined 06:33 xinming left 06:54 xinming joined 06:59 wtw joined 07:01 zby_home_ left 07:07 sftp joined 07:28 flussence left 07:29 flussence joined 07:35 passr joined, am0c_ joined
passr perl6: say [!=] 1,2,3,1 07:35
p6eval rakudo 9719f7: OUTPUT«Bool::True␤»
..niecza v13-381-g4158fa9: OUTPUT«Unhandled exception: assigning to readonly value␤ at /tmp/fYOEjkWoSB line 0 (ANON @ 1) ␤ at /home/p6eval/niecza/lib/CORE.setting line 2276 (reduceop @ 17) ␤ at /tmp/fYOEjkWoSB line 1 (mainline @ 2) ␤ at /home/p6eval/niecza/lib/CORE.setting line 3609 (…
..pugs b927740: OUTPUT«1␤»
07:35 am0c left 07:36 am0c_ is now known as am0c
passr hi perl6! why rakudo say true?say [!=] 1,2,3,1 07:37
moritz because it's the same as 1 != 2 != 3 != 1, which in turn in the same as (1 != 2) && (2 != 3) && (3 != 1) 07:39
which is True
passr moritz: If I want test all the numbers all diffrent, how to code? 07:43
moritz passr: you could use a hash, for example
rakudo: sub all_uniq(*@a) { my %h; for @a { return False if %h{$_}++ }; True }; say all_uniq(<1 2 3>), all_uniq(<1 2 3 1>) 07:44
p6eval rakudo 9719f7: OUTPUT«Bool::TrueBool::False␤»
moritz rakudo: sub all_uniq(*@a) { @a.uniq.elems == @a.elems}; say all_uniq(<1 2 3>), all_uniq(<1 2 3 1>)
p6eval rakudo 9719f7: OUTPUT«Bool::TrueBool::False␤»
passr moritz:thanks! 07:45
moritz rakudo: sub all_uniq(*@a) { so all(@a) == one(@a) }; say all_uniq(<1 2 3>), all_uniq(<1 2 3 1>)
p6eval rakudo 9719f7: OUTPUT«Bool::TrueBool::False␤»
Timbus i like that last one 07:53
its the solution i would have came up with, but at the same time, wouldn't have known how to do it 07:55
sorear welcome, passr 08:11
passr sorear:hi~ 08:12
08:24 snearch joined
passr rakudo: say ([1,2,3],[1,2,2],[4,5,6]).grep: {so all @$_ == one @$_} 08:27
p6eval rakudo 9719f7: OUTPUT«1 2 3␤»
moritz hrmpf 08:28
uses hashes then :-)
passr moritz: something wrong? 08:29
masak mornink, #perl6 08:33
passr: useless use of 'so'
passr moritz: thanks! 08:34
masak it makes sense in the all_uniq sub, not so much in the grep, which already boolifies for you.
08:34 mj41 joined
moritz shouldn't that have also returned [4, 5, 6] ? 08:35
passr but no output? 08:36
Timbus nom: say all @$_ == one @$_ for ([1,2,3]) 08:37
p6eval nom 9719f7: OUTPUT«all(one(Bool::False, Bool::False, Bool::True))␤»
Timbus got intified
08:40 TimToady left, daxim joined 08:41 TimToady joined
Timbus does one() not work on arrays or something or 08:41
08:41 mj41 left
passr maybe? 08:42
masak nom: say all @$_ for [1, 2, 3]
p6eval nom 9719f7: OUTPUT«all(1, 2, 3)␤»
masak nom: say one @$_ for [1, 2, 3]
p6eval nom 9719f7: OUTPUT«one(1, 2, 3)␤»
masak nom: say all([1,2,3]) == one([1,2,3]) 08:43
p6eval nom 9719f7: OUTPUT«all(one(Bool::True))␤»
masak nom: say all(1,2,3) == one(1,2,3)
p6eval nom 9719f7: OUTPUT«all(one(Bool::True, Bool::False, Bool::False), one(Bool::False, Bool::True, Bool::False), one(Bool::False, Bool::False, Bool::True))␤»
masak nom: say all([1,2,3].list) == one([1,2,3].list) 08:44
p6eval nom 9719f7: OUTPUT«all(one(Bool::True, Bool::False, Bool::False), one(Bool::False, Bool::True, Bool::False), one(Bool::False, Bool::False, Bool::True))␤»
Timbus holds his head and walks away
08:44 daemon left
masak that, believe it or not, is the correct response to anything junction-y. 08:44
Timbus yes but
masak use them below the headache threshold :)
08:45 cosimo left
Timbus nom: say all @$_ == one @$_ given [1,2,3] 08:45
p6eval nom 9719f7: OUTPUT«all(one(Bool::False, Bool::False, Bool::True))␤»
Timbus whats up with that
better shown
masak nom: say all(@$_) == one(@$_) given [1,2,3]
p6eval nom 9719f7: OUTPUT«all(one(Bool::True, Bool::False, Bool::False), one(Bool::False, Bool::True, Bool::False), one(Bool::False, Bool::False, Bool::True))␤»
masak Timbus: precedence.
Timbus nom: say all @$_ == one @$_ given [9,3,6]
p6eval nom 9719f7: OUTPUT«all(one(Bool::False, Bool::True, Bool::False))␤»
Timbus oh 08:46
masak Timbus: 'all' is a listop, so it eats '@$_ == one @$_'
case closed. :)
use parens.
08:46 cosimo joined
Timbus arg 08:46
masak .oo( use parens for you arg ) 08:47
your*
sorear all(@$_) perhaps better written .all
08:48 daemon joined
Timbus i blame passr for everything that just happened :I 08:48
masak :D
passr Thanks!Timbus,masak.
sorear -> sleep
passr masak: can you write any shorter code than that? 08:53
masak nom: say .all == .one given [1, 2, 3] 08:54
p6eval nom 9719f7: OUTPUT«all(one(Bool::True, Bool::False, Bool::False), one(Bool::False, Bool::True, Bool::False), one(Bool::False, Bool::False, Bool::True))␤»
masak nom: say so .all == .one given [1, 2, 3]
p6eval nom 9719f7: OUTPUT«Bool::True␤»
masak by sorear++'s suggestion. 08:55
passr wow!
08:55 lestrrat left, lestrrat joined 08:57 mj41 joined
moritz nom: say so .all == .one given [1, 2, 4, 1] 09:13
p6eval nom 9719f7: OUTPUT«Bool::False␤»
passr masak: how to write it shorter.about this code: my @a = ^4; my @b = @a X~ @a X~ @a X~ @a
masak niecza: my @a = ^4; @a = @a X~ @a for ^2; say @a 09:17
p6eval niecza v13-381-g4158fa9: OUTPUT«0000 0001 0002 0003 0010 0011 0012 0013 0020 0021 0022 0023 0030 0031 0032 0033 0100 0101 0102 0103 0110 0111 0112 0113 0120 0121 0122 0123 0130 0131 0132 0133 0200 0201 0202 0203 0210 0211 0212 0213 0220 0221 0222 0223 0230 0231 0232 0233 0300 0301 0302 …
masak niecza: my @a = ^4; @a X~= @a for ^2; say @a 09:18
p6eval niecza v13-381-g4158fa9: OUTPUT«0012300123100123100123232001231001231001232320012310012310012323330012310012310012323200123100123100123232001231001231001232333001231001231001232320012310012310012323200123100123100123233 1001231001232300123001231001231001232320012310012310012323200123100…
09:18 takesako joined
masak niecza: my @a = ^4; @a [X~]= @a for ^2; say @a 09:18
p6eval niecza v13-381-g4158fa9: OUTPUT«0000 0001 0002 0003 0010 0011 0012 0013 0020 0021 0022 0023 0030 0031 0032 0033 0100 0101 0102 0103 0110 0111 0112 0113 0120 0121 0122 0123 0130 0131 0132 0133 0200 0201 0202 0203 0210 0211 0212 0213 0220 0221 0222 0223 0230 0231 0232 0233 0300 0301 0302 …
daxim d.hatena.ne.jp/tokuhirom/20120130/1327912149 09:19
jnthn morning, #perl6
masak niecza: my @a = (@a = ^4) X~ @a; say @a
p6eval niecza v13-381-g4158fa9: OUTPUT«00 01 02 03 10 11 12 13 20 21 22 23 30 31 32 33␤»
masak niecza: my @a = (@a = ((@a = ^4) X~ @a) X~ @a); say @a
p6eval niecza v13-381-g4158fa9: OUTPUT«000 001 002 003 010 011 012 013 020 021 022 023 030 031 032 033 100 101 102 103 110 111 112 113 120 121 122 123 130 131 132 133 200 201 202 203 210 211 212 213 220 221 222 223 230 231 232 233 300 301 302 303 310 311 312 313 320 321 322 323 330 331 332 333…
masak niecza: my @a = ((@a = ((@a = ^4) X~ @a)) X~ @a); say @a 09:20
p6eval niecza v13-381-g4158fa9: OUTPUT«0000 0001 0002 0003 0010 0011 0012 0013 0020 0021 0022 0023 0030 0031 0032 0033 0100 0101 0102 0103 0110 0111 0112 0113 0120 0121 0122 0123 0130 0131 0132 0133 0200 0201 0202 0203 0210 0211 0212 0213 0220 0221 0222 0223 0230 0231 0232 0233 0300 0301 0302 …
09:22 dakkar joined 09:23 snearch left
moritz nom: print .base(4).fmt('%04s ') for ^(4**4) 09:26
p6eval nom 9719f7: OUTPUT«0000 0001 0002 0003 0010 0011 0012 0013 0020 0021 0022 0023 0030 0031 0032 0033 0100 0101 0102 0103 0110 0111 0112 0113 0120 0121 0122 0123 0130 0131 0132 0133 0200 0201 0202 0203 0210 0211 0212 0213 0220 0221 0222 0223 0230 0231 0232 0233 0300 0301 0302 0303 0310 031…
jnthn :) 09:27
moritz is that considered cheating? :-)
masak it's indistinguishable from the real solution :P
tadzik good morning! 09:29
moritz o/ tadzik 09:30
jnthn ...and easier to read. :)
tadzik: Wow, tadzik *in the morning*!
