perl6-projects.org/ | nopaste: sial.org/pbot/perl6 | evalbot: 'perl6: say 3;' | irclog: irc.pugscode.org/ | ~290 days 'til Xmas
Set by mncharity on 6 March 2009.
wayland76 @tell moritz_ What was wrong with the bot was we put it in an infinite loop waiting for input :) 00:13
lambdabot Consider it noted.
wayland76 @tell cspencer someone reorganised the test suite by putting lots of S29 stuff in S32. I reorganised the spec like this, but someone else (I forget who now) did the spectests 00:28
lambdabot Consider it noted.
00:52 WootKit joined 00:53 ujwalic left 00:55 nbrown joined 01:19 hercynium joined 01:20 justatheory left 01:38 WootKit left 01:41 Whiteknight left 01:43 kate21de1 left 01:50 DemoFreak left 01:57 justatheory joined 01:59 revdiablo joined 02:03 alester joined 02:04 Coleoid left 02:35 diakopter left 02:36 TimToady joined, justatheory left, diakopter joined 02:40 justatheory joined 02:48 alester left 02:53 justatheory left 02:54 justatheory joined 03:02 justatheory left 03:06 justatheory joined 03:25 justatheory left 04:04 SamB left 04:08 pidus left 04:17 meppuru joined 04:32 meppl left 04:35 meppuru is now known as meppl
wayland76 Oh, meppuru = meppl. I didn't know that :) 04:36
04:36 Kimtaro_ is now known as Kimtaro
meppl wayland76, "meppuru" is what happens, when japanese people try to pronounciate "meppl" 04:41
04:42 hercynium left 04:58 justatheory joined 05:01 finanalyst joined 05:06 c9s__ left 05:13 justatheory left
wayland76 meppl: :) 05:15
05:15 kjwcode joined 05:21 sri_kraih joined 05:29 REPLeffect joined 05:40 sri_kraih_ left 05:42 Kimtaro left 05:43 ewilhelm left 05:45 pidus joined 05:49 AzureStone left 05:50 AzureStone joined 05:51 Kimtaro joined 05:53 Tene joined 06:01 kate21de joined 06:05 ewilhelm joined 06:06 Tene_ left 06:09 nihiliad left 06:10 cai joined 06:15 REPLeffect left 06:17 kjwcode left 06:26 cai_ joined 06:27 ujwalic joined 06:33 araujo left 06:48 cai left 06:49 alc joined
pugs_svn r25754 | lwall++ | [STD] refactor ? twigil checking so %?FOO is handled correctly in strings 06:58
r25754 | lwall++ | [STD] canonicalize packaged vars before lookup
r25754 | lwall++ | [STD] allow $?CALLER::LINE and such via canonicalization to CALLER::<$?LINE>
r25754 | lwall++ | [STD] suppress complaints about DEBUG
07:12 kate21de left 07:13 kate21de joined 07:16 WootKit joined 07:19 WootKit left
pugs_svn r25755 | moritz++ | [evalbot] try to avoid infinite loops when some program reads from STDIN 07:23
07:26 p6eval left 07:27 p6eval joined 07:43 masak joined
pugs_svn r25756 | masak++ | [u4x/TODO] added 'word' grammar rule 07:45
masak S11:365:
use Dog:ver(Any):auth({ .substr(0,5) eq 'cpan:'}) 07:46
While "Any" here is cute and pronounceable, I'd think both "*" and "Whatever" would be more logical.
Also, can modules be re-entered without using 'augment' or some such? S11 doesn't seem to have an opinion on that.
moritz_ yes, * seems more logical
lambdabot moritz_: You have 1 new message. '/msg lambdabot @messages' to read it.
moritz_ @massages
lambdabot wayland76 said 7h 33m 42s ago: What was wrong with the bot was we put it in an infinite loop waiting for input :)
masak moritz_: I thought the evalbots had a 60-second timeout or some such. 07:47
moritz_ masak: so did I; apparently it didn't work in connection with IO
perhaps it only limits CPU time, not real execution time 07:48
masak that's explain it.
s/s/d/
moritz_ more general solutions are welcome, of course ;-)
masak more questions: Ruby has Module#instance_method -- is there an equivalent in Perl 6? gist.github.com/76158 07:49
moritz_ does the promote a method to a sub? 07:51
s/the/that/
masak not instance_method in itself, no. 07:52
moritz_ then I don't understand what it does 07:53
masak rather, it fetches a method that is supposed to be called on an object, but it isn't bound to an actual instance yet.
the do_a et al are already instance methods in Interpreter.
Dispatcher, then, is a hash of bytes to this kind of UnboundMethod objects. 07:54
finally, the loop in interpret(string) binds and calls.
I guess my question is: does Perl 6 have an equivalent of UnboundMethod?
moritz_ so all it does is taking a reference to a method?
masak right. 07:55
moritz_ &Class::method ?
masak and in the case of multis?
hm, maybe that isn't a problem.
moritz_ the & thing points to a collection of multis in the general case, I think 07:56
masak as it probably should. 07:58
is there a way to do the dispatch without the call?
moritz_ dunno 07:59
can you think of an application for that?
masak that would probably be highly desirable in some cases.
yes, something when you want pattern matching but not method calls for some reason.
maybe such a use case is bogus. you might as well make the call as well.
no, wait. delayed calls! 08:00
you do dispatch on the server side, and send a method object to the client side. or something.
moritz_ that sounds a bit like abuse of the dispatcher ;-)
in this case you can wrap the multi prior to dispatch, and unwrap later 08:01
masak ok.
moritz_ that said, maybe there *is* such an API, and I just don't know about it 08:02
masak understood. 08:03
Matt-W Morning 08:05
wayland76 Matt-W: good localtime() (with apologies to whoever I stole that from :) )
Matt-W :)
Did everyone have a wonderful weekend?
masak 08:06
moritz_ Matt-W: sure, but it was way too short
masak they all are.
Matt-W yup
wayland76 I did too. But life is too short :)
Matt-W mine finished with watching the police climb over my garden gate and drag a man out of the shed under the glare of the helicopter searchlight, so it was... unusual 08:07
masak wow.
never a boring moment, eh?
wayland76 I hope that was your weekend and not your life that ended that way :)
Matt-W but I do like to see stupid people get their comeuppance
wayland76 masak: Where is the Web stuff being discussed?
Matt-W and trying to hide from the helicopter's heat camera in a very wobbly shed is really, really stupid
masak wayland76: two places. #november-wiki and november-wiki as googlegroups 08:08
moritz_ snickers at today's xkcd 08:10
08:17 bacek_ left 08:24 kate21de left 08:30 DemoFreak joined
bacek good evening 08:58
masak bacek: marnin' 08:59
bacek masak: still lives in past? :) 09:00
masak bacek: yes, I'd like a few hours warning every day before the next one arrives. 09:01
I'm a late adopter in that sense, I guess.
bacek masak: Good point. You have to trade something to get something. 09:02
masak that's life.
09:04 ujwalic_ joined 09:06 pidus left
finanalyst evalbot: perl6: my $x='ab';my $y='xxyuabhh';$y~~m/$x/ ?? say 'variables work' !! say 'variables dont match'; 09:16
perl6: my $x='ab';my $y='xxyuabhh';$y~~m/$x/ ?? say 'variables work' !! say 'variables dont match';
p6eval rakudo 8bbc31: OUTPUT«Null PMC access in get_string()␤current instr.: 'parrot;PGE;Grammar;' pc 300 (EVAL_17:121)␤» 09:17
..elf 25756: OUTPUT«Unknown rule: infix:conditional␤It needs to be added to ast_handlers.␤ at ./elf_h line 2044␤»
..pugs: OUTPUT«Error eval perl5: "if (!$INC{'Pugs/Runtime/Match/HsBridge.pm'}) {␤ unshift @INC, '/home/evalenv/pugs/perl5/Pugs-Compiler-Rule/lib';␤ unshift @INC, '/home/evalenv/pugs/third-party/Parse-Yapp/lib';␤ eval q[require 'Pugs/Runtime/Match/HsBridge.pm'] or die $@;␤}␤'Pugs::Runti...
09:20 ujwalic left 09:21 rkendall joined 09:22 alc left
finanalyst perl6: my $x='ab';my $y='xxyuabhh';$y~m/$x/ and say 'variables work'; 09:24
p6eval elf 25756, pugs, rakudo 8bbc31: OUTPUT«variables work␤»
moritz_ ~ does a string concatenation 09:25
probably not quite what you want
finanalyst no it wasnt
pugs_svn r25757 | bacek++ | [t/spec] Fix subroutine form of max.
finanalyst but i thought pugs could match against variab;es 09:26
moritz_ pugs: my $var = 'a'; say "foobar" ~~ /$var/;
p6eval pugs: OUTPUT«Error eval perl5: "if (!$INC{'Pugs/Runtime/Match/HsBridge.pm'}) {␤ unshift @INC, '/home/evalenv/pugs/perl5/Pugs-Compiler-Rule/lib';␤ unshift @INC, '/home/evalenv/pugs/third-party/Parse-Yapp/lib';␤ eval q[require 'Pugs/Runtime/Match/HsBridge.pm'] or die $@;␤}␤'Pugs::Runti...
moritz_ rakudo: my $var = 'a'; say "foobar" ~~ /$var/;
p6eval rakudo 8bbc31: OUTPUT«Null PMC access in get_string()␤current instr.: 'parrot;PGE;Grammar;' pc 249 (EVAL_17:106)␤»
moritz_ it seems neither can
finanalyst matching against variables is known not to work in rakudo 09:28
bacek rakudo: say reduce -> $a, $b?, $c? { $a + ($b//0) + ($c//0) }, 1,2;
p6eval rakudo 8bbc31: OUTPUT«Cannot reduce() using a unary or nullary function.␤current instr.: 'die' pc 16009 (src/builtins/control.pir:204)␤»
bacek rakudo: my $f = -> $a, $b?, $c? { ... }; say $f.arity
p6eval rakudo 8bbc31: OUTPUT«1␤»
finanalyst i dont understand why it has such low priority for fixing
bacek ouch.... 09:29
rakudo: my $f = -> $a, $b, $c? { ... }; say $f.arity
p6eval rakudo 8bbc31: OUTPUT«2␤»
bacek summons masak
masak raaawr!
masak submits rakudobug
wait, isn't that correct?
bacek Team work in action :)
Why it's correct??? 09:30
masak it has two mandatory args.
bacek but I expect arity to be 3.
masak bacek: please provide evidence in spec for this.
then I'll submit the bug for you.
bacek masak: no, it's not a bug... 09:31
$args_required = &foo.arity;
masak right. 09:32
an interesting question is which method gives you a 3.
bacek rakudo: my $f = -> $a, $b, $c? { ... }; say $f.count
p6eval rakudo 8bbc31: OUTPUT«Method 'count' not found for invocant of class 'Block'␤current instr.: 'parrot;P6metaclass;dispatch' pc 320 (src/classes/ClassHOW.pir:161)␤»
wayland76 rakudo: my $f = -> $a, $b, $c? { ... }; say $f.maxarity
p6eval rakudo 8bbc31: OUTPUT«Method 'maxarity' not found for invocant of class 'Block'␤current instr.: 'parrot;P6metaclass;dispatch' pc 320 (src/classes/ClassHOW.pir:161)␤»
wayland76 Well, I liked it, anyway :) 09:33
bacek I can't spot proper method to find number of all arguments...
masak: but I still have bug for you :)
rakudo: say reduce -> $a, $b?, $c? { $a + ($b//0) + ($c//0) }, 1,2; 09:34
p6eval rakudo 8bbc31: OUTPUT«Cannot reduce() using a unary or nullary function.␤current instr.: 'die' pc 16009 (src/builtins/control.pir:204)␤»
masak haha :)
masak submits
bacek It's because of '.arity'...
