»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.
00:01 shinobicl left, benabik joined
Juerd moritz: At this moment, 1.5G is a bit too much. Note that feather3 is for the unsafe part; perhaps building parrot automatically could happen on feather1, after which the binaries are copied to feather3? 00:08
moritz: I could extend feather1's memory to 2G 00:10
00:13 icwiener left, icwiener_ joined 00:17 UrielRLeon joined 00:18 UrielRLeon left 00:21 ab5tract left 00:35 localhost left 00:36 localhost joined
sorear good * #perl6 00:44
Juerd: Why is there more than one feather?
00:45 shinobicl___ joined 00:47 shinobicl___ left 00:50 icwiener_ left 01:07 Yappoko____ left
sjn moritz++ # For blogging about the hackathon at perlgeek.de/blog-en/perl-6/2012-upc...athon.html :) 01:10
sorear o/ sjn
01:13 FACEFOX left
japhb sorear, how does niecza handle laziness? And is the implementation of laziness unified into one section of the code, or distributed across lots of files? 01:13
sorear see IterHasFlat in Kernel.cs 01:14
anything that iterates a list uses that
japhb Ouch, github trying to display Kernel.cs causes browser script timeouts. 01:21
benabik Javascript syntax highlight? 01:27
01:27 skipper left 01:42 leprevost joined 01:44 TESTKEYz joined 01:46 envi_ joined 01:50 lestrrat left 01:51 lestrrat joined
cbk1090 I need help/want to learn how to make modules for Perl 6. I have some code(gist.github.com/2143601) but I'm not sure if it will work or how to get it to work. I would like to have self contained classes or modules, any ideas/tips? 01:57
02:00 scott__ joined 02:01 shinobicl___ joined 02:02 FACEFOX joined 02:04 TESTKEYz left, TESTKEYz joined 02:05 whiteknight left
shinobicl___ rakudo: sub f(Int :$x, Int :$y){ return $x + $y; }; sub f(Int $x, Int $y){ return $x * $y; }; say f(1,1); say f(x=>2,y=>2); 02:06
p6eval rakudo 704a21: OUTPUT«===SORRY!===␤Redeclaration of routine f␤at /tmp/G7PDUGroVw:1␤»
shinobicl___ how could i declare both functions? One with positional and another with named parameters? 02:07
sorear if you want multiple declarations, you need to say "multi sub"
02:15 FACEFOX left, libertyprime left
shinobicl___ rakudo: class C { multi method new(:$a!, :$b, :$c, :$x = 0) {say "named $x"; }; multi method new($a, $b, $c, $x = 0){ say "positional $x"; }; }; my $var = C.new(a=>1,b=>2,c=>3); my $var2 = C.new(1,2,3); my $var3 = C.new(a=>1,b=>2,c=>3,x=>88); my $var4 = C.new(1,2,3,99); 02:16
p6eval rakudo 704a21: OUTPUT«named 0␤positional 0␤named 88␤positional 99␤»
shinobicl___ this works as expected.
rakudo: class B { method new($Z){say "B - $Z";}; }; my $var0 = B.new(100); class C is B { multi method new(:$a!, :$b, :$c, :$x = 0) {say "named $x"; callwith($x)}; multi method new($a, $b, $c, $x = 0){ say "positional $x"; }; }; my $var = C.new(a=>1,b=>2,c=>3); my $var2 = C.new(1,2,3); my $var3 = C.new(a=>1,b=>2,c=>3,x=>88); my $var4 = C.new(1,2,3,99);
p6eval rakudo 704a21: OUTPUT«B - 100␤named 0␤Too many positional parameters passed; got 2 but expected 1␤ in method new at src/gen/CORE.setting:552␤ in <anon> at src/gen/Metamodel.pm:2579␤ in any call_with_args at src/gen/Metamodel.pm:2576␤ in block <anon> at src/gen/CORE.setting:390␤ in …
shinobicl___ this... i don't know. I want to call the B.new with callwith, but it reports the "Too many positional parameters" error 02:17
02:19 wolfman2000 joined 02:21 libertyprime joined
shinobicl___ rakudo: class A { multi method new(Int $x){ say "A with Int $x";}; multi method new(Str $s){ say "A with Str $s"}; }; class B { method new(Int $y) { say "B with $y"; return callwith($y); }; }; my $var1 = A.new(1); my $var2 = A.new("str"); my $var3 = B.new("text"); 02:29
p6eval rakudo 704a21: OUTPUT«A with Int 1␤A with Str str␤Nominal type check failed for parameter '$y'; expected Int but got Str instead␤ in method new at /tmp/Bco7QQoMD4:1␤ in block <anon> at /tmp/Bco7QQoMD4:1␤␤»
shinobicl___ rakudo: class A { multi method new(Int $x){ say "A with Int $x";}; multi method new(Str $s){ say "A with Str $s"}; }; class B { method new(Int $y) { say "B with $y"; return callwith($y); }; }; my $var1 = A.new(1); my $var2 = A.new("str"); my $var3 = B.new(1); 02:30