;)
masak (tadzik in the morning)++ 09:31
nom: say (.base(4) for ^256).fmt('%04s') 09:32
p6eval nom 9719f7: OUTPUT«0000 0001 0002 0003 0010 0011 0012 0013 0020 0021 0022 0023 0030 0031 0032 0033 0100 0101 0102 0103 0110 0111 0112 0113 0120 0121 0122 0123 0130 0131 0132 0133 0200 0201 0202 0203 0210 0211 0212 0213 0220 0221 0222 0223 0230 0231 0232 0233 0300 0301 0302 0303 0310 031…
tadzik Achievement unlocked! :)
passr Is it impossible use reduce operator? [ ] 09:34
moritz I don't think so. 09:35
passr it must be shortest,I think? 09:36
moritz must it be?
nom: say [X~] [^4] xx 4
p6eval nom 9719f7: OUTPUT«0000 0001 0002 0003 0010 0011 0012 0013 0020 0021 0022 0023 0030 0031 0032 0033 0100 0101 0102 0103 0110 0111 0112 0113 0120 0121 0122 0123 0130 0131 0132 0133 0200 0201 0202 0203 0210 0211 0212 0213 0220 0221 0222 0223 0230 0231 0232 0233 0300 0301 0302 0303 0310 031…
moritz maybe :-)
passr I love it !! 09:37
moritz: That's all I need!Thanks! 09:40
masak moritz++ 09:42
passr masak++ too 09:44
masak nom: say [X~] ^4 xx 4
p6eval nom 9719f7: OUTPUT«0123012301230123␤»
masak just had to try :)
09:47 Psyche^ joined 09:48 Patterner left, Psyche^ is now known as Patterner
passr masak:precedence? 09:52
masak nom: say ^4 xx 4 09:53
p6eval nom 9719f7: OUTPUT«0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3␤»
moritz flattening, not precedence
masak passr: more like, without the array, you don't get X~ to cross four arrays against each other. 09:54
passr thanks guys 09:57
Is it impossible use reduce operator? ([1,2,3],[1,2,2],[4,5,6]).grep: {so all(@$_) == one(@$_)} 10:05
masak useless use of 'so' again.
moritz passr: I don't think it's impossible. But I don't want to spoil your fun when you try to figure it out on your own 10:06
passr :masak:thanks 10:07
moritz: Can you try to do that?
Timbus heh
passr o/ 10:08
moritz passr: not spoil your fun? sure thing.
10:09 am0c left
passr don't worry. moritz.i just want to find a shortest code for that.but I can't do that myself. 10:10
masak sure you can!
moritz passr: why not?
masak but why are you so obsessed with short code?
there are so many other things to optimize for. 10:11
passr I love short code
masak the ability to grok the code tomorrow, for example :)
passr moritz: If Ican do that ,I 'll not ask question here. 10:14
moritz passr: it's not an either-or 10:15
passr: you can try, show us some partial successes, and then tell us why you can't get further
passr: right now it juts feels like "golf that for me!", which is not the friendliest way to interact with us 10:16
passr moritz:I'll try
masak it's not unfriendly as such. just a bit lazy. :) 10:17
passr masak:not lazy,but stupid. 10:18
masak: not so smart as you 10:19
masak passr: I refuse to believe that, since I know my brain from the inside :) 10:20
the trick isn't to be smart all the time. the trick is to notice when you have smart days, or smart afternoons, and do something cool then. :) 10:21
I've been planning to put out a blog post saying "we're all stupid sometimes"
passr masak: You are my teacher
masak tools like RT and git and the test suite really help to even out the smart/stupid fluctuations.
Timbus another illegitimate student
D: 10:22
moritz the trick is also not to think of yourself as being stupid, because you start to believe it, and feel discouraged to try
masak Timbus: I was thinking that -- "oh no, not again" :P
moritz the trick is also not to think of yourself as being smart, because it implies some hard-bounded limit that you can't bend
which isn't the case
masak also, if you think you're smart, you might not be humble enough to learn when you need to. 10:23
moritz the trick is to think yourself of exploring, because that's what you should do
masak development. it's an exploration.
Timbus the only time im smart is when im doing something fundamentally dumb :<
see: my independent discovery of sleep sort rosettacode.org/wiki/Sorting_algori..._sort#Perl
moritz there have been interesting behavioral studies involving children who solved some tasks, and then one group was praised for being smart, and the other for working hard on the problem 10:24
the first group, when faced with a more difficult task, gave up much earlier, and thust performed worse
Timbus the "smart" children got a payrise, the "hard working" ones lobbied for one but didnt get it 10:25
passr Timbus: I love it .but i can't understand that code at the first time 10:26
10:27 lestrrat left 10:28 lestrrat joined 10:32 lestrrat left 10:33 lestrrat joined 10:35 passr left
masak everybody, thank you so much for your p6cc solutions. 10:37
moritz indeed. It is a joy to read, test and review 10:38
masak moritz and I are having way too much fun analyzing them and comparing them.
11:02 baest left 11:03 bloonix left
flussence perl6: [1..10].pick(*)».map: { sleep $_; say $_ } 11:07
p6eval pugs b927740: OUTPUT«decodeUTF8': bad data: '\187'␤»
..rakudo 9719f7, niecza v13-381-g4158fa9: OUTPUT«(timeout)8␤»
flussence heh, trying to be too clever :) 11:08
moritz >> doesn't parallelize
(yet)
11:09 Trashlord left 11:10 Trashlord joined
flussence OTOH, maybe I've just invented threaded bogosort... 11:10
11:42 icwiener joined 11:58 mupon joined
mupon nom: (1..10).map: { sleep $_; say $_ } 11:59
p6eval nom 9719f7: ( no output )
mupon it works
12:00 mupon left
flussence
.oO( methinks the audience missed the point... )
 12:05
jnthn was trying to work out what the point was 12:09
Maybe it's something like: "hi! I don't understand that map is lazy!"
Not sure if the "it works" was a statement of fact or satire ;) 12:10
masak I agree that it works. 12:21
12:22 fasad joined
masak just not the way the person might've expected. 12:22
but clearly, it's the expectations that were wrong.
fasad hi 12:40
masak \o 12:41
fasad so 'man perl6' says 'perl -v' should work just the same as 'perl6 --version' but for me it enters the REPL. Regression? or it never worked? 12:43
moritz I think it worked back in the old (alpha) days 12:44
fasad time to fix it? 12:46
masak submits rakudobug
moritz I hope you have the time to fix it :-)
fasad * fasad lowers his eyes 12:47
damn
fasad lowers his eyes
12:48 timotimo left 12:49 timotimo joined
masak fasad: it's probably easier than you think. 12:50
fasad masak: ok, offhand, what skills do i need?
masak ability to read code. 12:51
knowledge of where the flags are defined.
luck.
moritz (grep can help you, if you don't know where stuff happens)
fasad and masak+moritz's blessings :p 12:52
masak command_eval in nqp/HLL/Compiler.pm seems to have something to do with it all. 12:53
but there ought to be some place where the flags are *set*.
moritz that's typically in some Compiler.pm
jnthn Either src/main.nqp or src/Perl6/Compiler.pm I suspect 12:54
moritz so src/Perl6/Compiler.nqp
or main.nqp
jnthn er, yes
moritz or nqp's NQP::Compiler
or HLL::Compiler
masak see? it's easy! :) 12:55
fasad LOL
moritz ... except that it's not so easy as it seems 12:56
because my initial tests seem to show that no single-letter versions of multi-letter options work
12:59 rindolf joined
rindolf Hi all. I'm getting a compilation error with the latest Rakudo Star using Parrot-4.0.0. Is it a known issue? 12:59
moritz rindolf: no, but that's not the parrot that rakudo recommends 13:00
rindolf moritz: OK.
moritz rindolf: is there any reason for deviating from the recommendation?
rindolf moritz: well, I've been packaging the new parrot/Rakudo/etc. for Mageia Linux Cauldron. 13:01
moritz rindolf: anyway, what's the error?
rindolf moritz: let me pastebot it.
fasad rindolf: welcome :p
rindolf moritz: here - paste.debian.net/154062/ 13:02
moritz never seen that one 13:04
13:06 pomysl left, pomysl joined, pomysl left, pomysl joined
dalek p: b23ee76 | moritz++ | src/HLL/Compiler.pm:
fix -v short form of --version
 13:10
moritz fasad: now tell me if that patch really looks complicated to you 13:11
fasad moritz: no it doesn't *look* complicated, but there is so much knowledge & experience behind it which fails to meet the eye... 13:14
moritz fasad: not at all
fasad: I just grepped for 'version' in HLL/Compiler.pm
fasad: the only other knowledge I had was that %adverbs holds the command line options
there aren't that many places mentioning 'version', and only one that does something conditionally on the presence of something version-y 13:15
13:20 kaleem left, mtk joined
fasad ok, so i'm inspired now. 'perl6 -h' also does not work. so may i request that no one fixes it, & i''l fix it within 24hrs ! (coz i have to understand some git basics, get a github account!, etc.) 13:20
*i'll 13:21
moritz fasad: +1
masak ++fasad 13:23
go go go! :)
fasad: learn.github.com/p/intro.html looks good for learning Git. 13:24
fasad masak: thanks. 'coz i need a crash course ! 13:25
13:27 rindolf left
moritz phenny: tell rindolf fwiw on my machine (debian stable), rakudo compiles fine on a recent parrot 13:29
phenny moritz: I'll pass that on when rindolf is around.
dalek kudo/nom: d63e72d | moritz++ | src/ (2 files):
more typed syntax errors
 13:36
kudo/nom: 0683298 | moritz++ | src/ (2 files):
more typed exceptions
p: cf6203f | moritz++ | tools/build/PARROT_REVISION:
bump parrot revision to get some testing
 13:37
[Coke] b: say 668-100 # rt-testneeded 13:49
p6eval b 1b7dd1: OUTPUT«568␤»
moritz oh my, 100 testneeded again?
jnthn Dammit, I should stop fixing stuff :P
[Coke] rakudo: "foo" ~~ /<B::TOP>/ 13:50
p6eval rakudo 9719f7: OUTPUT«===SORRY!===␤Could not locate compile-time value for symbol B␤»
jnthn Hm, not a *bad* error, but would be better off saying that B is not predeclared or some such.