moritz_ rakudo: my $x = -> $a, $b?, $c? { ... }; say $x.count 09:35
p6eval rakudo 8bbc31: OUTPUT«Method 'count' not found for invocant of class 'Block'␤current instr.: 'parrot;P6metaclass;dispatch' pc 320 (src/classes/ClassHOW.pir:161)␤»
moritz_ rakudo: my $x = -> $a, $b?, $c? { ... }; say $x.arity
p6eval rakudo 8bbc31: OUTPUT«1␤»
bacek moritz_: Hey. I already checked it :) 09:36
moritz_ rakudo: my $x = { $_*2 }; say $x.arity
p6eval rakudo 8bbc31: OUTPUT«0␤»
moritz_ that however is wrong
masak submits 09:37
bacek rakudo: my $f = -> $a, $b, $c? { ... }; say $f.signature
ujwalic_ where to find for rakudo install dir layout
p6eval rakudo 8bbc31: OUTPUT«Signature<0xb6193c90>␤»
bacek rakudo: my $f = -> $a, $b, $c? { ... }; say $f.signature.perl
p6eval rakudo 8bbc31: OUTPUT«:(Any $a, Any $b, Any $c?)␤»
bacek rakudo: my $f = -> $a, $b, $c? { ... }; say $f.signature.params 09:38
p6eval rakudo 8bbc31: OUTPUT«Null PMC access in find_method()␤current instr.: 'infix:=>' pc 10086 (src/classes/Pair.pir:44)␤»
moritz_ rakudo++ # I didn't know it did that
masak that Null PMC access is already reported, btw. 09:39
bacek Which spec describes Signature ?
rakudo: my $f = -> $a, $b, $c? { ... }; say +$f.signature.params 09:40
p6eval rakudo 8bbc31: OUTPUT«3␤»
moritz_ probably S02 or S06, if any
bacek hooray. Here we go.
wayland76 Should be in S32/Callable as well
moritz_ rakudo: multi f() {...}; multi f($x) {...}; say $f.signature.params.WHAT 09:41
p6eval rakudo 8bbc31: OUTPUT«Scope not found for PAST::Var '$f' in ␤current instr.: 'parrot;PCT;HLLCompiler;panic' pc 146 (src/PCT/HLLCompiler.pir:102)␤»
moritz_ rakudo: multi f() {...}; multi f($x) {...}; say &f.signature.params.WHAT
p6eval rakudo 8bbc31: OUTPUT«Null PMC access in find_method()␤current instr.: '_block14' pc 84 (EVAL_21:45)␤»
moritz_ rakudo: multi sub f($z) {}; &f.arity.say 09:44
p6eval rakudo 8bbc31: OUTPUT«Null PMC access in find_method()␤current instr.: '_block14' pc 79 (EVAL_16:43)␤»
moritz_ that's RT #63568 09:45
masak aye.
moritz_ I'd expect an any() junction 09:46
but maybe I'm wrong there
bacek rakudo: my $f = -> $a, $b, $c? { ... }; say $f.signature.params[0].WHAT 09:47
p6eval rakudo 8bbc31: OUTPUT«Mapping␤»
bacek rakudo: my $f = -> $a, $b, $c? { ... }; say $f.signature.params[0]
p6eval rakudo 8bbc31: OUTPUT«Null PMC access in find_method()␤current instr.: 'infix:=>' pc 10086 (src/classes/Pair.pir:44)␤» 09:48
bacek rakudo: my $f = -> $a, $b, $c? { ... }; say $f.signature.params[0].perl
p6eval rakudo 8bbc31: OUTPUT«Null PMC access in find_method()␤current instr.: 'infix:=>' pc 10086 (src/classes/Pair.pir:44)␤»
bacek rakudo: my $f = -> $a, $b, $c? { ... }; say $f.signature.params[0].key
p6eval rakudo 8bbc31: OUTPUT«Method 'key' not found for invocant of class 'Hash'␤current instr.: 'parrot;P6metaclass;dispatch' pc 320 (src/classes/ClassHOW.pir:161)␤»
bacek rakudo: my $f = -> $a, $b, $c? { ... }; say $f.signature.params[0].kv
p6eval rakudo 8bbc31: OUTPUT«Null PMC access in find_method()␤current instr.: 'infix:=>' pc 10086 (src/classes/Pair.pir:44)␤»
bacek rakudo: my $f = -> $a, $b, $c? { ... }; say $f.signature.params[0].keys
p6eval rakudo 8bbc31: OUTPUT«Null PMC access in find_method()␤current instr.: 'infix:=>' pc 10086 (src/classes/Pair.pir:44)␤»
bacek rakudo: my $f = -> $a, $b, $c? { ... }; say $f.signature.params[0].values
p6eval rakudo 8bbc31: OUTPUT«Null PMC access in find_method()␤current instr.: 'infix:=>' pc 10086 (src/classes/Pair.pir:44)␤»
bacek rakudo: say reduce -> $a, $b, $c { $a + ($b//0) + ($c//0) }, 1,2; 09:54
p6eval rakudo 8bbc31: OUTPUT«3␤»
bacek rakudo: sub f (Code $c) { say +$c.signature.params }; f(-> $a, $b? {...}) 09:56
p6eval rakudo 8bbc31: OUTPUT«Method 'params' not found for invocant of class ''␤current instr.: 'parrot;P6metaclass;dispatch' pc 320 (src/classes/ClassHOW.pir:161)␤»
bacek summon masak again...
masak raaawr again! 09:57
masak submits
bacek rakudo: my Code $c = -> $a {...}; say $c.signature.perl
p6eval rakudo 8bbc31: OUTPUT«:(Any $a)␤»
bacek rakudo: my Code $c = -> $a {...}; say $c.signature.perl; sub f(Code $f) { say $f.signature.perl }; f($c); 09:58
p6eval rakudo 8bbc31: OUTPUT«:(Any $a)␤Failure␤»
bacek interesting... 09:59
Ah! 10:00
rakudo: my Code $c = -> $a {...}; say $c.signature.perl; sub f(Code $f is copy) { say $f.signature.perl }; f($c);
p6eval rakudo 8bbc31: OUTPUT«:(Any $a)␤Failure␤»
bacek rakudo: my Code $c = -> $a {...}; say $c.signature.perl; sub f(Code $f is rw) { say $f.signature.perl }; f($c);
p6eval rakudo 8bbc31: OUTPUT«:(Any $a)␤:(Any $a)␤»
10:01 cai_ left
bacek yes. ObjectRef doesn't proxy .signature properly. 10:01
or it is ClassHOW? 10:06
rakudo: my Code $c = -> $a {...}; say $c.signature.perl; sub f(Code $f is rw) { say $f.blah }; f($c); 10:11
p6eval rakudo 8bbc31: OUTPUT«:(Any $a)␤Method 'blah' not found for invocant of class 'Block'␤current instr.: 'parrot;P6metaclass;dispatch' pc 320 (src/classes/ClassHOW.pir:161)␤»
10:17 bacek_ joined
bacek rakudo: my Code $c = -> $a {...}; my ObjectRef $o = $c; say $o.signature.perl 10:17
hmmm... I broke evalbot again?
10:17 bacek left
p6eval rakudo 8bbc31: OUTPUT«Malformed declaration at line 1, near "ObjectRef "␤␤current instr.: 'parrot;PGE;Util;die' pc 129 (runtime/parrot/library/PGE/Util.pir:83)␤» 10:17
masak baest: isn't ObjectRef a Parrot type? 10:19
moritz_ I think it's a Rakudo type 10:20
10:20 SamB joined
masak ok. but it might still be an internal type, right? 10:21
moritz_ yes 10:22
bacek_ rakudo: my Code $c = -> $a {...}; my Any $a=$c; say $a.signature.perl 10:24
p6eval rakudo 8bbc31: OUTPUT«:(Any $a)␤»
bacek_ rakudo: my Code $c = -> $a {...}; sub f(Code $f) { say $f.HOW }; f($c); 10:25
p6eval rakudo 8bbc31: OUTPUT«get_string() not implemented in class 'P6metaclass'␤current instr.: 'print' pc 19094 (src/builtins/io.pir:26)␤»
bacek_ rakudo: my Code $c = -> $a {...}; sub f(Code $f) { say $f.perl }; f($c);
p6eval rakudo 8bbc31: OUTPUT«{ ... }␤» 10:26
bacek_ rakudo: class Foo { has $!foo = 42; method bar { say $!foo } }; my Foo $f=Foo.new; $f.bar; 10:27
p6eval rakudo 8bbc31: OUTPUT«42␤»
bacek_ rakudo: class Foo { has $!foo = 42; method bar { say $!foo } }; my Foo $f=Foo.new; $f.bar; sub blah($b) { $b.bar }; blah($f); 10:28
p6eval rakudo 8bbc31: OUTPUT«42␤42␤»
bacek_ rakudo: sub foo($x) { say ~$x }; foo([1..2]); 10:30
p6eval rakudo 8bbc31: OUTPUT«1 2␤»
10:30 ujwalic_ left
bacek_ rakudo: sub foo($x) { say ~$x }; foo(\(1..2)); 10:30
p6eval rakudo 8bbc31: OUTPUT«1 2␤»
bacek_ rakudo: sub foo($x) { say ~$x }; say foo(1|2); 10:32
p6eval rakudo 8bbc31: OUTPUT«1␤2␤Junction<0xb6145018>␤»
bacek_ rakudo: sub foo($x) { say ~$x }; foo(1|2);
p6eval rakudo 8bbc31: OUTPUT«1␤2␤»
bacek_ yak... 10:34
moritz_ what's wrong with that?
it autothreads, as it should
bacek_ If I understand correctly we has "parent-loop" for Code. 10:35
moritz_ I wonder if types in pointy blocks should default to Object instead of Any
bacek_ Code parents Sub and Any, Sub parent Routing, Routine parent is Block, Block's parent is Code...
Sub's parent Routine. 10:36
moritz_ Code is Any, and all others should inherit from Code
bacek_ moritz_: lemme check it in local repo. 10:37
moritz_ rakudo: say Code ~~ Any 10:38
p6eval rakudo 8bbc31: OUTPUT«1␤»
moritz_ rakudo: say Block ~~ Code
p6eval rakudo 8bbc31: OUTPUT«1␤»
moritz_ rakudo: say Routine ~~ Code
p6eval rakudo 8bbc31: OUTPUT«1␤»
moritz_ rakudo: say Sub ~~ Routine
p6eval rakudo 8bbc31: OUTPUT«1␤»
moritz_ rakudo: say Method ~~ Routine 10:39
p6eval rakudo 8bbc31: OUTPUT«1␤»
bacek_ rakudo: say Code ~~ Sub
moritz_ rakudo: say Routine ~~ Block
p6eval rakudo 8bbc31: OUTPUT«0␤»
rakudo 8bbc31: OUTPUT«1␤»
bacek_ hmm...
moritz_ maybe one of them is actually a role?