p6eval rakudo 704a21: OUTPUT«A with Int 1␤A with Str str␤B with 1␤Default constructor only takes named arguments␤ in method new at src/gen/CORE.setting:556␤ in <anon> at src/gen/Metamodel.pm:2579␤ in any call_with_args at src/gen/Metamodel.pm:2576␤ in block <anon> at src/gen/CORE.setting:3…
02:38 havenn joined 02:51 skids joined 02:58 s1n joined 03:12 shinobicl___ left 03:23 Chillance left 03:24 shinobicl___ joined 03:25 havenn left 03:32 alester joined 03:37 envi_ left 03:43 DarthGandalf left 03:52 orafu left, orafu joined 03:53 DarthGandalf joined 04:09 havenn joined 04:11 leprevost left 04:18 Yappo___ joined 04:27 Su-Shee left 04:29 Su-Shee joined 04:36 alester left 04:57 wolfman2000 left 05:38 birdwindupbird joined 05:40 kaleem joined, havenn left 05:51 fgomez left 05:58 NamelessTee joined 06:00 mucker joined 06:04 mucker left 06:05 mucker joined 06:20 thou left 06:21 TESTKEYz left 06:22 TESTKEYz joined 06:27 TESTKEYz left 06:34 shinobicl___ left
dalek p: ca74e29 | moritz++ | tools/build/PARROT_REVISION:
bump parrot revision to 4.2.0 release
 06:38
kudo/nom: 1a468d8 | moritz++ | tools/build/NQP_REVISION:
bump NQP revision to get parrot 4.2.0
06:47 mucker left
tadzik good morning #perl6 07:03
moritz \o tadzik
tadzik cbk1090: what do you mean by self-contained classes/modules? 07:04
cbk1090: you could look through the code of existing modules to see how they work: modules.perl6.org/ 07:05
07:09 daxim joined
sorear o/ tadzik 07:37
07:40 odoacre left 07:41 odoacre joined 07:45 mucker joined 08:05 bacek joined 08:11 bacek left, bacek joined 08:15 bacek left, bacek joined 08:16 bacek left
felher really likes the '0 but True' that is returned by 'shell' or such :) 08:16
08:17 bacek joined
moritz felher: I really hope that shell() will return something nicer and easier to inspect 08:17
08:17 bacek left 08:18 bacek joined, bacek left 08:19 bacek joined
felher moritz: What information should that nicer thing contain? 08:19
08:19 bacek left
moritz felher: well, I'm thinking of some more general object that qx// could also return 08:19
felher: which would then store the exit code, the STDOUT stream, the STDERR stream, possibly some timing information 08:20
sorear has an irrational dislike of 0 but True
moritz has a rational dislike of 0 but True 08:21
felher moritz: and will still return 'True' if i put it in boolean context and will still return 0 if i put it in numeric context?
moritz felher: yes
sjn AND it's defined :)
moritz though maybe an explicit .status would be clearer than just numeric context 08:22
tadzik yeah, 0 but True seems unsane
not insane, but unsane :)
08:23 am0c joined
moritz thing is, if you start passing around '0 but True' objects, people will come up with all sorts of crazy ways to guard against that 08:23
felher moritz: obvisouly, since this is a superset of the current '0 but True', im fine with that.
moritz p6: my $thing = 0 but True; if $thing ~~ Int && $thing ~~ Bool { say "I don't trust you" } 08:24
p6eval pugs: OUTPUT«*** Cannot cast from VBool True to VCode (VCode)␤ at Prelude.pm line 541, column 5-16␤»
..rakudo 1a468d, niecza v15-4-g1f35f89: ( no output )
moritz hm
sorear note, moritz, Bool ~~ Int 08:25
moritz felher: yes, I understand. Kinda like the context magic that Match possesses
p6: say Bool ~~ Int 08:26
p6eval rakudo 1a468d: OUTPUT«False␤»
..niecza v15-4-g1f35f89: OUTPUT«True␤»
..pugs: OUTPUT«␤»
moritz sorear: even if the spec says so, it feels wrong
because it doesn't respect Liskov in any way
felher moritz: it doesn't? Can't you use Bool everywhere you use int? (Or do i mix up liskov with something else) 08:27
nom: True.sin
p6eval rakudo 1a468d: ( no output )
felher nom: True.sin.say
p6eval rakudo 1a468d: OUTPUT«0.841470984807897␤»
sorear the only thing Int does that Bool doesn't is smartmatch 08:28
moritz felher: a Bool cannot hold all values that an Int can hold
sorear that only matters if Int/Bool is mutable, no?