Good we catch it at compile time, though :) 13:51
[Coke] jnthn: it's upgraded from "null PMC"
RT #64366 if you really want a better error msg. ;)
jnthn [Coke]: Then it's a huge improvement. That means compile time error, includes the name of the thing that's wrong, and not a Null PMC error
[Coke]: No, I'll take this as sufficient improvement to say the ticket is done
[Coke] so, 101 now. ;) 13:52
jnthn moritz++ will probably make a pass through this stuff when doing typed exception things anyway.
13:52 colomon left
[Coke] rakudo: class foo { has $!baz; method bar { my $!baz = 'lorem'; say $!baz; }; }; my $foo = foo.new; $foo.bar; 13:53
p6eval rakudo 9719f7: OUTPUT«===SORRY!===␤Cannot use ! twigil on my variable at line 1, near " = 'lorem'"␤»
moritz jnthn: yes, but I still need some ideas how to throw typed exceptions from the MOP
jnthn moritz: Well, that one's from World.
[Coke] rakudo: class foo { method bar { my $!baz = 'lorem'; say $!baz; }; };
jnthn moritz: But yes, you do. :) 13:54
p6eval rakudo 9719f7: OUTPUT«===SORRY!===␤Cannot use ! twigil on my variable at line 1, near " = 'lorem'"␤»
jnthn std: class foo { method bar { my $!baz = 'lorem'; say $!baz; }; };
p6eval std 48335fc: OUTPUT«ok 00:01 110m␤»
jnthn niecza: class foo { method bar { my $!baz = 'lorem'; say $!baz; }; };
[Coke] rakudo: class foo { has $!baz = 'ipsum'; method bar { my $!baz = 'lorem';say $!baz; }; }; my $foo = foo.new; $foo.bar;
fasad what is the relationship between nqp & rakudo ?
p6eval niecza v13-381-g4158fa9: OUTPUT«===SORRY!===␤␤Unable to resolve method add_method in class Any at /tmp/ZYTyuEsibA line 1:␤------> class foo { method bar { my $!baz ⏏= 'lorem'; say $!baz; }; };␤␤Unhandled exception: Check failed␤␤ at /home/p6eval/nie…
rakudo 9719f7: OUTPUT«===SORRY!===␤Cannot use ! twigil on my variable at line 1, near " = 'lorem'"␤»
moritz rakudo: try eval q[class foo { method bar { my $!baz = 'lorem'; say $!baz; }; };]; say $!.WHAT
p6eval rakudo 9719f7: OUTPUT«Exception()␤»
jnthn niecza?! :)
moritz fasad: much of rakudo's code is written in nqp 13:55
jnthn [Coke]: What ticket are you looking at for the my $!x decl, ooc?
[Coke] 69230
looks closable now.
(no longer silently fails in some cases.)
fasad moritz: nqp is a compiler or a language? 13:56
moritz fasad: nqp is a compiler for a subset of Perl 6 (and in that subset, much of rakudo is written)
jnthn [Coke]: Sounds good. I think Rakudo's response there is fine.
[Coke] std: std: 555 ~~!~~ 666 13:58
p6eval std 48335fc: OUTPUT«===SORRY!===␤Expecting a term, but found either infix ~~ or redundant prefix ~␤ (to suppress this message, please use space between ~ ~) at /tmp/sSDQhqr_KH line 1:␤------> std: 555 ~~!~~⏏ 666␤Parse failed␤FAILED 00:01 109m␤»…
[Coke] std: 555 ~~!~~ 666 13:59
p6eval std 48335fc: OUTPUT«===SORRY!===␤Expecting a term, but found either infix ~~ or redundant prefix ~␤ (to suppress this message, please use space between ~ ~) at /tmp/lT9_wQlZ86 line 1:␤------> 555 ~~!~~⏏ 666␤Parse failed␤FAILED 00:01 109m␤»…
[Coke] rakudo: 555 ~~!~~ 666
p6eval rakudo 9719f7: ( no output )
fasad moritz: oh, so nqp implemented some of Perl 6 features and using that primitive/incomplete set of features much of rakudo is written. correct?
[Coke] (RT#76436)
jnthn fasad: Yes
fasad: Also, nqp is bootstrapped, which means it is written in itself and can compile itself.
[Coke] rakudo: use Test; eval_lives_ok '1<&1' 14:00
p6eval rakudo 9719f7: OUTPUT«(timeout)»
fasad thinks of oroboros
[Coke] O_o
moritz fasad: most C compilers are also written in C 14:01
PerlJam fasad: this is perl, not python ;)
fasad PerlJam: sorry i'm not familiar with the python compiler 14:02
[Coke] nom: gist.github.com/1704540
p6eval nom 068329: OUTPUT«===SORRY!===␤Confused␤at /tmp/yOdOLj5os1:1␤»
moritz [Coke]: looks like a newline slipped in so that line 4 isn't a comment 14:03
[Coke] nom: gist.github.com/1704540
p6eval nom 068329: OUTPUT«Not enough positional parameters passed; got 0 but expected at least 1 in sub-signature of parameter @x␤ in sub infix:<merge> at /tmp/mYJTOtWcrz:1␤ in block <anon> at /tmp/mYJTOtWcrz:7␤ in <anon> at /tmp/mYJTOtWcrz:1␤»
moritz probably want $x? and $y? in the signature
fasad PerlJam: what's unique about python compiler? 14:04
*s/unique/different/
[Coke] rakudo: END { say "OH HAI" }; die "oh noes"
p6eval rakudo 068329: OUTPUT«oh noes␤ in block <anon> at /tmp/KMF2Dt0ZAK:1␤ in <anon> at /tmp/KMF2Dt0ZAK:1␤»
dalek ast: 31654a6 | moritz++ | S03-junctions/misc.t:
RT #109188, junctions with the List type object
ast: 2002a2f | moritz++ | S32-exceptions/misc.t:
more typed exceptions
PerlJam fasad: sorry, that was meant as humor. oroboros is a serpent eating its own tail. a python is a serpent. etc.
[Coke] moritz: that's for TimToady's script in RT#77474
(the gist) 14:05
fasad Perljam: oh!
masak PerlJam: oh!
fasad is a flickering tubelight
masak henceforth, explain all obscure jokes :)
[Coke] rakudo: END { say "OH HAI" }; die "oh noes"
p6eval rakudo 068329: OUTPUT«oh noes␤ in block <anon> at /tmp/iy3m4BMs6U:1␤ in <anon> at /tmp/iy3m4BMs6U:1␤»
[Coke] argh.
PerlJam I didn't realize it was that obscure :)
masak niecza: END { say "OH HAI" }; die "oh noes"
p6eval niecza v13-381-g4158fa9: OUTPUT«Unhandled exception: oh noes␤ at /home/p6eval/niecza/lib/CORE.setting line 1347 (die @ 3) ␤ at /tmp/2tAsrBgd0z line 1 (mainline @ 2) ␤ at /home/p6eval/niecza/lib/CORE.setting line 3609 (ANON @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 3610 (mo…
[Coke] rakudo: say ( { 1 + 1 }, { 2 + 2 } ).>>.() ; # RT#77670 14:06
p6eval rakudo 068329: OUTPUT«===SORRY!===␤Confused␤at /tmp/SoxLivAO_Z:1␤»
[Coke] rakudo: my %h; %h<a> = 3; %h.nonexist
p6eval rakudo 068329: OUTPUT«Method 'nonexist' not found for invocant of class 'Hash'␤ in block <anon> at /tmp/Eaa4YvK50_:1␤ in <anon> at /tmp/Eaa4YvK50_:1␤»
jnthn rakudo: say ( { 1 + 1 }, { 2 + 2 } )>>.() 14:07
p6eval rakudo 068329: OUTPUT«2 4␤»
jnthn std: say ( { 1 + 1 }, { 2 + 2 } )>>.()
p6eval std 48335fc: OUTPUT«ok 00:01 111m␤»
jnthn std: say ( { 1 + 1 }, { 2 + 2 } ).>>.()
p6eval std 48335fc: OUTPUT«ok 00:01 111m␤»
moritz nom: gist.github.com/1704555
p6eval nom 068329: OUTPUT«===SORRY!===␤Variable @y is not predeclared at line 2, near " };\nmulti "␤»
moritz nom: gist.github.com/1704555 14:08
p6eval nom 068329: OUTPUT«Ambiguous dispatch to multi 'infix:<merge>'. Ambiguous candidates had signatures:␤:(, @y)␤:(@x, )␤␤ in block <anon> at /tmp/73ZfGPG0we:9␤ in <anon> at /tmp/73ZfGPG0we:1␤»
fasad i have rakudo installed, so do i need to clone nqp too for my task & if yes, will it interfere with my rakudo installation ?
[Coke] jnthn: does your recent work make rt.perl.org/rt3/Ticket/Display.html?id=62244 closable?
moritz fasad: how did you install rakudo? via the git repo?
fasad moritz: yes. i have full rakudo source code. 14:09
moritz fasad: then you already have a copy of nqp in the nqp/ dir
fasad: cd nqp; git checkout master; git pull # now you're ready for hacking
nom: gist.github.com/1704555
p6eval nom 068329: OUTPUT«1␤» 14:10
jnthn [Coke]: It's obsolete. The way we do NCI has changed, the way we do namespaces has changed, etc.
[Coke]: Mention Zavolaj in the closing message, perhaps.
"We have working NCI now, see ..."
[Coke] is zavolaj shipped with star? 14:12
moritz star: use NativeCall; 14:13
p6eval star 2012.01: ( no output )
jnthn [Coke]: yes
moritz yes!
[Coke]: I've responded to the hamming ticket 14:14
jnthn star: use NativeCall; sub fork() returns Int is native() { * }; say fork(); 14:15
p6eval star 2012.01: OUTPUT«0␤29346␤»
fasad moritz: so i have made the necesarry changes to the code. obviously i need to recompile rakudo with 'make install', right ?
moritz fasad: well, first you recompile nqp
fasad: run 'make' 14:16
in the nqp dir
[Coke] moritz++ jnthn++
jnthn 'make install' in nqp dir
masak the new Hamming code... outputs '1'?