everything is as I expect, except for Code ~~ Sub 10:40
wayland76 Code is the parent of sub
bacek_ yes, I misread Sub.pir. 10:41
moritz_ rakudo: say Sub ~~ Code
p6eval rakudo 8bbc31: OUTPUT«1␤»
moritz_ rakudo: say Code ~~ Sub
p6eval rakudo 8bbc31: OUTPUT«0␤»
moritz_ ah
wayland76 Not parent, the other thing
moritz_ rakudo: say Routine ~~ Block
p6eval rakudo 8bbc31: OUTPUT«1␤»
wayland76 Block does Code
Routine does Block
Sub does Routine
See S32/Callable :)
moritz_ isn't Callable the role that it should do? 10:42
moritz_ confused
wayland76 Sorry, S32/Callable includes all related roles
(Block Sub, Routine, etc)
Kind of like Containers has a bunch of different things in it 10:43
moritz_ yes, I understood that 10:46
well, I have to do other stuff anyway :/
wayland76 is multitasking 10:49
mikehh built rakudo [99bbc31] on parrot r37235 - spectest passes (35min for the spectest) 10:57
masak mikehh++ 10:59
11:10 meppl left 11:33 ruoso joined
ruoso Hello! 11:34
wayland76 hello (except I'm mostly not here :) ) 11:37
masak hello (just got back) 11:39
ruoso I think I'll take the fact that most of SMOP need to be refactored to the new capture 11:46
and also refactor the several parts of SMOP into separated modules
to make the dependencies more clear...
masak ruoso++ 11:56
12:02 ihrd joined
rkendall Hi, just have a question about the legal aspects of Rakuo. The Perl Foundation looks after Perl 5, The Parrot Foundation looks after Parrot, but where does Rakudo fit in? (eg. managing trademarks, licence etc) 12:04
moritz_ TPF I think 12:05
rkendall ...and what about other implementations of Perl 6? 12:06
moritz_ that depends on the implementation
pugs is copyright "the pugs contributors", and they are listed in CREDITS
rkendall thanks moritz - should probably mention something about that on the website
i.e. Rakudo.org 12:07
moritz_ anyway, better ask pmichaud first
rkendall ok
noticed talk about 'contributor agreements' 12:08
but I guess it's still being worked out 12:09
moritz_ when it was still part of the parrot repo, every committer had to send in a license agreement
now that we moved out we need some kind of policy
but that's the pumpking's decision 12:10
rkendall ok, thanks
ruoso really would love to be able to stop worrying with things like licensing... 12:20
ruoso .oO( How I Learned to Stop Worrying and Love the Public Domain ) 12:22
moritz_ well, you could, if everyting was public domain :-)
ruoso but I actually learned that public domain is not a good idea everywhere 12:24
in Portugal, and it seems most of Europe, public domain means "property of the state"
which looks very bad
in Brazil public domain simply means "not subject to copyright/author-right laws" 12:25
moritz_ in Germany you can't place code in the "public domain" in the brazil/US sense
because you always retain some rights as the author, even if you don't want to
so it's better to chose a particular license 12:26
ruoso in Brazil, it's just a matter of making a public statement "this is in the public domain"
there was even an interesting case in Brasil 12:28
where a famous singer
visited a country-side community, that had some traditional songs
the "musica de quixabeira"
and then he recorded an album stating that musics were public domain
the people in the community entered in a lawsuit to revert that, 12:29
and they actually won
12:38 skids left
ruoso TimToady, have you thought a bit more about the packages aliasing issue? I'm quite convinced the idea of "use" cloning the packages is very good, while if you really want to access a package globally, you would need to access GLOBAL::My::Package::<$VARIABLE> 12:39
that way every package can have its meaning lexically re-defined.. 12:43
and it should also allow several interesting optimizations, since the use of globals would be very well determined... 12:44
everything without a GLOBAL:: in the front might be considered local to that lexical scope
with the benefit of making the use of globals very de-huffmanized 12:49
rkendall masak: very glad to hear about Web.pm 13:00
masak rkendall: thank you. :)
ruoso masak, I suppose you don't mind I throw out crazy ideas for Web.pm here, do you?
masak ruoso: au contraire, please do. 13:01
ruoso one of the most important featuers I've been using in Catalyst is the fact that it is engine-agnostic
masak ruoso: but know that the main discussion takes place on #november-wiki, as we speak and in general.
ruoso oh
at freenode?
masak aye.
ruoso ETOOMANYCHANNELS.... need to leave some 13:02
masak has 10 channels open in Emacs ERC -- could easily have a few more
Matt-W masak: is that on this network? 13:03
masak Emacs rocks.
Matt-W: aye.
ruoso masak, it's not the client that's blocking me... it's the server
masak oh.
13:06 gfldex is now known as Guest82337, gfldex_ joined 13:14 finanalyst left 13:18 Guest82337 left 13:22 ujwalic joined
ruoso pugs: class A {}; role B { method foo { say 1 }}; role C { method foo { say 2} }; my $a = A.new; $a does B; $a does C; $a.*foo() 13:45
p6eval pugs: OUTPUT«*** Unknown syntactic construct: Syn "does" [Var "$a",Val (VType (mkType "B"))]␤ at /tmp/kKlMbMIb5Q line 1, column 92-101␤»
ruoso rakudo: class A {}; role B { method foo { say 1 }}; role C { method foo { say 2} }; my $a = A.new; $a does B; $a does C; $a.*foo()
p6eval rakudo 8bbc31: RESULT«[]»
ruoso rakudo: class A {}; role B { method foo { say 1 }}; role C { method foo { say 2} }; my $a = A.new; $a does B; $a does C; $a.?foo() 13:46
p6eval rakudo 8bbc31: RESULT«undef»
ruoso rakudo: role B { method foo { say 1 }}; role C { method foo { say 2} }; class A does B does C { }; A.*foo()
p6eval rakudo 8bbc31: OUTPUT«A conflict occurred during role composition due to method 'foo'.␤current instr.: '!meta_compose' pc 342508754 ((unknown file):-1)␤»
moritz_ rakudo: class A { multi method b ($x, $y?) { say "b1" }; multi method b($x) { say "b2" } }; A.new.+b(3) 13:48
p6eval rakudo 8bbc31: OUTPUT«b1␤b2␤»
moritz_ rakudo: class A { multi method b ($x, $y?) { say "b1" }; multi method b($x) { say "b2" } }; A.new.*b(3)
p6eval rakudo 8bbc31: OUTPUT«b1␤b2␤»
moritz_ rakudo: class A { multi method b ($x, $y?) { say "b1" }; multi method b($x) { say "b2" } }; A.new.?b(3)
p6eval rakudo 8bbc31: OUTPUT«b1␤»
ruoso hmmm
they need to be multi
right
rakudo: role B { multi method foo { say 1 }}; role C { multi method foo { say 2} }; class A does B does C { }; A.*foo() 13:49
p6eval rakudo 8bbc31: OUTPUT«1␤2␤»
ruoso moritz_++
rakudo: role B { multi method foo { say 1 }}; role C { multi method foo { say 2} }; class A does B does C { }; A.foo() # that should fail
p6eval rakudo 8bbc31: OUTPUT«Ambiguous dispatch to multi 'foo'. Ambiguous candidates had signatures:␤:(Object self)␤:(Object self)␤␤current instr.: 'parrot;P6metaclass;dispatch' pc 130 (src/classes/ClassHOW.pir:93)␤»
13:50 skids joined
moritz_ rakudo: role B { multi method foo { say 1 }}; role C { multi method foo { say 2} }; class A does B does C { }; A.*foo() 13:51
p6eval rakudo 8bbc31: OUTPUT«1␤2␤»
bacek_ bacek@icering:~/src/parrot/languages/rakudo.bacek$ ./perl6 -e 'say (1..4).reduce: { $^a + $^b + $^c }' 13:52
too few arguments passed (2) - 3 params expected
bacek@icering:~/src/parrot/languages/rakudo.bacek$ ./perl6 -e 'say (1..4).reduce: -> $a, $b, $c? { $a + $b + ($c//0) }'
10
Hooray. I did it :)
masak bacek_++ 13:53
bacek_ And it's pure Perl6 implementation :)
moritz_ and let me guess, it's less than 20 lines long 13:54
bacek_ moritz_: 21 including empty lines and comments about workarounds 13:55
pmichaud what's the meaning of a 3-argument reduce?
bacek_ nopaste.snit.ch/15826 13:56
moritz_ it gets the previous result plus two items on each call
pmichaud is that going into the synopsis somewhree? 13:57
ruoso hmm
I would expect it to consume n items at a time
moritz_ just like @list.map: -> $a, $b { ... } gets two at a time
pmichaud: it seems I have to add it :)
pmichaud: see current p6l thread
pmichaud moritz_: I did look at the thread. 13:58
I didn't see anything about multi-argument reduce.
bacek_ At least it spectested. 13:59
moritz_ "The same question also arises for "magical" List.reduce and similar methods."
13:59 mikehh left
pmichaud moritz_: that question was with respect to how to handle lists that weren't multiples of the body arity 13:59
moritz_: it didn't say how reduce should choose to pick its elements
but if we're going to declare that reduce always takes n-1 items at a time, that's fine with me. 14:00
moritz_ that seems like the obvious generalization for me
PerlJam good morning #perl6
moritz_ which works for all operators that are commutative and left-associative 14:01
pmichaud I'm spectesting ".count" for blocks.
moritz_ nice
14:01 pidus joined
moritz_ rakudo: say { $_ }.signature.perl 14:02
p6eval rakudo 8bbc31: OUTPUT«Failure␤»
PerlJam reduce wouldn't *always* take n-1 items at a time. The first time it takes n items :)
pmichaud PerlJam: no, that's not how I interpreted it.
it could just take the first item and start with that. 14:03
PerlJam oh, I see.
er, maybe 14:04
pmichaud thus say 10.reduce: -> $a, $b, $c? { $a + $b + ($c // 0) } returns 10, having never called the block.
bacek_ rakudo: sub foo ($a, $b?) {...}; say $a.arity; say $a.count
p6eval rakudo 8bbc31: OUTPUT«Scope not found for PAST::Var '$a' in ␤current instr.: 'parrot;PCT;HLLCompiler;panic' pc 146 (src/PCT/HLLCompiler.pir:102)␤»
bacek_ rakudo: sub foo ($a, $b?) {...}; say &foo.arity; say &foo.count
p6eval rakudo 8bbc31: OUTPUT«1␤Method 'count' not found for invocant of class 'Perl6Sub'␤current instr.: 'parrot;P6metaclass;dispatch' pc 320 (src/classes/ClassHOW.pir:161)␤»
pmichaud bacek_: I said I'm spectesting it now. :-) 14:05
bacek_ pmichaud: :)
pmichaud I don't commit until after the spectest passes.
bacek_ It should be specced also... Or I can't find where .count is described.
moritz_ speaking of which, running rakudo on an optimized parrot build seems quite stable, all spectests pass
pmichaud bacek_: Signature.count is described in S06
I'm presuming there's an equivalent .count for Code 14:06
bacek_ pmichaud: only in context of want().
PerlJam pm: I interpretted Larry's email on arity to mean "carp if there aren't the required number of parameters". your example has a block with 2 required parameters, so it shouldn't return 10. IMHO
pmichaud PerlJam: by that reasoning, [+] 10 should carp. 14:07
bacek_ PerlJam: there is 3 params. One of them optional.
moritz_ pmichaud: no, [+] uses explicit default values
pmichaud moritz_: the default values are for empty list.
moritz_ perl6: say [+];
p6eval pugs: OUTPUT«0␤»
..elf 25757: OUTPUT«Parse error in: /tmp/4Ab5R3K0lw␤panic at line 1 column 0 (pos 0): Can't understand next input--giving up␤WHERE: say [+];␤WHERE:/\<-- HERE␤ STD_red/prelude.rb:99:in `panic'␤ STD_red/std.rb:76:in `scan_unitstopper'␤ STD_red/std.rb:224:in `comp_unit'␤ STD_red/std.rb:210:in
..`_UNI...