PerlJam jnthn: did you just show how to fork bomb the bot?
masak that can't be right :)
14:16 kaleem joined
moritz PerlJam: no, it was not a loop 14:16
jnthn PerlJam: No, that'd be like
[Coke] I suspect we have to harden safe with NativeCall in place. :P 14:17
PerlJam moritz: that exercise is trivial :)
[Coke]: indeed
jnthn stаr: use NativeCall; sub fork() returns Int is native() { * }; say fork() while 1;
[Coke] b: say 666-103; #rt
p6eval b 1b7dd1: OUTPUT«563␤»
jnthn ;-)
dalek ast: b09f2a9 | moritz++ | S05-metasyntax/angle-brackets.t:
fix and unfudge RT #64464 tests
 14:18
jnthn
.oO( oh fork() :-) )
 14:19
moritz nom: say 'alive'
p6eval nom 068329: OUTPUT«alive␤»
jnthn .u stаr
phenny jnthn: U+0073 U+0074 U+0430 U+0072
jnthn ;-)
[Coke] nom: our Str a
p6eval nom 068329: OUTPUT«===SORRY!===␤In "our" declaration, typename Str must be predeclared (or marked as declarative with :: prefix) at line 1, near " a"␤» 14:20
masak std: our Str a
p6eval std 48335fc: OUTPUT«===SORRY!===␤Multiple prefix constraints not yet supported at /tmp/z5kimkhvrl line 1:␤------> our Str ⏏a␤Malformed our at /tmp/z5kimkhvrl line 1:␤------> our Str ⏏a␤ expecting any of:␤ multi_declarator␤ typen…
[Coke] RT #76830
fasad moritz: nqp 'make' & 'make install' done. next? 14:22
moritz fasad: run ./nqp -h 14:23
fasad: see if that works
fasad: if it does, go back to the rakudo dir, and run perl Configure.pl && make
14:27 Yary_ joined
moritz nom: class A { my $!foo } 14:27
p6eval nom 068329: OUTPUT«===SORRY!===␤Cannot use ! twigil on my variable at line 1, near " }"␤»
14:31 Yary left 14:32 Yary_ is now known as Yary
masak S12 has example uses of 'my' and the ! twigil. 14:32
but you know that :)
moritz std: my $!foo;
p6eval std 48335fc: OUTPUT«ok 00:01 108m␤»
moritz fwiw I plan to test RT #69230 properly with a typed exception 14:39
14:39 PacoAir joined 14:42 Yary left
jnthn masak: Oh, that role private attribute thingy? 14:46
14:46 Yary joined
PerlJam jnthn: class attrs too 14:47
jnthn PerlJam: There's no such thing as class attrs.
my $.x; # a lexical and an accessor
dalek ast: 56d7207 | moritz++ | S12-enums/basic.t:
RT #77982, spooky interactions between enums
ast: 33e961e | moritz++ | S32-exceptions/misc.t:
RT #69230, my $!foo
masak right.
that's what I meant.
no, wait.
jnthn S12 may want some cleanup here
masak the spec mentions 'my %!something' too.
jnthn I mean, we *could* say that my $!x; is just a lexical.
masak looks
PerlJam S12:776
moritz but that would be very strange
jnthn But, why? That screws up your ability to read code and know what the scope of a $!foo thingy is.
moritz because the ! twigil means 'attribute'
jnthn Right. 14:49
moritz \o/ first RT closed that is tested by typed exceptions
jnthn nah, "my $!foo" just makes no sense to me.
masak I can sort of see the symmetry in it.
flussence
.oO( heh... calling fork() in a repl has some amusing side effects )
masak you can do 'has $looknotwigil' too.
moritz no reason to mess it up even further
fasad moritz: perl Configure.pl && make done. next 'make install' ? 14:50
jnthn masak: Yeah, I'm not entirely fond of that either, but can live with "sigilless = you figure it out, dear compiler"
er
twigilless
masak moritz: there will be people who think "class attribute" and expect "my $!twigilmeansnothing" to work. 14:51
PerlJam masak: only if we prime them to expect that
moritz masak: that's why we'll have a good error message for them 14:52
PerlJam moritz: so ... what's the answer to "class attributes"? :)
an ordinary lexical?
moritz I think so 14:53
PerlJam do they have to create the accessor?
14:53 Yary left
masak yes, an ordinary lexical. 14:53
'my @.classattr' creates the accessor.
it's because we allow the dot that allowing the bang makes sense.
14:53 am0c joined
masak it's just filling out the table with sensible defaults. 14:53
I oppose changing the status quo. 14:54
moritz well
let's look at what the twigil mean:
jnthn masak: The status quo is an error saying don't do that. :)
moritz . make me an accessor
! attribute
(no twigil) let the compiler figure it out 14:55
from that table, I don't see how 'my $.foo' implies we should allow 'my $!foo'
dalek ast: 15c94ed | moritz++ | S12-attributes/instance.t:
RT #108670, accessor name clash
 14:56
kudo/nom: 5fcfed5 | moritz++ | src/Perl6/Actions.pm:
fix thinko
kudo/nom: cb00ad1 | moritz++ | src/ (2 files):
typed exceptions for "my $!foo"
masak it's because you made the table up as you think it is, not how I see it :)
. make me an accessor
! no accessor
(no twigil) I don't know or don't care about attributes 14:57
moritz my $x; # no accessor either 14:58
class A { has $.x; method foo() { $!x # look, I can use that, even though it means "no accessor", and we have an accessor
fasad moritz: 'perl6 -h' now works for me. next?
moritz fasad: go back to nqp, and commit your changes
git add src/HLL/Compiler.pm 14:59
git commit -m 'fix -h option'
fasad: and tell me your github ID
fasad moritz: Siddhant 15:00
^my github ID
15:00 icwiener left
moritz fasad: you now have commit access 15:00
so once you've committed your stuff, do a git push 15:01
masak moritz: sure, it means "no accessor" if you did "has $foo". if you did "my $foo" it even means "no attribute" :) 15:02
fasad moritz: from the rakudo dir or nqp ?
moritz fasad: nqp
masak: well, if, you think '! means attribute', you don't have to do that distinction in the first place 15:03
masak: ie it gives you a more consistent picture which doesn't need further distinctinos
PerlJam And if you guys can't agree on how to think about it, imagine the newbie Perl 6 programmer's quandry!
moritz now apply ocaml's razor... erm, whatever it's called... :-)
PerlJam: well, she'll read the version of whoever last edited the UP6 chapter :-) 15:04
PerlJam heh
fasad moritz: i get "fatal: remote error: You can't push to git://github.com/perl6/nqp.git Use [email@hidden.address] 15:05
moritz fasad: then edit the .git/config files, substituting git://github.com/perl6/nqp.git by [email@hidden.address]
fasad: and then try again to git push 15:07
masak PerlJam: moritz argues out of orthogonality. I'm arguing out of DWIM.
PerlJam: there's not really a right or wrong here. just whether to be strict or permissive. 15:08
moritz well, not just orthogonality, but also readability
flussence `git remote set-url --push origin [email@hidden.address] should work too
moritz ie if you can safely make the connection ! => attribute
masak people who think they'll get class attributes can still do that. 15:09
I don't see the problem.
15:09 birdwindupbird left
moritz or git config remote.origin.pushurl [email@hidden.address] 15:09
masak people who know that class attributes don't exist will be surprised the variable is shared between instances, but if they're surprised at that, they haven't understood the class anyway.
and will need to read source anyway. 15:10
PerlJam in any case, the spec needs updating I think.
masak I agree it's not clear enough. 15:12
moritz: the 'safely make the connection ! => attribute' argument would be much stronger if it weren't for 'has $sans_twigil' 15:13
moritz masak: no, that only influences the opposite direction 15:14
but I won't shed a tear if we'd abolish that one too
15:14 phenny left
PerlJam me either 15:14
masak I'm fine with shedding both, yes. 15:15
but I'm not sure everyone is.
15:15 kaare_ joined 15:16 phenny joined
fasad moritz: on 'git push' i'm getting 'Permission denied (publickey).' 15:16
jnthn A case for has $sans_twigil; is that iirc the Perl 5 MOP work will have things that look that way
moritz fasad: have you told github about your public ssh key?
PerlJam jnthn: sounds weak 15:17
(this is Perl *6* after all ;)
masak fasad: help.github.com/linux-set-up-git/
jnthn PerlJam: Yeah, it's not a strong argument, just an existing one :)
masak another case for 'has $sans_twigil': some people hate twigils. 15:18
[Coke] brings up perl6.org in IE5.5
masak this will be a kind of safety valve for them.
dalek ast: 5514f34 | moritz++ | S04-statements/given.t:
RT #99912, regex in when-block failed to set $/
 15:22
moritz that kinda fails to impress me too
we don't have a safety valve for those who hate curlies
and it's much harder to program in Perl 6 without curlies than without private attributes :-) 15:23
15:25 baest joined
masak if everyone else is fine with removing all('has $sans_twigil', 'my $!fake_twigil'), I am too. 15:27
but not just one of them.
15:27 pothos left
moritz +2 then :-) 15:27
erm, -2
you know what I mean :-)
15:27 pothos joined
masak :P 15:28
dalek ast: c0c67ef | moritz++ | S02-literals/misc-interpolation.t:
RT #104594, lexicals in blocks in double-quoted strings
 15:30
p: e6c8684 | (Siddhant Saraf)++ | src/HLL/Compiler.pm:
fix -h short form of --help
moritz fasad++ 15:31
down to 97 testneeded tickets
masak fasad++
fasad * pops open a champagne and pours out to everyone in #perl6 15:32
Cheers !
felher If i want to specify: 'everything that is matched by <all-words> but not by <bad-words>', is "regex good-words { <!bad-words> <all-words>" in a
fasad pops open a champagne and pours out to everyone in #perl6
felher * the way to go?
masak er.
could you repeat that so that it makes sense? :) 15:33
felher If i want to specify: 'everything that is matched by <all-words> but not by <bad-words>', is "regex good-words { <!bad-words> <all-words>
Damn it.
felher brb
masak felher: it sounds like you need a negative lookahead or something.
or maybe '$string ~~ /<@good-words>/ && $string !~~ /<@bad-words>/' 15:34
moritz <!bad-words> is a negative look-ahead
jnthn fasad++ # contributing to Rakudo :)
felher is back with a new x-server. 15:35
moritz,masak: thnx. so if i need a negative-lookahead and <!bad-words> is one, i'll use that :)
masak <!before <bad-words>> is another. 15:36
I'm not really sure what the difference is.
jnthn The latter burns more CPU cycles :P
fasad jnthn: :) this was my first ever code contribution to any open source project. Which was your first?
moritz felher: but what you want often has many nasty edge cases, dependent on what exactly you want
jnthn fasad: Fixing some Win32 build bug in Parrot, I think. :)
moritz for example you should anchor stuff with word boundary 15:37
felher moritz: yep, thnx :) 15:38
fasad moritz, masak: what was your first code contribution ever to an open source project ?
*were
masak hm, good question. 15:40
might've been to Pugs, actually. but I'm not 100% sure.
moritz fasad: I'm not sure I remember. Either some p5 module, or pugs docs
oh wait, code. then probably test cases for complex numbers for pugs 15:41
fasad will always remember
masak moritz: I thought your first contribution was some web page fix. 15:43
to the Pugs project, I mean. 15:44
wasn't that why you got a commitbit?
moritz masak: I just mentioned that a link was broken, but I didn't fix it
masak oh!
flussence come to think of it, perl6 is the first thing I remember contributing code for too...
fasad :) 15:45
moritz around that time I also fixed some code in igal (a html gallery generator written in perl), and some stuff in a syntax hilighter
can't say for sure which one was first
15:47 wtw left
fasad is reading Bootstrapping (compiler) on Wikipedia & trying to wrap his head around the concept :) 15:48
masak fasad: it's not so tricky. you write the compiler in such a way that it can compile its own source code. 15:49
first you compile it with something else. then you compile it with itself.
that's all, really.
fasad yeah, but the first time you read it ... :) 15:51
but then you realize that that you were taught English in English, & things start making sense 15:53
moritz that's a nice way to look at it. 15:54
masak yeah. it's a good thing brains are better at filling in the blanks than programs :)
15:59 skids joined 16:01 lrnperl6 joined
fasad so en.wikipedia.org/wiki/Bootstrapping..._compilers does not list perl 6. Should it? 16:02
masak Perl 6 is not a compiler. 16:03
it's a language specification.
jnthn decommute &
masak I guess both Rakudo and Niecza can be said to be partly self-hosting, but not fully. 16:04
well, maybe Niecza is at this point.
moritz nqp is self-hosting 16:05
fasad actually the heading is a bt misleading, it should be 'List of >programming languages having< self-hosting compilers' 16:06
^^based on the content of the section
PerlJam "Languages that have self-hosting compilers" sure sounds like Perl 6 should be on that list 16:07
16:07 am0c left
PerlJam fasad: but, yeah, the title is all off 16:07
fasad is now fixing the wikipedia article 16:08
16:10 GlitchMr joined 16:11 MayDaniel joined
fasad is perl 5 also self-hosting? #cause i've heard 'only perl can parse perl' 16:12
scratch that. perl's compiler is written in C, IIRC
16:13 spine joined
arnsholt The Perl 5 compiler is written in C, yes 16:13
16:13 nsh left
arnsholt "Only perl can parse Perl" means that only perl (the program) can parse Perl (the language) 16:14
(Which is accomplished by deep magical sorcery, if you wondered =)
PerlJam though the magic has been bubbling closer to the surface lately. 16:15
skids wonders as to the meaning of "as" trait on a parameter as used in Rakudo src/core. Can't find "as" in spec except as a sub trait for inner return type specifier. 16:16
16:16 MayDaniel left
masak Perl 6 is all about building the awesomeness from the ground up and then making it perform. Perl 5 is all about condensing the awesomeness from Unix mythology and then trying to make it extensible and less magical while not breaking backwards compatibility. :) 16:17
skids: "as" was de-spec'd and replaced by constructs such as Int(Any) 16:18
flussence "as" is gone? when'd that happen?
(is it the as I'm thinking of?) 16:19
masak what "as" are you thinking of? :)
flussence the one in function signatures... 16:20
skids masak: just the answer I was hoping for, since I was missing the Int(Num) feature... so I can use that until things congeal. masak++
flussence ah, that one's still there.
I think :/ 16:21
skids "sub A () as Int" is still there. 16:22
flussence yeah, I can find that one... can't find "sub A ($a as Int)"
masak hm, yes. maybe "as" survives for return casting. 16:23
16:24 lrnperl6 left 16:25 kaleem left
skids That would apparently now be "sub A (Mu(Int) $a)" or somesuch. Though I'm not quite sure what Mu(Int) would acheive where Int(Mu) makes sense to me. 16:26
flussence std: sub A(Int(Str) $b) { ... } 16:27
p6eval std 48335fc: OUTPUT«Potential difficulties:␤ $b is declared but not used at /tmp/g21e5TUidb line 1:␤------> sub A(Int(Str) ⏏$b) { ... }␤ok 00:01 111m␤»
flussence perl6: sub A(Int(Str) $b) { say $b.WHAT }; A(1); A("1"); 16:28
p6eval niecza v13-381-g4158fa9: OUTPUT«Int()␤Unhandled exception: Nominal type check failed in binding $b in A; got Str, needed Int␤ at /tmp/9me2s0mHyg line 0 (A @ 1) ␤ at /tmp/9me2s0mHyg line 1 (mainline @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 3609 (ANON @ 3) ␤ at /home/p6eval…
..pugs b927740: OUTPUT«*** ␤ Unexpected "Int"␤ expecting word character, "|", ":", "*", parameter name or ")"␤ at /tmp/5yh5OIv89L line 1, column 7␤»
..rakudo cb00ad: OUTPUT«===SORRY!===␤Malformed block at line 1, near "(Int(Str) "␤»
flussence niecza's doing *something*, but I'm not sure what. 16:29
16:30 DarthGandalf left 16:34 DarthGandalf joined 16:47 fasad_ joined, fasad left
TimToady perl6: my @elems = <1 2 3 1>; say @elems == set @elems; 16:48
p6eval rakudo cb00ad: OUTPUT«===SORRY!===␤CHECK FAILED:␤Undefined routine '&set' called (line 1)␤»
..pugs b927740: OUTPUT«*** No such subroutine: "&set"␤ at /tmp/smunr20Bbi line 1, column 28-48␤»
..niecza v13-381-g4158fa9: OUTPUT«Bool::False␤»
TimToady niecza++ 16:49
niecza: my @elems = <1 2 3 1>; say @elems == set @elems;
p6eval niecza v13-381-g4158fa9: OUTPUT«Bool::False␤»
TimToady niecza: my @elems = <1 2 3 4>; say @elems == set @elems;
p6eval niecza v13-381-g4158fa9: OUTPUT«Bool::True␤»
TimToady just checking :)
eviltwin_b was just wondering about that :) 16:50
TimToady anyway, that's likely to be faster than almost anything else 16:51
16:52 mj41 left
ashleydev TimToady: why faster? 16:56
TimToady only constructs one object, which is really a hash 16:58
and only uses it for its size
at least, it's likely to be faster when set operations optimize away all the => True bits 16:59
sorear good * #perl6
TimToady howdy 17:00
masak sorear! \o/
TimToady I'm getting really tired of seeing things like all(one(Bool::False, Bool::False, Bool::True)) 17:01
I'm not sure .gist should add the Bool
after all, you'd think the "gist" would be shorter
jnthn nom: say True 17:02
p6eval nom cb00ad: OUTPUT«Bool::True␤»
TimToady instead it *adds* meta info
jnthn Hm, should be trivial to fix.
TimToady nom: print True,"
p6eval nom cb00ad: OUTPUT«===SORRY!===␤Confused␤at /tmp/aJ1ejYtB4P:1␤»
moritz nom: say ~True
TimToady nom: print True,"\n";
p6eval nom cb00ad: OUTPUT«True␤»
TimToady nom: print True.perl
p6eval nom cb00ad: OUTPUT«Bool::True»
TimToady I think maybe only that one should add the type
it's a lot easier to understand all(one(False, False, True)) 17:03
sorear TimToady: pastie.org/3283332 ?
jnthn TimToady: Will make it happen. 17:04
TimToady yes, that's what I am proposing
sorear shall I commit?
or are you about to
TimToady お願いします！ 17:05
masak decommutes
sorear dalek? 17:07
[Coke] nom: say (22-1)/4 ; # averge numbers of submissions per not-me. 17:08
TimToady had a long weekend, apparently
p6eval nom cb00ad: OUTPUT«5.25␤»
sorear I guess maybe it's blocking on email
[Coke] O_o
jnthn > say all(one(True, False), True) 17:09
all(one(True, False), True)
:)
17:09 wolfman2000 left
TimToady \o/ 17:09
sometimes we are very slow to see the obvious...
[Coke] nom: next say all for one time 17:10
p6eval nom cb00ad: OUTPUT«all()␤»
TimToady nom: next 17:11
p6eval nom cb00ad: OUTPUT«next without loop construct␤ in block <anon> at /tmp/5NnTtLEWFM:1␤ in <anon> at /tmp/5NnTtLEWFM:1␤»
17:11 fasad_ is now known as fasad
TimToady how does that not get the error in yours? 17:11
oh, there's a for, duh
for one time { say Cool } 17:12
perl6: for one time { say Cool }
p6eval pugs b927740: OUTPUT«*** No such subroutine: "&Cool"␤ at /tmp/2K8M7nvC6J line 1, column 20-25␤»
..rakudo cb00ad, niecza v13-381-g4158fa9: OUTPUT«Cool()␤»
TimToady we now have an official one-time loop :) 17:13
[Coke]++ 17:14
jnthn gently pokes at dalek
dalek ecs: f4488b0 | sorear++ | S12-objects.pod:
Change .gist on enums to not include the type

Because .gist should be shorter, and the types are rarely needed. .perl stays as is, for disambiguation.
17:15 colomon joined
felher jnthn: Oh, now that i see you. Does nom do backtracking into subrules yet? :) 17:16
jnthn felher: Yeah, I fixed that just before the last compiler release 17:17
(So that compiler release has it, and the Star release from this weekend has it too)
[Coke] nom: for all time { say Cool} 17:18
p6eval nom cb00ad: OUTPUT«Cool()␤»
dalek ast: 8d71270 | jnthn++ | S02-types/bool.t:
Update Bool.gist tests.
felher jnthn++: Awesome! Thanks :)
dalek kudo/nom: 683a387 | jnthn++ | src/core/ (2 files):
Enumerations should .gist to just the key, not including the full package name.