..rakudo 8bbc31: OUTPUT«say requires an argument at line 1, near " [+];"␤␤current instr.: 'parrot;PGE;Util;die' pc 129 (runtime/parrot/library/PGE/Util.pir:83)␤»
pmichaud rakudo: say [+](); 14:08
p6eval rakudo 8bbc31: OUTPUT«0␤»
moritz_ pmichaud: and [+] 10 would then be (default) + 10
14:08 ihrd left
PerlJam moritz_: and the default depends on the operation? 14:09
moritz_ PerlJam: yes
perl6: my $x; $x *= 3;
p6eval pugs: RESULT«\0»
..rakudo 8bbc31: RESULT«3»
..elf 25757: OUTPUT«Unknown rule: infix_postfix_meta_operator:=␤It needs to be added to ast_handlers.␤ at ./elf_h line 2044␤»
bacek_ &prefix:<[*]> ::= &reduce.assuming(&infix:<*>, 1);
S03. 14:10
PerlJam moritz_: that tends to sound crazyish to me. How is perl to know the defaults for user-defined ops?
pmichaud S03: If fewer than two arguments are given, a dispatch is still attempted
with whatever arguments are given, and it is up to the receiver of that
dispatch to deal with fewer than two arguments.
moritz_ PerlJam: by letting the user define a nullary sub of the same name
pmichaud (this is for reduce metaoperator)
moritz_ or unary, for that matter
pmichaud also from S03: 14:11
By default, if there is one argument, the built-in reduce operators
return that one argument.
That seems to argue against moritz's [+] 10 statement above.
moritz_ aye 14:12
pmichaud thus I'd say that @list.reduce should return @list[0] if @list has only one argument.
s/argument/element.
at least, that's a reasonable interpretation.
and then all subsequent calls to the reduction function are using .count-1 elements from the list. 14:13
moritz_ that sounds good to me
bacek_ that what I've implemented in reduce :)
pmichaud I'm not advocating for a particular interpretation; I'm just trying to understand what the actual interpretation should be.
fwiw, I don't like the shifting/splicing in bacek's reduce implementation 14:15
iterators are the answer. 14:16
14:16 c9s__ joined
pmichaud in particular, doing things like my @list = $values.list; creates an Array. 14:16
when $values might not be an Array.
PerlJam reduce is repeated function application to turn N things into 1 thing. What would turn N things into M things? 14:17
moritz_ PerlJam: for example map
pmichaud PerlJam: map?
PerlJam no, map isn't quite the same.
14:17 pmurias joined
pmichaud PerlJam: why isn't it? 14:17
moritz_ rakudo: (1..10).map( {$^a + $^b} ).perl.say 14:18
p6eval rakudo 8bbc31: OUTPUT«[3, 7, 11, 15, 19]␤»
moritz_ that turns 10 things into 5 things
now identifier N with 10 and M with 5...
pmichaud rakudo: (1..5).map( { $^a, $^a*2 } ).perl.say
p6eval rakudo 8bbc31: OUTPUT«[1, 2, 2, 4, 3, 6, 4, 8, 5, 10]␤»
PerlJam sure, but it doesn't do the repeated function application using the M things as input to the next cycle like reduce does 14:19
14:21 bacek_ left, bacek_ joined
bacek_ yak... my ISP getting worth. 14:21
rakudo: multi sub foo($a) {...}; multi sub foo($a, *@list) {...}; foo(1) 14:22
p6eval rakudo 8bbc31: OUTPUT«Ambiguous dispatch to multi 'foo'. Ambiguous candidates had signatures:␤:(Any $a)␤:(Any $a, Any @list)␤␤current instr.: '_block14' pc 76 (EVAL_18:44)␤»
bacek_ Is it expected failure?
moritz_ summons jnthn 14:23
bacek_ thinks jnthn drinks gorilka in Ukrain :)
pmurias ruoso: re capture refactoring does it amount to changing .invocant to .postcircumfix:<[ ]>(0) and .positional($i) to .postcircumfix:<[ ]>($i+1)
ruoso .postcircumfix<[ ]> provides more semantics then .positional 14:24
.positional doesn't do any kind of flattening
.[] needs to
but basically
dalek kudo: 2d5f157 | pmichaud++ | src/classes/Code.pir:
Add a .count method to Code objects.
ruoso .invocant turns into .positional(0) 14:25
and .positional($i) turns into .positional($i+1)
pmurias ruoso: what sort of flattening to we want .[] to do?
bacek_ rakudo: sub Reduce(Code $c, $prev, *@list) { @list ?? Reduce($c, &$c($prev, |@list.splice(0,$c.count-1)), @list) !! $prev }; say Reduce { $^a+$^b }, 1..5
p6eval rakudo 8bbc31: OUTPUT«12345␤»
bacek_ rakudo: sub Reduce(Code $c, $prev, *@list) { +@list ?? Reduce($c, &$c($prev, |@list.splice(0,$c.count-1)), @list) !! $prev }; say Reduce { $^a+$^b }, 1..5
ruoso pmurias, say (1,(1,(1,(1,2,3),2,3),2,3),2,3)
p6eval rakudo 8bbc31: OUTPUT«12345␤»
bacek_ ouch
rakudo: sub Reduce(Code $c, $prev, *@list) { +@list ?? Reduce($c, &$c($prev, |@list.splice(0,$c.count-1)), @list) !! $prev }; say Reduce({ $^a+$^b }, 1..5) 14:26
p6eval rakudo 8bbc31: OUTPUT«12345␤»
ruoso pmurias, .[] implies list context, so implies flattening
I guess
bacek_ rakudo: sub Reduce(Code $c, $prev, *@list) { +@list ?? Reduce($c, &$c($prev, |@list.splice(0,$c.arity-1)), @list) !! $prev }; say Reduce({ $^a+$^b }, 1..5)
p6eval rakudo 8bbc31: OUTPUT«12345␤»
pmurias ruoso: is positional specced? 14:27
ruoso pmurias, capture is not specced yet
bacek_ rakudo: sub Reduce(Code $c, $prev, *@list is rw) { +@list ?? Reduce($c, &$c($prev, |@list.splice(0,$c.arity-1)), @list) !! $prev }; say Reduce({ $^a+$^b }, 1..5) 14:28
p6eval rakudo 8bbc31: OUTPUT«12345␤»
ruoso pmurias, but I'd assume it's ok to provide that
14:29 c9s___ joined
bacek_ rakudo: sub foo($a, *@b) { say $a }; foo(1..3) 14:31
p6eval rakudo 2d5f15: OUTPUT«123␤»
bacek_ looks like a bug...
Or I don't understand param passing at all.
pugs: sub foo($a, *@b) { say $a }; foo(1..3) 14:32
p6eval pugs: OUTPUT«123␤»
moritz_ bacek_: 1..3 is a Range object
which is bound to $a
bacek_ pugs: sub foo($a, *@b) { say $a }; foo(list(1..3))
p6eval pugs: OUTPUT«123␤»
bacek_ rakudo: sub foo($a, *@b) { say $a }; foo(list(1..3))
p6eval rakudo 2d5f15: OUTPUT«1 2 3␤»
moritz_ perl6: sub foo($a, *@b) { say $a }; foo(|1..3)
p6eval pugs: OUTPUT«123␤»
..rakudo 2d5f15: OUTPUT«argument doesn't array␤current instr.: 'infix:..' pc 8089 (src/classes/Range.pir:279)␤»
..elf 25757: OUTPUT«Undefined subroutine &GLOBAL::prefix__124 called at (eval 127) line 6.␤ at ./elf_h line 4377␤»
bacek_ rakudo: sub foo($a, *@b) { say $a }; foo(1,2,3) 14:33
p6eval rakudo 2d5f15: OUTPUT«1␤»
moritz_ perl6: sub foo($a, *@b) { say $a }; foo(|(1..3).list)
p6eval pugs: OUTPUT«123␤»
..rakudo 2d5f15: OUTPUT«1␤»
..elf 25757: OUTPUT«Undefined subroutine &GLOBAL::prefix__124 called at (eval 128) line 6.␤ at ./elf_h line 4377␤»
moritz_ I agree with rakudo here
bacek_ rakudo: sub TheReduce(Code $c, $prev, *@list is rw) { +@list ?? TheReduce($c, &$c($prev, |@list.splice(0,$c.arity-1)), @list) !! $prev }; sub Reduce(Code $c, $a) { TheReduce($c, |$a.list) }; }; say Reduce({ $^a+$^b }, 1..5) 14:35
p6eval rakudo 2d5f15: OUTPUT«Syntax error at line 1, near "}; say Red"␤␤current instr.: 'parrot;PGE;Util;die' pc 129 (runtime/parrot/library/PGE/Util.pir:83)␤»
bacek_ rakudo: sub TheReduce(Code $c, $prev, *@list is rw) { +@list ?? TheReduce($c, &$c($prev, |@list.splice(0,$c.arity-1)), @list) !! $prev }; sub Reduce(Code $c, $a) { TheReduce($c, |$a.list) }; say Reduce({ $^a+$^b }, 1..5)
p6eval rakudo 2d5f15: OUTPUT«15␤»
bacek_ hooray. Single line reduce implementation :)
rakudo: sub TheReduce(Code $c, $prev, *@list is rw) { +@list ?? TheReduce($c, &$c($prev, |@list.splice(0,$c.arity-1)), @list) !! $prev }; sub Reduce(Code $c, $a) { TheReduce($c, |$a.list) }; say Reduce({ $^a+$^b +$^c}, 1..5); 14:36
p6eval rakudo 2d5f15: OUTPUT«15␤»
bacek_ rakudo: sub TheReduce(Code $c, $prev, *@list is rw) { +@list ?? TheReduce($c, &$c($prev, |@list.splice(0,$c.arity-1)), @list) !! $prev }; sub Reduce(Code $c, $a) { TheReduce($c, |$a.list) }; say Reduce({ $^a+$^b * $^c}, 1..8);
p6eval rakudo 2d5f15: OUTPUT«too few arguments passed (2) - 3 params expected␤current instr.: '_block71' pc 557 (EVAL_19:181)␤»
bacek_ And even works as yet-to-be-specced :) 14:37
rakudo: sub TheReduce(Code $c, $prev, *@list is rw) { +@list ?? TheReduce($c, &$c($prev, |@list.splice(0,$c.arity-1)), @list) !! $prev }; sub Reduce(Code $c, $a) { TheReduce($c, |$a.list) }; say Reduce( -> $a, $b, $c? { $a+$b * ($c//1)}, 1..8);
pmurias an
p6eval rakudo 2d5f15: OUTPUT«36␤»
bacek_ rakudo: sub TheReduce(Code $c, $prev, *@list is rw) { @list ?? TheReduce($c, &$c($prev, |@list.splice(0,$c.count-1)), @list) !! $prev }; sub Reduce(Code $c, $a) { TheReduce($c, |$a.list) }; say Reduce( -> $a, $b, $c? { $a+$b * ($c//1)}, 1..8); 14:39
p6eval rakudo 2d5f15: OUTPUT«77␤»
bacek_ Is .tailcall "cheap" in parrot?
14:42 c9s__ left
pmichaud regardless of whether it's cheap, I'm not a fan of a recursive reduce implementation. 14:45
14:45 rodi joined
bacek_ It's just for fun :) 14:46
14:47 c9s__ joined 14:49 pmurias left
dalek kudo: 065175b | pmichaud++ | docs/spectest-progress.csv:
spectest-progress.csv update: 317 files, 7124 passing, 0 failing
 14:53
14:55 c9s__ left 14:56 c9s___ left
bacek_ pmichaud: is this nopaste.snit.ch/15830 better? 14:57
pmichaud bacek_: no, because it's modifying the invocant. 14:59
14:59 c9s___ joined
bacek_ pmichaud: otherwise it will require .list. Will it lead to eager flattening? 15:00
pmichaud .list isn't eager. 15:01
moritz_ rakudo: my $s = sub ($x, $y!, $z) { ... }; say $s.count
p6eval rakudo 065175: OUTPUT«3␤»
bacek_ pmichaud: ah. ok
pmichaud rakudo: ( -> $a, $b? $c? { ... } ).count.say
p6eval rakudo 065175: OUTPUT«Syntax error at line 1, near "( -> $a, $"␤␤current instr.: 'parrot;PGE;Util;die' pc 129 (runtime/parrot/library/PGE/Util.pir:83)␤» 15:02
pmichaud rakudo: ( -> $a, $b?, $c? { ... } ).count.say
p6eval rakudo 065175: OUTPUT«3␤»
pmichaud rakudo: ( -> $a, $b?, $c? { ... } ).arity.say
p6eval rakudo 065175: OUTPUT«1␤»
15:03 frioux joined, PZt joined
masak rakudo: -> $a, $b?, $c? { ... }.count.say 15:03
p6eval rakudo 065175: OUTPUT«3␤»
masak parens redundant.
pmichaud when in doubt, add parens. :-)
masak I guess I'm simply not in doubt, then. :) 15:04
pmichaud perhaps. I always vote for what's clearer to read.