 17:19
17:22 daxim left 17:27 timotimo left
colomon sorear: ping? 17:28
[Coke] ooh, nclark doing ticket triage on perl6.
sorear colomon: pong 17:29
dalek ecza: fb2bb1e | sorear++ | lib/CORE.setting:
Enumerations should .gist to just the key, not including the full package name.
sorear Nicholas in #p5p -- funny he's not here
release day! 17:31
colomon \o/ 17:33
sorear will, ahem, cheat and run a spectest on the p6eval box 17:35
colomon: pong?
colomon sorear: I've been messing about with KeySet while waiting for the doctor this morning. I can see Hash is doing some sort of proxy thing, ideally I need to figure out how to overload it so that I can remove False things from the KeySet in an efficient fashion.
17:36 kaleem joined
dalek kudo/nom: ef22e37 | jnthn++ | src/core/Str.pm:
Forbid :8(non-Str-here), as per spec.
 17:38
jnthn nudges testneeded back up by 1 again ;) 17:39
Also found two tickets to close that were already fixed and tested. :)
Guess I should get on with my slides for $dayjob-event now...
sorear or maybe I'll use feather instead, seems to be faster 17:40
17:46 Chillance joined
[Coke] Juerd++ #feather 17:49
jnthn++ moritz++ queue attrition 17:50
Juerd Also, feather as a secure sudo which is rare these days :) 17:52
[Coke] nom: enum Boo <True False Meh>; say any(Bool::True,Boo::True)
p6eval nom 683a38: OUTPUT«any(True, True)␤»
[Coke] nom: enum Boo <True False Meh>; say any(Bool::True,Boo::True).perl
p6eval nom 683a38: OUTPUT«any(Bool::True, Boo::True)␤»
17:52 GlitchMr left 17:58 dakkar left
[Coke] that's ok für gist? 17:58
sorear I think so 17:59
17:59 alester joined
[Coke] hokay. 18:00
18:03 ksi joined 18:10 icwiener joined
sorear sweet, 'All tests successful' 18:12
*however* the bootstrap failed!
colomon all tests successful here as well.
[Coke] sorear: OOC, did you run with TEST_JOBS=n ? 18:14
18:15 supernovus joined
dalek kudo/nom: ce5ccae | moritz++ | tools/build/NQP_REVISION:
bump NQP revision to get option handling and charclass fixes
 18:22
sorear [Coke]: no 18:23
fff. bootstrap is not cooperating at all 18:24
looks like I won't get to the release until tonight, if at all :|
18:24 kmwallio joined
moritz fwiw bumping the parrot revision seem to have helped with the setting build times 18:25
from ~16min down to about 4min for the .pm -> .pir step
which is closer to what I remember from a while go
tadzik ooh 18:27
what did Parrot do?
18:28 pmurias joined
sorear apparently I majorly broke something somewhere 18:28
pmurias!
pmurias sorear: hi
phenny pmurias: 28 Jan 00:07Z <colomon> tell pmurias gist.github.com/1691700 # p5test results
pmurias: 02:01Z <sorear> tell pmurias If you see this message before Monday, I'd like it if you could give a better description of your changes for announce.v14 18:29
pmurias pmurias: re release annoucement, the description of changes is perfect
benabik pmurias: Talking to yourself again? ;-) 18:30
pmurias again?
sorear: re release annoucement, the description of changes is perfect
supernovus Guh, tracking down grammar related bugs can be a pain. :p
tadzik do you know Grammar::Tracer yet? :) 18:31
pmurias sorear: i'm currently studing for my (numerical analysis)-- exam, but i hope to get much more work done on the p5 interop after the exam 18:32
supernovus Not yet. At this point, I'm willing to try anything. I just want to make Exemel work properly under nom. I don't remember having this many problems porting from alpha to ng, but given how long ago it was, I probably did ;-)
tadzik :)
pmurias sorear: did you read the use :from<perl5> messages? 18:34
supernovus It can now parse <thing name="blah"/> just fine, and stores the attributes "name" => "blah". But if the element has content, such as <a href="woot">this</a> it parses it, but doesn't find the "href" => "woot" part.
18:35 fasad left
[Coke] supernovus: you have any WIP on a branch? 18:36
moritz tadzik: re what did parrot do, either they unbroke some earlier breakage, or maybe it was just my installation that was somehow busted
supernovus [Coke]: github.com/supernovus/exemel the master branch has my latest tests, and includes all the stuff I've done so far to get it to work with nom. 18:38
[Coke] how do you test it? just "prove t/"?
sorear pmurias: yes 18:39
pmurias: I don't have a good answer
supernovus [Coke]: the t/03-query.t is the file I'm currently playing with.
[Coke]: and I don't expect it to pass at the moment even if the grammar bug is tracked down, as I added some extra stuff to t/query.xml to test the bug more. I may pull out the extra stuff so that the file is in a state where it should pass if all the bugs were fixed. 18:40
18:41 tjrandall joined, tjrandall left
pmurias sorear: i'll propably try reimporting stuff from the module twice, as serializing arbitary perl5 modules seems too crazy 18:41
sorear pmurias: I agree 18:42
sorear tries reverting the latest LTM changes
supernovus there query.t is now in its proper state again
err, query.xml
tadzik btw, any comments on github.com/rakudo/rakudo/compare/n...prints-WHY ? 18:45
sorear There is no way I am going to finish the release in the next ten minutes - tonight it is 18:46
[Coke] nom: my %a = { $.idattr => 3 }; say %a.perl 18:47
p6eval nom ef22e3: OUTPUT«Null PMC access in find_method('idattr')␤ in block <anon> at /tmp/nLqcRL3iFS:1␤ in <anon> at /tmp/nLqcRL3iFS:1␤»
[Coke] masak, is there a ticket for that? 18:48
18:49 estrabd joined, estrabd left 18:50 estrabd joined
masak [Coke]: no. 18:51
[Coke] enjoy. ;) 18:52
supernovus: you are returning False and then trying to use it as a list, I tihnk.
supernovus [Coke]: I know that part. It should never get there. The test failing isn't the issue, the fact that the attributes are entirely missing is. :-) 18:53
[Coke] do you have a smaller snippet than 03*.t that demonstrates the failure?
supernovus Sure, I'll write a quick snippet that will show exactly what's wrong. 18:54
[Coke] If nothing else, we got another RT out of it. :P
dalek ecza: 74ba1ec | sorear++ | / (2 files):
Revert "Ignore arguments to rules for LTM purposes"

This reverts commit 6915253d906f76e33c0a292a3cb5129310da238e. Seems to break the bootstrap and I don't have time to study why just yet.
ecza: 852f0ff | sorear++ | lib/Cursor.cs:
Fix use of variables in propagation
sorear bootstrap succeeded; will finish release in a few hours 18:55
supernovus [Coke]: Okay, take a look at the newly added what-is-wrong.p6 in the main folder. 18:59
[Coke] supernovus: gist.github.com/1705999 19:02
the only difference I see between the 2 lines is the <?xml> prefix. 19:03
supernovus hmm, that's not the output I'm getting at all...
[Coke] but the comment makes me think I should be seeing something else.
$ perl6 --version 19:04
This is perl6 version 2012.01-45-g9719f7d built on parrot 3.11.0 revision RELEASE_3_11_0
supernovus This is perl6 version 2012.01-43-g396f827 built on parrot 3.11.0 revision RELEASE_3_11_0
I will rebuild and see what happens.
[Coke] looks like mine is slightly newer.
(mine's probably a day or 2 old at this point.) 19:05
supernovus Rebuilding from git to see if that helps. I last rebuilt 2 days ago. 19:07
masak submits [Coke]'s rakudobug 19:08
nom: my %a = { $.idattr => 3 }; say "alive" 19:11
p6eval nom ce5cca: OUTPUT«Null PMC access in find_method('idattr')␤ in block <anon> at /tmp/TfNWmA8_Lz:1␤ in <anon> at /tmp/TfNWmA8_Lz:1␤»
masak nom: class A { my %a = { $.idattr => 3 } }; say "alive"
p6eval nom ce5cca: OUTPUT«Null PMC access in find_method('idattr')␤ in <anon> at /tmp/IjSkXH_snl:1␤ in block <anon> at /tmp/IjSkXH_snl:1␤ in <anon> at /tmp/IjSkXH_snl:1␤»
masak nom: my %a = { self.idattr => 3 }; say "alive"
p6eval nom ce5cca: OUTPUT«===SORRY!===␤Symbol 'self' not predeclared in <anonymous> (/tmp/g6qCPc4vDg:1)␤»
masak nom: class A { my %a = { self.idattr => 3 } }; say "alive"
p6eval nom ce5cca: OUTPUT«===SORRY!===␤Symbol 'self' not predeclared in <anonymous> (/tmp/XwlqA54gqp:1)␤»
masak nom: class A { method foo { my %a = { self.idattr => 3 } } }; say "alive"
p6eval nom ce5cca: OUTPUT«alive␤»
masak nom: say "before"; my %a = { self.idattr => 3 }; say "alive" 19:12
p6eval nom ce5cca: OUTPUT«===SORRY!===␤Symbol 'self' not predeclared in <anonymous> (/tmp/0Sw1hpQTdv:1)␤»
masak nom: say "before"; my %a = { $.idattr => 3 }; say "alive"
p6eval nom ce5cca: OUTPUT«before␤Null PMC access in find_method('idattr')␤ in block <anon> at /tmp/i4okZIdPp8:1␤ in <anon> at /tmp/i4okZIdPp8:1␤»
masak nom: class A { has $.idattr; my %a = { $.idattr => 3 } }; say "alive" 19:14
p6eval nom ce5cca: OUTPUT«Null PMC access in find_method('idattr')␤ in <anon> at /tmp/F_HeJ0RhYK:1␤ in block <anon> at /tmp/F_HeJ0RhYK:1␤ in <anon> at /tmp/F_HeJ0RhYK:1␤»
TimToady 'my $!classattr' makes much more sense when you think about 'trusts' and private accessors
masak ooh 19:15
moritz so 'my' doesn't (just) declare a lexical in that case?
jnthn masak: We must have a ticket for that $.a in a place it ain't meant to be issue. 19:16
TimToady anything with ! also gets a private accessor (which is not called when you just say $!foo)
the private accessor is only for trusts-style calls
jnthn Oh, that idea lives on? 19:17
jnthn thought it'd been gotten rid of...
That's mean that "has $!a; has @!a" is problematic as well as the . variants that we recently disallowed. 19:18
dalek ast: 6cfc6bd | moritz++ | S12-class/stubs.t:
test for stubbed-but-not-filled-in class
TimToady how are you implementing $obj!THAT::CLASS::a then? 19:19
jnthn TimToady: I'm not at the moment; if you want such an accessor you have to explicitly declare it. 19:20
TimToady: I'm still a little bothered about generating them all when they won't be used the majority of the time. I guess some lazy technique would do.