15:04 eternaleye left
masak that's a bit subjective, though. 15:04
pmichaud having a term with spaces receive a .method just seems odd
pugs_svn r25758 | bacek++ | [t/spec] Added test for List.reduce with optional args. 15:05
15:05 bacek_ is now known as bacek
pugs_svn r25759 | moritz++ | [t/spec] tests for .count on subs 15:05
bacek pmichaud: and this one? nopaste.snit.ch/15831
masak pmichaud: if you say so. I don't find that harder to read than terms without spaces. but nevermind.
moritz_ pmichaud: two of the new tests fail, and I have no idea why
did I mess up something fairly obvious? 15:06
pmichaud moritz_: in r25759?
moritz_ yes
pmichaud looking.
I don't know if named params count. 15:07
i.e., I think .count might be only for positionals.
moritz_ might be 15:08
ie we need specs
pmichaud clearly: sub foo($a, :$b) { ... }; foo(1,2); #fail
moritz_ right
pmichaud in the example I just gave, foo can take only one positional argument.
so I'd think .count == 1 15:09
moritz_ rakudo: (-> $a, :$b! { ... }).arity.say
p6eval rakudo 065175: OUTPUT«2␤»
moritz_ we should be consistent
pmichaud I think that's the parrot named argument bug there.
Parrot has a (serious) bug in that it treats named arguments as positionals.
and there's not a way to turn that off yet. 15:10
bacek rakudo: +(-> $a, :$b! { ... }).signature.params
p6eval rakudo 065175: RESULT«2»
bacek rakudo: +(-> $a, :$b! { ... }).count
pmichaud I might be able to get signature.arity and signature.count to dtrt, though.
p6eval rakudo 065175: RESULT«1»
pmichaud it's tough to rely on .signature because non-rakudo subs don't have a .signature. 15:11
masak why not?
pmichaud because Parrot subs don't have signatures?
masak hm. 15:12
bacek Let's rewrite everything in Perl6 using Q:PIR!
masak fair point.
bacek hides
15:12 nihiliad joined
pmichaud bacek: with my @list = $values.list; we're still creating an Array. I'd prefer to see my $list = $values.list; or my $list := $values.list; to avoid that. 15:13
Also, why do people have an aversion to 'self' ?
I think that self.list reads far more clearly than $values.list 15:14
bacek Because it's in spec. I prefer self.list too 15:15
pmichaud The S29/S32 spec should be considered "provisional" in many respects.
even if $values is given in S32, self.list is still valid in the body of the code. 15:16
afk for a bit. 15:18
15:19 PacoLinux joined, NoirSoldats left
masak doesn't have an aversion to 'self', he just forgets to write it often 15:21
moritz_ you can even write @.list
masak I know.
the problem is not that I don't like to write 'self'. 15:22
the problem is that I forget I need to.
pugs_svn r25760 | bacek++ | [t/spec] Put explicit undef as last argument of n-ary List.reduce 15:23
moritz_ rakudo: class Any is also { method listsay { .say for @.list } }; <a b c>.listsay; 3.listsay
p6eval rakudo 065175: OUTPUT«a␤b␤c␤3␤»
moritz_ woot
it even works :-)
that could simplify some built-in methods 15:24
bacek pmichaud: nopaste.snit.ch/15832 this is latest attempt.
afk & sleep.
2:30 here.
15:28 NoirSoldats joined 15:33 NoirSoldats left 15:37 frioux left 15:38 frioux joined, NoirSoldats joined 15:39 mikehh joined 15:45 ingy joined
dalek kudo: d16d3c3 | pmichaud++ | src/classes/Code.pir:
Temporary fix to get .signature to work on ObjectRefs (fixes RT #63744).

dereference objects prior to dispatch.		 15:45
15:51 justatheory joined, Tene_ joined 15:59 [particle] joined, pmurias joined
diakopter pmurias: hi 15:59
TimToady do we need a primitive way to match a Signature against an Iterator? which presumably iterates the iterator only if the signature matches, and to the extent it matches? 16:00
and that explicitly says its okay to leave more values in the Iterator?
*it's 16:01
diakopter "why not?"
16:01 moritz_ sets mode: +o TimToady
moritz_ TimToady: that was the easy part. now you have to come up with a syntax :-) 16:01
TimToady that's not the question you're supposed to ask when you're designing
16:01 Tene left
TimToady or not the only question... 16:01
you also need a good answer for "why?" 16:02
but in this case it seems as though people are struggling to invent these semantics in terms of current syntax
moritz_ because that's what map, grep, reduce and similar methods all need
TimToady so perhaps there's something missing
diakopter i know; kidding. :D a method on Iterator or a method on Signature? or a sub? 16:03
TimToady well, Iterator.nibble(Signature) is probably good enough
moritz_ what does 'nibble' mean?
diakopter don't bite off more than you can chew?
moritz_ chop off, byte by byte?
rodi "look at the next one, bite only if it matches" 16:04
TimToady well, I guess the question is whether .ACCEPTS should be doing it
but it seems like the default match of a signature should want to match everything
rodi moritz_: I think Knuth calls what you described a "nybble" :-P 16:05
TimToady we could have an explicit marker in a signature that says "okay to have more args here", but then we'd have to modify the user's sig in map et al.
nybble is half a byte
ruoso TimToady, once an iterator is consumed, there's no turning back 16:07
it's the role of a "lazy list" to keep the consumed values stored
TimToady indeed, but that argues for a primitive that can try to match without consuming until "commit" point
ruoso what do you mean by "commit" point/ 16:08
?
16:08 NoirSoldats left
TimToady the point at which the signature is happy with the match 16:08
it's a transaction that either succeeds or fails
diakopter peek
ruoso well, there only way a signature can match an iterator without consuming it
TimToady or pushback
ruoso is for it to be *@items
TimToady *@items should consume it 16:09
ruoso no, it binds lazily as well
TimToady well, semantically it consumes it
ruoso ok, we're talking about different things then
let's take a practical scenario
TimToady the iterator should report "done" after handing off the rest to @items
16:09 [particle]2 joined, NoirSoldats joined
ruoso let's assume our iterator is reading from a pipe and splitting using IFS/IRS 16:10
pugs_svn r25761 | dankogai++ | updated to Cabal-1.6.0.2
masak TimToady: do I need to do 'augment' when re-opening modules?
ruoso either you created a lazy list from that iterator
which will store the consumed items
TimToady masak: yes, I think
masak ok. 16:11
16:11 M_o_C joined
ruoso but if you don't have such a lazy list, the item you consumed is not stored anywhere else 16:11
and a pipe doesn't accept undo
masak TimToady: S11:365: use Dog:ver(Any):auth({ .substr(0,5) eq 'cpan:'}) -- why 'Any' and not '*' or 'Whatever'?
TimToady fossil 16:12
pugs_svn r25762 | dankogai++ | updated to HTTP-4000.0.4
TimToady that was written before * existed
masak ah.
ruoso TimToady, I think I'm looking at a barrier you're not 16:13
TimToady, I see a barrier between the iterator itself and the lazy list that is being used as a proxy
TimToady well, I'm not picky, if matching Signature against a lazy list is good enough for map etc. 16:14
but somewhere we need that transactional support
ruoso TimToady, pipes are not transactional
TimToady and the bit that says it's okay to leave more stuff
pugs_svn r25763 | dankogai++ | updated to zlib-0.5.0.0
TimToady but pipe handles can be, if they support pushback
diakopter TimToady: so, it treats it as immutable until it matches, then it actually consumes all those up to the commit point? or just the ones that matched? 16:15
ruoso then it's a lazy list that talks to that iterator, that's what I mean
or a lazy scalar of some sort
that happens to also be an Iterator
16:17 NoirSoldats left
TimToady I don't care what does it, but we have to support partial signature matching suffiently well to know whether we ran out of arguments in mid-signature, or ran out of signature in mid arguments, or they came out even 16:17
and in each case, retain enough control to Do The Right Thing for whatever the current operator is, map, reduce, whatever 16:18
ruoso for that matter, lazy lists seems to suffice
TimToady List ~~ Signature should fail on partial match by default, it seems like 16:19
masak rakudo: say "abcdefgh".substr(*-2) # would it be unreasonable to make this work like the .[*-2] case? 16:20
p6eval rakudo d16d3c: OUTPUT«Use of uninitialized value␤gh␤»
pugs_svn r25764 | dankogai++ | updated to cabal-install-0.6.2
ruoso TimToady, a partial Signature match is not a match... is it?
TimToady so we need some API to say not to do the default of "match all"
ruoso isn't that declaring arguments as optional? 16:21
TimToady Capture ~~ Signature is certainly "match all" semantics
ruoso ahhhh
I see what you mean
moritz_ maybe with | on one side?
TimToady no, it's declaring the rest of the list as optional
I don't think we want to modify the Signature to do it
pmichaud agreed, modifying the signature seems wrongish. 16:22
ruoso I always assumed the operators that did this were fetching n args from the list
TimToady but it's also declaring it as transactional, so it doesn't eat any of the list if the match fails
ruoso depending on the arity of the code
TimToady see overnight discussion above
nobody is having good success doing the transaction right in terms of current primitives 16:23
nobody is thinking about the rollback part of it
pmichaud correct. 16:24
agreed.
16:24 [particle] left
TimToady if signature matching is to be useful in the general case, we have to handle this correctly, or it would be like regex that randomly put the pos in the wrong place on backtracking 16:25
moritz_ maybe we could have a @(signature) form that creates an object that slices a list into captures?
ruoso honestly, I've been neglecting transactions
pmichaud feels like it ought to be a method like 'map'; indeed, it's probably the foundation for map/reduce/etc.
16:25 ejs joined
moritz_ I don't think it's easy to build reduce on top of map 16:25
16:25 rkendall left
pmichaud moritz_: I agree, which is why I think there's something else that map and reduce and grep are built on top of. 16:26
TimToady well think about it guys, I have to go get ready for $work
ruoso hmmm
maybe Iterator itself provides a method to match a signature
ok... this is where TimToady started few minutes ago 16:27
ruoso looping
pmichaud perhaps we have a 'map'-ish method on Iterator
ruoso pmichaud, the thing is that we shouldn't require iterators to be transational 16:29
diakopter iterators (that do work that can't be rolled back (to generate/pump each item)) need to cache the results in case the signature gives up, and iterators whose (possibly lazy) feeds are state-dependent shouldn't do that at all
pmichaud the method could be transactional without the iterator having to be.
16:30 pmurias left
ruoso ok... lazy lists do fullfit that part 16:30
ruoso needs to read the overnight discussino
16:32 masak left
ruoso can't find it... any link? 16:33
pmichaud it's primarily in scrollback on #perl6
the stuff about implementing .reduce in Perl 6
moritz_ maybe TimToady also meant the p6l thread?
pmichaud possibly, but there's not a lot of detail there other than how map/reduce should handle argument lists that aren't multiples of some function of the arity 16:34
at any rate, the p6l thread is groups.google.com/group/perl.perl6....581cb8940#
ruoso hmmm... that doesn't seem to explain why each implementation of such features to use the arity of the code to fetch elements from the input 16:36
er... why that would be a bad idea, i mean 16:37
diakopter to support this, each Iterator needs to know whether its generating function 1) supports rollback 2) supports redo (if anything the function invokes is not deterministic) and 3) is pure (stronger version of supporting redo)
ruoso but I think using it as a lazy list seems very much ok 16:38
because then the iterator doesn't need to handle redo 16:39
nor anything else
the lazy list consumes the iterator, but the values won't go away
but I think we have other question here 16:40
which is how to say to the binding process
"try to bind as many as possible, returning what's left"
diakopter but if you're trying a succession of signatures... 16:42
hm
ruoso using lazy lists
so, if that doesn't suceed, you simply use the one you sent before
the real question then is... 16:43
is the Signature responsible for the binding?
hmm
actually 16:44
Signature.ACCEPTS might set $?
sorry
$/
16:46 mikehh left
ruoso then, if successfull, the match object would say how many items it matched 16:47
16:49 mikehh joined
ruoso and in fact, you could even call &code.(|$/) 16:53
16:54 bacek left 16:57 Exodist joined 17:00 namenlos left 17:01 namenlos joined, krunen left 17:02 bacek joined 17:06 namenlos left 17:07 PacoLinux left, namenlos joined 17:14 finanalyst joined, jferrero joined
pugs_svn r25765 | dankogai++ | updated to Cabal-1.6.0.2 17:15
17:16 c9s___ left 17:18 cspencer joined 17:24 pmurias joined
pmurias diakopter: hi 17:25
cspencer moritz_: ping 17:26
lambdabot cspencer: You have 1 new message. '/msg lambdabot @messages' to read it.
moritz_ cspencer: pong 17:27
cspencer moritz_: looks like you beat me to the map/pairs fixes 17:28
moritz_ cspencer: I did indeed
cspencer i've got a question re: the implementation though
moritz_ shoot 17:29
cspencer if i remember correctly, pmichaud was concerned about the laziness of lists being affected by various things in the old implementation
ie) expanding the list out in full
moritz_ right 17:30
cspencer in the pairs method, in the case that @indices was defined, if we did something like "for (@indices Z self.list[|@indices]) ...." instead of iterating over the whole list, is that a bad thing? 17:31
ie) will that cause the list to be fully flattened out?
moritz_ that wasn't the concern in the .pairs method 17:32
cspencer ah ok
moritz_ the concern was in the .map method
cspencer oh, that was with map
right
moritz_ where you used @array.end
cspencer i'm getting my concerns muddled here :)
moritz_ that's... concerning ;-)
cspencer it is indeed ;)
pugs_svn r25766 | dankogai++ | updated to Cabal-1.6.0.2 17:33
moritz_ anyway, i'm updating the self.list things to @.list 17:34
it's the same, just a bit more ideomatic
pmichaud looks nicer, too.
cspencer alrighty
what does the "@." do? 17:35
does it access self in a list context?
pmichaud $.list calls self.list in scalar context
@.list calls self.list in list context
lambdabot No module "calls self.list in list context" loaded
moritz_ lambdabot: shut up 17:36
17:36 moritz_ sets mode: +v lambdabot
cspencer ok :) 17:36
pmichaud afk, lunch 17:38
mikehh built rakudo on parrot r37239 linux AMD64 spectests pass - t/spec/S32-list/reduce.rakudo TODO passed: 14 17:48
moritz_ same here 17:50
17:51 PacoLinux joined
mikehh make spectest took 37 min 17:52
moritz_ I compile parrot with --optimize, that makes it a bit faster
mikehh I also use --optimize - took me 35 min on linux i386 17:54
pasteling "ruoso" at 201.45.49.162 pasted "role ideas for the web.pm framework..." (156 lines, 3.9K) at sial.org/pbot/35421
mikehh back in the days we had threading working it used to take just over 10 minutes 17:57
17:58 AndyAway is now known as alester
moritz_ takes 17min here for one pass... you need to buy a faster computer ;-) 18:10
18:10 jferrero left 18:11 araujo joined
mikehh more memory would probably help too 18:12
moritz_ does it swap?
bbl
18:14 ujwalic left
mikehh not that I noticed - its an old Pentium D dual core with 2GB - I am trying for a quad core with a minimum of 4GB 18:14
dalek kudo: 960d251 | (Moritz Lenz)++ | src/setting/Any-list.pm:
[setting] replace self.list with @.list in some places
 18:16
mikehh my son has one my DIL just got one - I am waiting 'cos I only run Linux 18:17
18:18 pmurias left
ruoso can I do a ~~ inside a closure inside a regex and get it as a part of the match? 18:23
/ bla { $foo ~~ $bar } ble /
I see it will execute
and if the block returns false, I think it will stop processing
but... will it take the $/ set inside the block?
basically 18:24
I have a Pattern
how do I use it inside a regex so that it integrates with the matching
(i.e., $/.to, $/.from etc 18:25
TimToady why isn't it just / bla $bar ble /? 18:28
if $bar is a Regex, it should dwym
bare {...} pays no attention to the bool return value, btw
it's used only for side effects
<?{...}> runs an assertion 18:29
18:31 c9s__ joined, schmalbe joined 18:32 Psyche^ joined
TimToady but STD does things like $¢ = $*PARSER.bless($¢); to change the current cursor 18:33
which is not exactly single assignment... 18:34
18:35 justatheory left, justatheory joined 18:39 mikehh left, mikehh joined
ruoso TimToady, consider I start to use Match objects to match in other places then regexes 18:41
but then I want to mix using those (usually Patterns) with regex 18:42
that would allow me to write something that would take part in matching, but is written as a simple ACCEPTS method 18:43
that sets $/ and returns bool
TimToady I almost never want to deal with $/ in STD; it's usually better to deal with cursors, because $/ is one level too "global", and $/ doesn't really understand everything a cursor needs to know, like what the current language is 18:45
ruoso right... so, any idea on how to solve my problem?
TimToady in theory positions are opaque, so you need something typologically correct to translate the submatch's idea of position into the surrounding Cursor's notion of position 18:47
that being said, STD just uses codepoint positions :) 18:48
because it's really just the p5 subset of p6 that STD is written in, more or less
at least, the parts that don't end up being emulated by Cursor.pmc 18:49
18:49 Patterner left, Psyche^ is now known as Patterner
ruoso so, having a way to call a closure that should leave $/ set on sucess is not that bad, is it? 18:49
or at leasy 18:50
18:50 dusty joined
ruoso *least 18:50
to use a Pattern object as part of the match
where it should set CALLER::<$/>
TimToady then your closure needs to extract the positional information from that and generate a new $¢ from the old one 18:51
ruoso I mean having a syntax to hide that
TimToady (btw, $/ turns into $M in STD.pmc) 18:52
I don't think it's worth it witout more use cases to look at
ruoso something{ $_ starts at pos, and closure sets $/ }
or something<$pattern-object> 18:53
that would call ACCEPTS on $pattern-object sending $_ starting on current position
TimToady I'm not saying it won't be worth it eventually, but we need to see the actual usage patterns to know what the best sugar would be, or whether it occurs huffmanly often enough to deserve sugar
18:53 armagad left
ruoso TimToady, my use case is about matching a string against an arbitrary data structure... 18:54
in this case
TimToady I should probably get rid of $¢ = $*PARSER.bless($¢); though in any case
ruoso a url
against a linked list of objects
18:55 dusty left
TimToady right now you can just say $¢ = $¢.cursor($newpos) to set the new position, in STDese 18:56
ruoso ok.. but that's not really Perl 6... 18:57
TimToady yes it is
$¢ is specced somewhere
ruoso but ideally it would create an entire submatch, wouldn't it? 18:59
how could it do that?
TimToady S05:800 19:00
well, same way STD does, presumably :)
it's all just method calls...
we can argue about the names of the method calls, of course 19:01
but Cursor should be considered somewhat canonical in the same way that STD itself is
19:03 nihiliad left
TimToady Cursor uses variants of .SUBSUME to attach a submatch 19:04
ruoso hmmm...
I think I just had a better idea for the specific problem I'm dealing with
which is to pre-build a regex representing the entire data structure 19:05
TimToady depends on which kind of lazy you're aiming at :)
if you can cache it, it's probably worth prebuilding
ruoso yeah... the specific case of URL ~~ ChainedAction seems fine to prebuild 19:06
TimToady also depends on the pass/fail statistics
ruoso but there are cases where matching against a data structure will be indeed interesting
TimToady "we'll burn that bridge when we get to it", as they say...
ruoso does / ^ $.foo / mean " the value returned by $.foo at the start of the string? or do ^ mean something else now 19:08
?
or is that just plain wrong? 19:09
19:10 nihiliad joined
TimToady $.foo is a perfectly fine variabl name, as far as the regex is concerned 19:10
19:10 c9s__ left
TimToady so it's just like / ^ $foo / 19:10
19:10 c9s__ joined
TimToady that is, match whatever $foo matches at the start of the string 19:10
use ^^ for start of line 19:11
$ for end of string, $$ for end of line
19:11 japhb left
ruoso right 19:11
TimToady std: / ^ $.foo / 19:12
p6eval std 25766: OUTPUT«ok 00:02 34m␤»
ruoso TimToady, so, if $regex contains a capture, will it be in a submatch of / foo $bar baz / ? or does it look like "in-line"? 19:16
TimToady I don't understand your question at all 19:17
did you mean if $bar contains a capture
if $bar is a string it's just matched inline like a literal
ruoso $bar = / (positional) $<named> := bar /; 19:18
"foo positional bar baz" ~~ / foo $bar baz /
TimToady that's matched inline unless you do $<newname> := $bar 19:19
ruoso right...
thankfully, that was what I expected ;)
TimToady or <newname=$bar>
should generally use the <$bar> form if you expect a regex rather than a string, just for documentation 19:20
but it's still anonymous because of the $ in front
same way that <.foo> makes <foo> anonymous, really 19:21
ruoso cool 19:22
so if I "return $url ~~ $regex" in the end, it will return all the captured values
looks just fine
TimToady also ($bar) will capture submatches in $0, though STD makes little use of this 19:26
ruoso so then I use the match as argument to the methods in the (|$/) form 19:28
pasteling "ruoso" at 201.45.49.162 pasted "updated role ideas for the web.pm framework..." (163 lines, 4.1K) at sial.org/pbot/35424 19:32
ruoso TimToady, if you look at the code I pasted, you'll see that I build a lot of regexes, and then use a for loop to try to match each option... 19:36
presumably, if this was all set into a single grammar
an optimizer would be able to reuse match parts (considering the start is very much similar for a lot of regexes) 19:37
is there a way to build in runtime a regex that would do that as a single match?
19:39 M_o_C left
ruoso considering I re-use the same regex object all over the place, it should be fairly optimizable 19:40
19:43 c9s__ left
skids ruoso: also that might be useful to abuse regex as a trie where you add a rule into a regex for every value you want to add to it, and occasionally somehow ask for the regex to be recompiled. 19:43
That kinda inverts code and data though. 19:44
ruoso skids, I would actually assume that when you do /foo $bar baz/ it would be free to optimize as much as it wans 19:45
so if I'm reusing $bar in a lot of regexes 19:46
it should be able to optimize it
19:46 japhb joined
ruoso TimToady, in that code, I'd like to have a token in the end that would look like / [ $regex1 { make $action1 } | $regex2 { make $action2 } | ... ]/ 19:48
but that can only be composed programatically,
skids ruoso: Oh I guess we're not really talking about the same thing. I thought you were asking about collapsing dynamically built up regexe.
ruoso We are
I'm just assuming it is free to collapse it as it pleases 19:49
pugs_svn r25767 | pmichaud++ | [spec]: Update S05: remove <word>, add <ww>, revise <wb>, <ws>, <space>, and <blank>. 19:50
19:55 NoirSoldats joined 19:57 c9s__ joined 20:06 icwiener joined 20:07 c9s__ left
dalek kudo: d8a1926 | pmichaud++ | docs/guide_to_setting.pod:
Some refinements to guide_to_setting.pod .