TimToady they don't need to be autogenerated unless you trust someone
jnthn True.
TimToady: Would "has $!x" only generate an ro private accessor? 19:21
TimToady: And "is rw" needed for a rw one?
e.g. has $!x is rw;
TimToady I suspect all private accesses are born lvalues 19:22
jnthn Another option would be not to generate accessors methods, but instead to spot the lack of one at compilation time and just compile it down to an attribute access.
TimToady otherwise we don't know the intended state of $!x when 'has $.x is rw'
and it seems to me if you trust yourself to modify $!x, you trust others too
jnthn That is, $x!Foo::Bar::a would first look for a private method 'a', and failing one of those being in the private methods table, it'd consider whether there's an attribute. 19:23
TimToady yes, private access is just a sub call really, and can be inlined
jnthn Yeah, we already optimize those somewhat in Rakudo
We also detect missing privates at CHECK time
I think 19:24
TimToady omits pun on Unix
jnthn :P
TimToady or were you speakiing of deserters?
jnthn TimToady: Well, what I'm really getting at is, would we expect Foo.^methods(:private) to come back with a bunch of generated private accessors? 19:25
If yes, that probably means they have to be generated up front. If not, we can play tricks more easily. :)
Though of course meta-objects can lie as much as they want too :)
TimToady ^methods could generate them lazily for all that :) 19:26
jnthn suspected that answer was coming... :)
moritz (test plan)++ # caught my accidental use of say() instead of is()
dalek ast: ecd8da2 | moritz++ | S02-types/hash.t:
RT #77922, Hash.new
 19:27
supernovus Okay, now I am really confused. The output from [Coke]'s rakudo is entirely different than mine. I just used a fresh compile. gist.github.com/1706138
19:28 mdxi is now known as sboyette
[Coke] supernovus: your updated sample doc works for me still. 19:28
supernovus: did you install rakudo?
supernovus [Coke]: Yeah, it's installed. 19:29
[Coke] I'm on feather, if that helps.
supernovus $ perl6 --version This is perl6 version 2012.01-52-gce5ccae built on parrot 4.0.0 revision RELEASE_4_0_0-85-g8a1265a
jnthn Where's the code for what-is-wrong.p6?
jnthn can try it
supernovus jnthn: github.com/supernovus/exemel
I am going to try it on another machine and see what happens. 19:31
TimToady and if you guys think 'has $x' is so terrible, I can only imagine what you'll think of: 19:32
std: class X { has \x = 42; method sayit { x.say } }
p6eval std 48335fc: OUTPUT«ok 00:01 109m␤»
TimToady :P
dalek ast: bddc607 | moritz++ | S12-attributes/instance.t:
RT #74274, sneaking in access to private attributes through the back door
 19:33
moritz masak: ^^ one of your favorite bugs being close right now.
[Coke] TimToady: yah, I don't particularly like it. ;)
19:33 birdwindupbird joined
TimToady std: class X { has \x is public = 42; method sayit { x.say } } 19:33
p6eval std 48335fc: OUTPUT«===SORRY!===␤Term definition requires an initializer at /tmp/VuUWfstGP8 line 1:␤------> class X { has \x ⏏is public = 42; method sayit { x.say } }␤Confused at /tmp/VuUWfstGP8 line 1:␤------> class X { has \x ⏏is pub…
TimToady innersting 19:34
19:34 Trashlord left
jnthn supernovus: Here is output I get: gist.github.com/1706190 19:34
supernovus jnthn: So yours is like mine, in that the attributes magically vanish. 19:36
jnthn Yeah.
jnthn looks around for a pmichaud :) 19:37
supernovus Okay, I'm using Debian sid on the current testing box (which is actually a 32 bit VM running on virtualbox). How about jnthn and [Coke] ?
jnthn supernovus: It's certainly a bug. 19:38
supernovus: If you change
rule element {
to
regex element {
Then it captures properly.
jnthn whines about the bstack cut logic
dalek ast: ba5ee93 | moritz++ | S06-traits/is-copy.t:
RT #82810, ||= on an "is copy" array parameter
 19:39
supernovus Interesting. I'll try that and see what happens. Still doesn't explain why the bug doesn't show up on [Coke]'s computer. 19:40
jnthn supernovus: No, I'm confused about that too
TimToady if the 'b' stands for backtracking, I'd point out that keeping backtracking info outside of cursors is fraught with peril; I much prefer the view that backtracking is simply throwing away unwanted cursors 19:41
jnthn I'm running a 64-bit build on Windows
TimToady: $!bstack is an attribute of Cursor, which contains marks. It works in conjunction with $!cstack, which is a stack of cursors, and those are thrown away during backtracking involving captures or subrules. 19:42
supernovus 64-bit Windows and 32-bit Linux are both showing the same bug. I'm going to try running this on 64-bit Linux and see what that does.
[Coke] if you're on feather, it's /home/coke/sandbox/perl6pass/rakudo/install/bin/perl6
(which is also a 32bit linux.) 19:43
jnthn TimToady: Or so far as I understand the design, anyway. ;-)
[Coke] dunno what's up, either.
19:44 GlitchMr joined
supernovus I don't know if I still have a feather account, it's been three lifetimes and several iterations of rakudo since I last logged in there :-) 19:44
19:44 twirp joined
TimToady ah, in that case, I'd say the problem is that the stack of cursors is getting out of sync with your control flow somehow; in STD-based grammars, this is managed by using lexically scoped lazy lists, so when you exit a scope it automatically trims the lazy lists that need to go away, and leaves the rest 19:44
jnthn TimToady: It's something along those lines. It basically throws away cursors that it should not. 19:45
TimToady well, I'm just guessing ignorantly, so take with eighteen grains of salt
jnthn :)
Well, with a little luck pmichaud++ may have a tuit or two to look at in sometime soon :) 19:46
*it
19:46 Vlavv_ joined, GlitchMr left
TimToady basically, the "cut" in a STD engine is the difference between the lazy lists you throw away, and the lazy list you return 19:47
supernovus Well, none of novus@feather, supernovus@feather or huri@feather work, so I'm assuming I don't have an account on there anymore.
TimToady STD is basically emulating continuations with lazy lists of closures 19:48
masak moritz: nice!
19:49 Vlavv` left 19:51 kaleem left 19:52 pmurias_ joined
jnthn TimToady: Ah, that makes sense. 19:52
[Coke] checks.
jnthn OK, I really should stop peeking at that and do the $dayjob slides... :)
[Coke] supernovus: yah, don't see those.
19:53 pmurias left
supernovus okay, changing "rule element" to "regex element" now has all the included tests passing. I can't guarantee it's perfect, as I'm sure the test coverage should be a heck of a lot more, but basic XML parsing, manipulation, querying and output is now ready for the newest rakudo :-) 19:58
masak supernovus++ 19:59
supernovus There, I updated the "State of modules on nom" list to show Exemel as working again. Next up DateTime::Utils. 20:06
jnthn supernovus++
moritz supernovus++ indeed
[Coke] supernovus++ 20:09
dalek ast: 4808a9d | moritz++ | S05-metasyntax/charset.t:
RT #89470, char classes with exclusions
 20:13
moritz jnthn: when you have some spare tuits, it would be very helpful for me if you could implement named arguments in nqp regexes (<.typed_panic: 'X::Malformed', what => "Block">). My attempts to wire up <LANG('MAIN', 'arglist')> for that task have failed :/ 20:16
tadzik huh, nom compilation fails for me
gist.github.com/1706413
moritz tadzik: uhm 20:17
20:17 birdwindupbird left
moritz tadzik: SymbolTable has long been renamed to World 20:17
tadzik: something is very outdated in your nom (maybe the Makefile?) 20:18
jnthn moritz: In NQP, right?
tadzik tries git clean -fdx
moritz jnthn: yes
jnthn: I want to use it Grammar.pm
jnthn nom: grammar G { rule x(:$y) { { say $y } '' }; rule TOP { <x(y => 42)> } }; G.parse('') 20:19
p6eval nom ce5cca: OUTPUT«Too many positional parameters passed; got 2 but expected 1␤ in regex x at /tmp/rY6d3MZgc7:1␤ in regex TOP at /tmp/rY6d3MZgc7:1␤ in method parse at src/gen/CORE.setting:7642␤ in block <anon> at /tmp/rY6d3MZgc7:1␤ in <anon> at /tmp/rY6d3MZgc7:1␤»
jnthn Hm, needs doing there too
20:19 birdwindupbird joined
moritz I thought it worked in nom.. 20:20
20:20 icwiener left
moritz nom: grammar G { method x(:$y) { say $y }; rule TOP { <x: y => 42> } }. G.parse('') 20:20
p6eval nom ce5cca: OUTPUT«===SORRY!===␤Confused␤at /tmp/FdxsaWZFZf:1␤»
tadzik argh, I'm on rakudo/master :/ 20:23
moritz didn't we mean to delete that branch? 20:24
tadzik I don't think it matters when you do 'g co master'
and then 'git pull' appears to be up-to-date
moritz deletes the branch
tadzik: then delete it locally too
git branch -D master
tadzik I will 20:25
flussence 'git remote prune origin' is a good idea to run once in a while too :)
jnthn .u ௐ 20:26
phenny U+0BD0 TAMIL OM (ௐ)
masak I love seeing RT tickets get closed. 20:27
moritz there's just a wee problem; I tend to write tests for those tickets that seem sensible and not too hard to test for 20:28
jnthn .u ഇ 20:29
phenny U+0D07 MALAYALAM LETTER I (ഇ)
20:29 jaldhar left
moritz so over time the 'testneeded' tag accumulates a bunch of weird tickets 20:29
which makes it not much fun to attack
skids tries to wrap head around S14:516 20:33
All the example have role Foo[Bar] but then we have Array[Bar] whichis a class, not a role... 20:34
20:36 nwc10 joined
TimToady we can have punned roles as classes; in this case it would presumably override the default pun of Array[Any] 20:37
though we still need to detangle the 'of' type from the default value 20:39
skids Ah, so Array is supposed to be a punned class?
erm, punned role as a class.
jnthn Well, Foo[Bar] in Rakudo desugars to Foo.^parameterize(Bar) and the meta-class can do what it wants.
TimToady the Array role would presumably be a slightly more concrete version of Positional 20:40
where Array assumes mutability
jnthn (So people can implement custom parametric types.)
Note that this call may well happen at compile time, like many other MOP calls.
TimToady sure, mostly just thinking about user-visible syntax, as is my wont 20:41
skids so my Array[Recipe] %book; is really my Mu does Array[Recipe] %book;?
TimToady no
moritz std: $@*ARGS
p6eval std 48335fc: OUTPUT«===SORRY!===␤Unsupported use of $@ variable as eval error; in Perl 6 please use $! at /tmp/t94nayz7EE line 1:␤------> $@⏏*ARGS␤Parse failed␤FAILED 00:01 107m␤»
jnthn More like my %book := Hash[Array[Recipe]].new 20:42
moritz jnthn: which method in World do i use to check if a symbol is in scope? find_symbol? 20:43
jnthn moritz: What do you mean by "in scope"? 20:44
moritz jnthn: declared, in visible in the lexical scope we're currently in
s/in/and/
jnthn moritz: find_symbol works provided the thingy has a known compile time value. It throws if not. But some things aren't known at compile time at all 20:45
moritz: already_declared may do it
moritz: is_lexical is perhaps more what you're after 20:46
Though it doesn't do multi-part names
moritz that's OK
I just want to search for 'self' 20:47
jnthn ah
moritz std: $.foo
p6eval std 48335fc: OUTPUT«===SORRY!===␤Variable $.foo used where no 'self' is available at /tmp/HWV9y7rZDq line 1:␤------> <BOL>⏏$.foo␤Check failed␤FAILED 00:01 108m␤»
jnthn Is this for the $.foo bug?
moritz yes
at least that's what I'm trying to fix :-) 20:48
jnthn moritz: Yeah, that's best done by the way STD does it
moritz: Which is by tracking invocant availability in some other way that also handles $!x being used in submethods
er, $.x not being allowed there, I mean
moritz I thought I'd just check if 'self' is declared, to catch those cases that generate a Null PMC Access 20:50
20:50 nwc10 left
TimToady the remaining 'as Int' should probably be replaced with --> Int(Any) 20:51
moritz std: class A { has $.x; has $.y = $.x } 20:52
p6eval std 48335fc: OUTPUT«===SORRY!===␤Virtual call $.x may not be used on partially constructed object at /tmp/euGptSo2wj line 1:␤------> class A { has $.x; has $.y = ⏏$.x }␤Check failed␤FAILED 00:01 110m␤»
moritz SVG::Plot uses that syntax heavily :/
TimToady if you really want to cheat, you can say $.y = self.x 20:53
jnthn moritz: Was the virtualness intended or not?
TimToady but we don't turn on virtualness of $.x by default there
because it often is a think-o
moritz jnthn: no. I just wanted dependent defaults
TimToady $.y = $!x would work too
jnthn std: class A { has $.x; has $.y = $!x }
masak TimToady: +1 on --> Int(Any)
p6eval std 48335fc: OUTPUT«ok 00:01 110m␤»
TimToady funny thing about --> Int(Any) is it's opposite on what the type looks like from the outside, in a contravariant kind of way 20:54
moritz jnthn: maybe I'll tackle $.foo in the STD way tomorrow
jnthn moritz: +1 20:55
masak TimToady: er. it's params that are contravariant, and return values are covariant :)
TimToady sub foo (Int(Any) $x --> Int(Any)) {...} looks to the outside world like (Any --> Int)
masak TimToady: I finally learned that by heart when re-learning Java generics not long ago.
TimToady masak: yes, I always get those backwards :)
masak :P
TimToady and to the inside like (Int --> Any) 20:56
masak TimToady: it's quite easy. it's not really params vs return values, it's consumers vs producers.
consumers are naturally contravariant, producers naturally covariant.
TimToady sure, and that's why it makes sense to use Int(Any) in both places
masak er. perhaps I should call'em "sinks" and "sources". that seems to be the prevailing terminology.
TimToady we don't have to tell people it will do what they expect :) 20:57
masak :)
supernovus Hmm, another interesting bug/feature. In DateTime::Math there are + and - subs. The one for adding seconds to a DateTime object works fine: multi infix:<+>(DateTime $dt, Numeric $x) is export { ... } but the one for substracting seconds from a DateTime object never gets dispatched to: multi infix<->(DateTime $dt, Numeric $x) is export { ... } instead a default infix:<->(\$a, \$b) { ... } from the CORE.setting gets called.
masak read-write containers can't be co-/contravariant, because they try to be both sources and sinks, and therefore end up being neither.
TimToady they're invariant in their variance :) 20:58
20:58 twirp left
supernovus and no, the typo between infix and <-> is not in the actual code, that's just me mistyping things instead of copy/pasting. 20:59
skids
.oO(In soviet Perl6 the context of "sink" alters expectations of "kitchen")
 21:00
TimToady masak: though copy-in/copy-out params can coerce both ways; we don't really have those 21:07
is copy is rw ?
21:09 cognominal left 21:10 cognominal joined
TimToady probably not something we need to encourage right now, but might be useful for explicit boxing/unboxing coercions, I suppose 21:10
21:12 birdwindupbird left
supernovus guh, hand testing shows a bunch of the infix: stuf is not dispatching to the custom versions, but instead is going to the "default" versions in the setting. For instance >, >=, <=, !=, cmp, and <=> (as well as +) are going to the custom version, but <, == and of course - aren't. This seems slightly odd to me. 21:13
benabik Are they defined differently in the setting? 21:14
21:19 timotimo joined, PacoLinux joined
supernovus benabik: not that I can tell. For instance, multi infix:<+>(\$a, \$b) { ... } in the setting gets overridden by multi infix:<+>(DateTime $dt, Numeric $x) is export { ... } in the library, but multi infix:<->(\$a, \$b) { ... } DOES NOT get overriden by multi infix:<->(DateTime $dt, Numeric $x) is export { ... } in the library. There seems to be no reason for the inconsistencies. 21:20
PerlJam supernovus: show the code? 21:21
21:22 zby_home joined
supernovus PerlJam: github.com/supernovus/temporal-utils the library in question is lib/DateTime/Math.pm6 and the test in question is t/DateTime-math.t 21:22
21:28 kaare_ left 21:29 PacoLinux left
skids If I'm getting things right, the only reason I can't seem to figure out how to write a class that is cool with being assigned integers is because that should try postcircumfix:<( )> -- as long as there's an "as Int" a.k.a. Int(Mu) there -- but A) postcircumfix:<( )>'s actually gets an unchewed capture and B) just NIY? Or do I miss a trick? 21:38
supernovus nom: class Foo {}; multi infix:<+>(Foo $a, Numeric $b) { say "Foo went up $b"; }; multi infix:<->(Foo $a, Numeric $b) { say "Foo went down $b"; }; my $a = Foo.new; $a + 1; $a - 1; 21:41
p6eval nom ce5cca: OUTPUT«Foo went up 1␤No applicable candidates found to dispatch to for 'Numeric'. Available candidates are:␤:(Mu:U \$v, Mu %_!)␤␤ in method Numeric at src/gen/CORE.setting:649␤ in sub infix:<-> at src/gen/CORE.setting:2233␤ in block <anon> at /tmp/MJakwXcjJO:1␤ in <anon>…
skids Heh, well I suppose that one way around it :-) 21:42
PerlJam nom: class Foo {}; class Bar {}; multi infix:<+>(Foo $a, Bar $b) { say "up" }; multi infix:<->(Foo $a, Bar $b) { say "down" }; my $a = Foo.new; my $b = Bar.new; $a + $b; $a - $b; 21:47
p6eval nom ce5cca: OUTPUT«up␤No applicable candidates found to dispatch to for 'Numeric'. Available candidates are:␤:(Mu:U \$v, Mu %_!)␤␤ in method Numeric at src/gen/CORE.setting:649␤ in sub infix:<-> at src/gen/CORE.setting:2233␤ in block <anon> at /tmp/TmAPl2na81:1␤ in <anon> at /tmp/Tm…
21:48 BlueMatt joined
supernovus PerlJam: Yeah, it seems -, == and < are the operators that don't seem to be overridable at the moment. All the rest work fine. Beats me. :p 21:48
benabik nom: class Foo {}; multi infix:<->(Foo $a, Numeric $b) { say "down" }; multi infix:<+>(Foo $a, Numeric $b) { say "up" }; my $a = Foo.new; $a - 1; $a +1
p6eval nom ce5cca: OUTPUT«No applicable candidates found to dispatch to for 'Numeric'. Available candidates are:␤:(Mu:U \$v, Mu %_!)␤␤ in method Numeric at src/gen/CORE.setting:649␤ in sub infix:<-> at src/gen/CORE.setting:2233␤ in block <anon> at /tmp/BWzKASfJsI:1␤ in <anon> at /tmp/BWzKA…
benabik Not order dependent. :-)
21:49 BlueMatt left
PerlJam mine showed that it's trying to turn something Numeric though. (there was no Numeric in my code, just the setting) 21:51
supernovus Yeah, it's getting dispatched to the infix:<->(\$a, \$b) { $a.Numeric - $b.Numeric } version from the setting rather than the custom version. 21:52
21:53 imarcusthis joined 21:56 imarcusthis left 21:57 skids left 21:59 imarcusthis joined
spine nom: (1=>1).map({ .key => 1}).say; 22:02
p6eval nom ce5cca: OUTPUT«Method 'key' not found for invocant of class 'Any'␤ in block <anon> at /tmp/QxKemmos8k:1␤ in <anon> at /tmp/QxKemmos8k:1␤»
spine nom: (1=>1).map({ ; .key => 1}).say;
p6eval nom ce5cca: OUTPUT«1 1␤»
spine I don't get it why those produce different output...
22:02 cooper left, cooper joined
supernovus Well, I marked DateTime::Utils as working in nom, and noted the exception for DateTime::Math, which is broken until we can figure out the issue with certain infix operators not being overridable outside the setting. 22:03
22:09 mj41 joined 22:25 cj joined 22:35 ksi left 22:39 bbkr1 joined 22:43 zby_home left 22:48 vlixes joined 23:12 kmwallio left 23:37 mj41 left 23:44 spine left, PacoAir left 23:47 whiteknight joined 23:57 supernovus left

Powered by Cro and the Raku® Programming Language.

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

Thank you!