 20:11
ruoso TimToady, if <$bar> in-lines a Regex object... could <@bar> inline several? 20:12
alester Itr drives me nuts that you have to check for $hash{key}'s existence in PHP. 20:16
I'm going to cry from overtyping.
pmichaud ...check for $hash{key}'s existence? 20:17
ruoso TimToady, while <@bar> would mean the same as [<$bar[0]>|<$bar[1]>|<$bar[2]>]
pmichaud ruoso: <@bar>: each element is treated as a subrule
ruoso with OR or AND? 20:18
pmichaud OR -- same as @bar without the <...>
ruoso cool
pmichaud S05:1283
ruoso it's fun when things works according to your expectations...
pmichaud but they aren't "inlined" -- they're subrule calls.
ruoso Ok, it doesn't change much, since each of them is going to have a { make ... } 20:19
pmichaud in particular, by default they're non-capturing subrule calls.
so if they have a { make ... } then the results need to be captured somewhere. 20:20
20:21 NoirSoldats left
ruoso hmm 20:22
but if I do $<foo> := <@bar>
does it capture ir?
pmichaud yes, it should.
I suspect that <foo=@bar> would also work. 20:23
ruoso hmm
now consider I want both the value returned by "make" 20:24
and the sub-match for each item of @bar
pmichaud it's an alternation, so only one item of @bar matches.
ruoso right... I meant that one, actually... 20:25
pmichaud the thing that <@bar> matched would be $<foo>
the value returned by make would be $<foo><?>
ruoso right
while the submatch would still be $<foo> 20:27
pmichaud correct.
ruoso so... my $regexN = / something { make 'bla' } /; my @regexes = ($regexN,...); my $rx = / $<foo> := <@regexes>/; if $foo ~~ $rx { say $<foo>; say $<foo><?> } 20:29
pmichaud keeping in mind that say $<foo> and say $<foo><?> will produce the same result. 20:30
20:31 NoirSoldats joined
ruoso right... but if "something" actually was a match that had another captures 20:31
would I still be able to access that captures?
pmichaud sure, you can access the captures through $<foo> (more)
but stringifying $<foo> will give you its stringified result object by default. 20:32
ruoso ok
pmichaud if you want the text of whatever matched, use $<foo>.text
20:32 schmalbe left
pmichaud it's all described at S05:2373 20:35
20:37 NoirSoldats left 20:38 cspencer left
pasteling "ruoso" at 201.45.49.162 pasted "yet another version of the ideas for the web.pm framework" (168 lines, 4.2K) at sial.org/pbot/35425 20:40
ruoso pmichaud, Dispatcher.compile in the pasted code above is the code that tries to do that... would you mind taking a look to see if it's ok?
pmichaud looking 20:41
depends on what you mean by "freezes the regexes"
ruoso it means it will, only once, generate the regex to match URL x ChainedAction 20:42
which is used in the "dispatch" method, just below
pmichaud but yes, since the $.regex is effectively a closure, it should continue to refer to whatever values were in @subregexes at the time it was captured. 20:43
in particular, if I have
20:43 mberends joined
pmichaud my @a = ...; my $regex = / <@a> /; @a[1] = ...; 'xyz' ~~ $regex 20:43
then the match will use whatever value of @a[1] is in the second assignment
ruoso right... but since any other reference to that value is lost... it ends up "almost" read-only 20:44
pmichaud correct.
20:45 Tene_ is now known as Tene
pmichaud I wonder if |$<action> will also include '?' as a named parameter. 20:45
ruoso is there a way to test it? 20:46
pmichaud not at present... but naively I would expect %($<action>) to include all of the elements in its hash interface, which would include <?> 20:47
ruoso so it'd be better if I add an inner capture to avoid it
pmichaud or we might want some confirmation/clarification from TimToady
ruoso / $<foo> := [ foo bar <baz ] / is valid, right? 20:49
oops
*<baz> that is
pmichaud yes.
you end up with both $<foo> and $<baz>
ruoso I mean... I can bind a non-capturing group to a capture variable
20:49 Kimtaro_ joined, Kimtaro_ left
pmichaud $<foo> := [ foo bar <baz> ] means that $<foo> will capture the text inside of the brackets 20:49
ruoso but not the submatch?
pmichaud the sub match would be $<baz> 20:50
if you want the submatch to be inside of $<foo>, it's
ruoso right... so I should use ( ) then
pmichaud $<foo> := ..... right, what you wrote.
with ( ) it's $<foo><baz>
I have to go pick up kids from school... bbi30
pasteling "ruoso" at 201.45.49.162 pasted "even yet another version of the ideas for the web.pm framework" (167 lines, 4.3K) at sial.org/pbot/35426
pmichaud your basic approach seems workable, at any rate. 20:51
ruoso pmichaud, thanks for the input
21:02 meppl joined 21:05 nihiliad left 21:07 Kimtaro left, araujo left, NoirSoldats joined 21:10 araujo joined
ruoso is there a predefined order in the invocation of $foo.*method ? 21:19
or is it up to the HOW?
in the sense of "as returned by $foo.can
21:20 nihiliad joined
PerlJam See perlcabal.org/syn/S12.html#Calling_...of_methods 21:22
21:26 NoirSoldats left 21:28 namenlos_ joined, namenlos left
ruoso TimToady, btw... I'm already considering ¢foo as the way to store captures without enforcing any context... 21:31
21:36 NoirSoldats joined
moritz_ pmichaud: "Using them often makes the code less clear, and can sometimes be correct" - did you mean "incorrect"? 21:39
pugs_svn r25768 | wayland++ | Added real/imaginary to Complex, and true/not to Object 21:43
moritz_ pmichaud: I'm too lazy to wait for your answer ;-) Just pushed it to github, if you disagree, please revert 21:44
dalek kudo: 95ce390 | (Moritz Lenz)++ | docs/guide_to_setting.pod:
[docs] thinko in guide_to_setting.pod
 21:45
moritz_ wayland76: any objections to calling them .re and .im?
wayland76: they occur quite frequently in Complex code to deserve a better huffmanization, IMHO
PerlJam moritz_: why not just .r and .i then? 21:46
moritz_ PerlJam: the re() and im() are what's used in mathematics
s/the/because/ 21:47
and .i means something different already
rakudo: say 4.i
p6eval rakudo 95ce39: OUTPUT«Method 'i' not found for invocant of class 'Int'␤current instr.: 'parrot;P6metaclass;dispatch' pc 320 (src/classes/ClassHOW.pir:161)␤»
moritz_ anywa, that should be 4i
PerlJam 4i != 4.i :-) 21:48
21:48 Kimtaro joined
moritz_ 41 == 4.i 21:48
if you believe the spec
S02-bits.pod
898:The complex type defaults to NaN + NaN.i. A buf type of known size
pugs_svn r25769 | wayland++ | Changed real to re, and imaginary to im, as perl moritz_++ suggestion 21:49
wayland76 s/perl/per/ :)
moritz_ I make that typo as well, once in a while 21:50
wayland76 finger-o :)
PerlJam I often do that too.
moritz_ especially at 11pm
PerlJam usually it's with the word "perlson" :)
wayland76 Well, it's 8:47am, so I don't have that excuse
Ooh, I like "perlson" :)
moritz_ "before coffee" counts as excuse too ;-) 21:51
wayland76 (although it does make me wonder whether we shouldn't also have perldaughters :) )
My whole life is "before coffee" :)
moritz_ they are too rare ;-)
wayland76: there you go. Perlfect (sic) excuse ;-) 21:52
wayland76 :)
21:52 mberends left 21:53 jferrero joined
pugs_svn r25770 | lwall++ | .i should be \i 21:54
r25770 | lwall++ | .text should be .Str
wayland76 @tell masak in regards to CPAN6, see cpan6.org/ and S22 :) 21:57
lambdabot Consider it noted.
pmichaud wayland76: I don't think true/not can be 'is export'
moritz_: you're correct, I meant "incorrect". Thank you for correctly correcting me.
22:01 wknight8111 joined
pmichaud TimToady: I'm a little confused about Match.text and Match.Str 22:01
if Match.Str returns the substring of the original target, then when/why/how does stringification of a Match object return the stringificationof the result object? 22:02
pugs_svn r25771 | wayland++ | Removed "is export" from true/not, as per pmichaud++ s suggestion
moritz_ $<?>.Str
pmichaud moritz_: that's not what the spec currently says in S05, though.
S05:2414: In string context it evaluates to the stringified value of its 22:03
I<result object>, which is usually the entire matched string:
(where "it" == "Match object" in this sentence)
so, a Match object in string context should return the stringified value of its result object.
22:04 alester left
pmichaud But if .Str is defined to return the original matched text.... there seems to be a conflict there. 22:04
22:06 ejs left, ejs joined 22:09 donaldh joined 22:18 [particle]2 left
pugs_svn r25772 | lwall++ | clarify semantics of invocant type checking on exported methods 22:19
22:21 icwiener left
pugs_svn r25773 | wayland++ | Updated in line with Larry's changes to S05 22:24
TimToady I don't think ~ and + should be using the result object anymore 22:25
pmichaud Okay.
TimToady can always say ~$<?> if that's what you mean
pmichaud so, they always stringify/numify the matched text?
TimToady yes
pmichaud okay.
In many ways that's very very helpful to PCT.
(changing it to not stringify the result objet)
*object 22:26
TimToady objét d'arte
pmichaud Another question that arose earlier: would %() include an entry for '?' ?
(if a result object was set)
TimToady I'm of two minds on that
pmichaud okay, it's a Junction then. :-) 22:27
TimToady it's really hard to write a parameter named ?
pmichaud Agreed.
TimToady so maybe only if *% is used
well, that's not quite what you were asking 22:28
I think it's definitely returned by %()
pmichaud Okay.
and methods still have an implicit slurpy hash, yes?
TimToady yes
pmichaud Okay.
I'm fine with that.
While looking through the specs on <?> I noticed an inadvertent pun on the <?> metasyntax operator. 22:29
Or perhaps it's not inadvertent?
TimToady thought about that, decided it probably was neither pos nor neg
pmichaud Excellent. answers++ 22:30
TimToady++
TimToady and probably doesn't interfere unless someone tries to match $<foo><?>
which seems silly anyway
pmichaud I've gotta update a lot of PCT docs and add some deprecation notices to Parrot. :-| 22:32
pugs_svn r25774 | lwall++ | ~ and + now just stringify/numify the matched string, not the result object 22:34
22:35 PZt left 22:36 PZt joined 22:40 nihiliad left
pugs_svn r25775 | lwall++ | more result object demotion 22:46
r25776 | moritz++ | [t/spec] test for Complex.re and .im 22:47
pmichaud I'm curious if .re and .im are only defined for Complex, or are they defined for Any ? 22:48
moritz_ IMHO they are sufficiently domain specific to go into Complex 22:49
pmichaud so, my $a = <3>; say $a.re; # method not found?
moritz_ *if* we make sure that methods that are supposed to return Complex always return Complex objects, and don't be "smart" about promoting to Num or Int
pmichaud: yes
22:50 skids left
pmichaud There's also been discussion that "3+4i" would numify to a Complex. 22:50
moritz_ hm.
22:50 jferrero left
pmichaud (I don't know that this is specced yet.) 22:50
22:51 japhb left
pmichaud in STD, <number> will match Complex numbers, though. 22:51
TimToady and Rats 22:52
pmichaud which means that 3+4i is now a term and not an infix:<+> and postfix:<i>, yes?
TimToady yes
likewise 22/7
pmichaud and thus numification of "3+4i" is a Complex
TimToady hmm, I bet it doesn't parse "4i" right though... 22:54
pmichaud ...but 3 + 4i is not a term?
TimToady correct
pmichaud how about +"3 + 4i" ?
TimToady but then do we allow spaces on input
moritz_ but still an expression, I hope?
TimToady yes, still an expression 22:55
maybe + squeezed out whitespace before parsing?
pugs_svn r25777 | moritz++ | [t/spec] tests for RT #63744
pmichaud then.... +"3 4 5" ? 22:56
moritz_ would be 3, in good ol' perl 5 tradition?
TimToady not if spaces squeezed first 22:57
moritz_ all spaces? not only leading?
TimToady +"3 + 4i"
moritz_ that will cause confusion
pmichaud moritz: if we don't squeeze internal spaces... what TT said
moritz_ I'm more in favor of optimizing the common case (non-complex numbers) 22:58
TimToady well, squeeze is the wrong word, remove
pmichaud I'm being called to dinner -- bbl.
TimToady that can be optimized, but if we get to the end of a normal number and it looks like there's more, it gives us more options 22:59
we do have to be a little careful not to make it so powerful that a malicious user can feed us "4 + BEGIN { run 'rm -rf /' }i" 23:00
so we can't just tell the user to parse with EXPR 23:01
hopefully <number> is sufficiently restricted to be broken out of like that
23:05 donaldh left 23:10 gbacon left
TimToady std: -Inf+NaNi 23:12
p6eval std 25777: OUTPUT«Undeclared name:␤ NaNi used at 1 ␤ok 00:02 34m␤»
moritz_ std: -Inf+NaN\i 23:13
p6eval std 25777: OUTPUT«ok 00:02 34m␤»
TimToady parses as expression, not term
meaning you can't type it at the prompt
maybe literal form should allow optional \
still not sure whether literal form should require 0+4i or allow 4i 23:15
23:16 gbacon joined
TimToady well, it only adds 16 more forms to the <number> lexer table 23:20
moritz_ std: 2 R** 3 23:21
p6eval std 25777: OUTPUT«ok 00:02 34m␤»
moritz_ std: 2 RR** 3
p6eval std 25777: OUTPUT«ok 00:02 34m␤»
TimToady std: 2 RRRRRRRRRRRRRRRRRRRRRRRRR** 3 23:23
p6eval std 25777: OUTPUT«ok 00:02 34m␤»
TimToady std: 2 XXXXRRRXXXRRRXXRXRXRXXR** 3
p6eval std 25777: OUTPUT«ok 00:02 34m␤»
wayland76 We need some vowel hyperoperators :) 23:29
std: 2 ROAR 3
p6eval std 25777: OUTPUT«##### PARSE FAILED #####␤Syntax error (two terms in a row?) at /tmp/Arz3FY209V line 1:␤------> 2 ROAR 3␤ expecting any of:␤ infix or meta-infix␤ infix stopper␤ standard stopper␤ terminator␤FAILED 00:02 34m␤»
moritz_ sense great opportunities for obfus
wayland76 std: 2 RandR 3
p6eval std 25777: OUTPUT«Undeclared name:␤ R used at 1 ␤ok 00:02 34m␤»
wayland76 std: 2 RorR 3
p6eval std 25777: OUTPUT«Undeclared name:␤ R used at 1 ␤ok 00:02 34m␤»
wayland76 std: 2 Ror 3
p6eval std 25777: OUTPUT«ok 00:02 34m␤» 23:30
wayland76 std: 2 Rand 3
p6eval std 25777: OUTPUT«ok 00:02 34m␤»
wayland76 std: rand Rand rand
p6eval std 25777: OUTPUT«ok 00:02 34m␤»
wayland76 hehe
moritz_ ;-)
wayland76 That would help obfus :)
TimToady std: $drugs RX= 42
p6eval std 25777: OUTPUT«##### PARSE FAILED #####␤Can't cross a item assignment because it's too fiddly at /tmp/PqBEge01rg line 1:␤------> $drugs RX= 42␤Other potential difficulties:␤ Variable $drugs is not predeclared at /tmp/PqBEge01rg line 1:␤------> $drugs RX= 42␤FAILED
..0...
TimToady std: 123 »XXX« 321 23:32
p6eval std 25777: OUTPUT«ok 00:02 34m␤»
wayland76 rakudo: say rxrokr ~~ "ok"
p6eval rakudo 95ce39: OUTPUT«Could not find non-existent sub rxrokr␤current instr.: '_block14' pc 82 (EVAL_16:42)␤»
wayland76 rakudo: say rx:rokr ~~ "ok"
p6eval rakudo 95ce39: OUTPUT«Statement not terminated properly at line 1, near ":rokr ~~ \""␤␤current instr.: 'parrot;PGE;Util;die' pc 129 (runtime/parrot/library/PGE/Util.pir:83)␤»
wayland76 rakudo: say rx,ok, ~~ "ok"
p6eval rakudo 95ce39: OUTPUT«Could not find non-existent sub rx␤current instr.: '_block14' pc 53 (EVAL_16:38)␤»
wayland76 rakudo: say rx/ok/ ~~ "ok" 23:33
p6eval rakudo 95ce39: OUTPUT«Could not find non-existent sub rx␤current instr.: '_block14' pc 53 (EVAL_16:38)␤»
wayland76 rakudo: say rx:i/ok/ ~~ "ok"
p6eval rakudo 95ce39: OUTPUT«Statement not terminated properly at line 1, near ":i/ok/ ~~ "␤␤current instr.: 'parrot;PGE;Util;die' pc 129 (runtime/parrot/library/PGE/Util.pir:83)␤»
TimToady perl6: my $rx = rx/ok/;
p6eval elf 25777: RESULT«Regexp.new(!!!)␤»
..rakudo 95ce39: OUTPUT«Statement not terminated properly at line 1, near "/;"␤␤current instr.: 'parrot;PGE;Util;die' pc 129 (runtime/parrot/library/PGE/Util.pir:83)␤»
..pugs: RESULT«\{regex}»
TimToady looks like rx is unimpl 23:34
wayland76 std: rx/rx/ RX= "rx"
moritz_ aye
p6eval std 25777: OUTPUT«##### PARSE FAILED #####␤Can't cross a list assignment because it's too fiddly at /tmp/UOkDWU0VEK line 1:␤------> rx/rx/ RX= "rx"␤FAILED 00:02 34m␤»
wayland76 std: rx/rx/ RXand "rx"
p6eval std 25777: OUTPUT«ok 00:04 34m␤»
wayland76 std: rx/rx/ RX+ "rx"
p6eval std 25777: OUTPUT«ok 00:04 34m␤»
wayland76 std: rx/rx/ RX~~ "rx" 23:35
23:35 Limbic_Region joined
p6eval std 25777: OUTPUT«ok 00:04 34m␤» 23:35
wayland76 No idea what the code would do though :)
TimToady bbl &
moritz_ match 'rx' against rx/rx/
an X on two single items doesn't make a difference 23:36
23:36 bacek_ joined
moritz_ rakudo: say 2 X+ 3 23:36
p6eval rakudo 95ce39: OUTPUT«5␤»
moritz_ bed& 23:37
wayland76 Cool. That's what I hoped it'd do :) Could help with obfus too :) 23:38
szabgab rakudo: say 12 ~~ /^ \d+ $/ ?? 'ok' !! 'not' 23:41
p6eval rakudo 95ce39: OUTPUT«ok␤»
23:41 ejs left
szabgab rakudo: grammar Y { regex TOP { ^ \d+ $ } }; say 12 ~~ Y ?? 'ok' !! 'not' 23:41
p6eval rakudo 95ce39: OUTPUT«not␤»
szabgab why is that not?
wayland76 Now you're scaring me with obfus ideas :) 23:42
(btw, TOP has been deprecated in favour of a parse method, IIRC)
szabgab rakudo: grammar Y { regex TOP { ^ \d+ $ } }; say 12 ~~ Y::TOP ?? 'ok' !! 'not' 23:43
p6eval rakudo 95ce39: OUTPUT«Method 'ACCEPTS' not found for invocant of class 'PGE;Match'␤current instr.: 'infix:~~' pc 19380 (src/builtins/match.pir:18)␤»
szabgab rakudo: grammar Y { regex TOP { ^ \d+ $ } }; say 12 ~~ /<Y::TOP>/ ?? 'ok' !! 'not' 23:44
p6eval rakudo 95ce39: OUTPUT«ok␤»
szabgab aha
wayland76 (oh, the TOP deprecation is in the spec, but may not be in rakudo yet)
23:44 pmurias joined
szabgab where in the spec? 23:44
wayland76 rakudo: grammar Y { regex TOP { ^ a\d+ $ } }; say 12 ~~ /<Y::TOP>/ ?? 'ok' !! 'not'
diakopter pmurias: I'm back now
p6eval rakudo 95ce39: OUTPUT«not␤»
szabgab TOP is just an arbitrary word here, isn't it? 23:46
wayland76 szabgab: Not sure, but you'll not that TOP doesn't appear at all any more, except as part of STOP
szabgab: TOP had some special meaning, but not any more
szabgab yeah, I just grepped the Specs too
wayland76 But you can still use it as a rule name
szabgab I did not know that
23:47 skids joined
wayland76 Here's what I wanted... 23:48
Matching against a C<Grammar> treats the grammar as a typename,
not as a grammar. You need to use the C<.parse> or C<.parsefile>
methods to invoke a grammar.
From S03
diakopter pmurias: did you get a chance to build mono from svn?
szabgab I don't understand that 23:49
Matching against a "Grammar" would mean if I declare XX as a grammar and then rry to match: $str ~~ XX 23:50
wayland76 Don't use X unless you want to do obfus -- X is a hyper-operator 23:55
...but to address your point...
rakudo: grammar example {}; say example.perl 23:56
p6eval rakudo 95ce39: OUTPUT«example␤»
wayland76 rakudo: grammar example { rule TOP { ^ \d+ $ } }; say example.perl
p6eval rakudo 95ce39: OUTPUT«example␤»
23:57 pmurias left
wayland76 rakudo: grammar example { rule TOP { ^ \d+ $ } }; say example ~~ /a/ 23:57
p6eval rakudo 95ce39: OUTPUT«Null PMC access in get_integer()␤current instr.: 'parrot;PGE;Match;new' pc 75 (compilers/pge/PGE/Match.pir:75)␤»
wayland76 rakudo: grammar example { rule TOP { ^ \d+ $ } }; say example ~~ "12"
p6eval rakudo 95ce39: OUTPUT«0␤»
wayland76 rakudo: grammar example { rule TOP { ^ \d+ $ } }; say example ~~ 12
p6eval rakudo 95ce39: OUTPUT«Use of protoobject as value␤0␤»
23:58 mikehh left
wayland76 rakudo: grammar example { rule TOP { ^ \d+ $ } }; say example ~~ "12" 23:58
p6eval rakudo 95ce39: OUTPUT«0␤»
wayland76 rakudo: grammar example { rule TOP { ^ \d+ $ } }; say example.parse("12")
p6eval rakudo 95ce39: OUTPUT«12␤»
wayland76 Those last two lines illustrate the point it's trying to make

Powered by Cro and the Raku® Programming Language.

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

Thank you!