08:28 NamelessTee left
moritz r: say Int.new(5) == 5 08:28
p6eval rakudo 1a468d: OUTPUT«Default constructor only takes named arguments␤ in method new at src/gen/CORE.setting:556␤ in block <anon> at /tmp/0VbTP_GsuS:1␤␤»
sorear I don't think it's really correct to say an Int "holds" a value 08:29
moritz my point is that Bool is more of a subset type of Int than an inherited type
08:29 mj41_nb joined
moritz well, it is a value 08:29
sorear "my Int $x" can hold considerably fewer values than "my Any $x"
Int is a subtype of Any :D 08:30
moritz I know, I know
sorear is half asleep, will leave when this is done
sorear times out, --> sleep 08:32
felher o/ sorear :)
08:33 fglock joined 08:35 bacek joined
moritz moritz.faui2k3.org/tmp/perl6-lines.txt statistics about how much we spoke in here 08:36
08:36 bacek left
moritz currently too lazy to turn it into a chart 08:36
08:36 bacek joined
tadzik svg::plot ttr! 08:37
08:38 bacek left, bacek joined
felher Hm, regarding liskov. Wikipedia says: "Let q(x) be a property provable about objects x of type T. Then q(y) should be provable for objects y of type S where S is a subtype of T.". That means that subtypes do have to beave exactly like 'T', doesn't it? Since 'does not have Method x' or 'does reject value of ...' are properties, every subtype that would have such a method violates the liskov-principle? 08:39
08:39 bacek left
felher *behave 08:39
08:40 bacek joined
moritz no, I think "property" is more narrowly defined here 08:40
08:40 bacek left 08:41 bacek joined
felher Ah, so property is more like 'can do'-properties? 08:43
08:43 bacek left
moritz yes 08:43
or think of it as an assertion 08:44
08:44 bacek joined
moritz for example for role Real, we dearly hope that is_approx $x.sqrt ** 2, $x; holds 08:44
08:45 bacek left 08:46 bacek joined, bacek left 08:47 bacek joined, bacek left
felher Okay, i think i got it on a 'not-so-formal' level. I try to get the paper never the less, just to see how it's defined. Thanks :) 08:48
moritz on a not-so-formal level it just says "substitution an object by the same object of a subclass shouldn't break stuff" 08:49
*substitutiting 08:50
it seems SVG::Plot is quite broken when using it with a single data set :( 08:54
and now I've tried to add some debugging output 08:56
and got a segfault :/ 08:57
tadzik moritz: do you use it precompiled?
moritz r: note 'oh hai'
p6eval rakudo 1a468d: OUTPUT«oh hai␤»
tadzik it did segfault for me when precompiled
moritz tadzik: I've precompiled SVG.pm and XML::Writer, but not SVG::Plot
tadzik I see 08:58
you could try removing those as well, maybe it'll help
moritz tries
09:03 bacek joined
moritz nom: my @a = eager gather for 1..10 { .say; take $_ } 09:05
p6eval rakudo 1a468d: OUTPUT«1␤2␤3␤4␤5␤6␤7␤8␤9␤10␤»
felher "Our paper is intentionally written in a descriptive and informal style" ... damn it ... :) 09:10
moritz read it anyway, it might still contain some interesting formalism :-)
felher moritz: yeah, quite possible :) 09:14
moritz moritz.faui2k3.org/tmp/perl6-lines.svg 09:15
I finally found the bug
felher nearly lookes like a sine ... :) 09:16
moritz SVG::Plot was using [min] and [max] to find the upper and lower bounds
and I fed it strings, not numbers
so [min] and [max] worked lexicographically, having $min > $max
which moved nearly all of the plot points outside the viewport 09:17
felher "11" < "2" .. :) 09:18
09:18 lestrrat left
moritz p6: say '11' lt '2' 09:19
p6eval pugs: OUTPUT«1␤»
..rakudo 1a468d, niecza v15-4-g1f35f89: OUTPUT«True␤»
09:19 lestrrat joined
masak good morning, #perl6 09:45
tadzik dzien dobry masak 09:46
masak dzien dobry, o tadziku!
tadzik :P
you usually omit the "o" here, unless you want it to sound ancient-ish :) 09:47
masak I... I sorta had that feeling :)
tadzik sounds a bit like "o królu"
masak moritz: it hurts sometimes that infix:<min> and infix:<max> don't coerce, like all the arithmetic, stingithmetic, and comparison ops do.
moritz masak: I really wonder if we want +min, ~min, +max, ~max operators 09:48
masak: or something along these lines
it feels we're missing a primitive
masak aye. 09:49
I think that would be good.
masak .oO( ?min, ?max -- exotic ways to spell && and || ) 09:50
09:51 dakkar joined
masak nom: my $o = class {}; say $o 09:53
p6eval rakudo 1a468d: OUTPUT«<anon>()␤»
09:53 NamelessTee joined
masak nom: my $o = class {}; say [max] $o, Inf 09:53
p6eval rakudo 1a468d: OUTPUT«Inf␤»
masak nom: my $o = class {}; say [min] $o, Inf
p6eval rakudo 1a468d: OUTPUT«Inf␤»
masak grrr :/
moritz what would zou want that to do? 09:54
s/z/y/
masak always treat Inf as bigger than anything else, including user-defined types. 09:55
moritz hm
masak the only thing that Inf isn't bigger than is Inf.
moritz what about NaN? 09:56
masak Inf is bigger than NaN.
moritz how much bigger? ;-)
masak or maybe I don't care about that one, I'll let the IEEE people fight that one out.
Inf/NaN arithmetic isn't consistent, so its specifics don't interest me that much :)
09:58 envi_ joined 10:02 kaleem_ joined
colomon r: my $o = class {}; say [min] $o, Inf 10:02
p6eval rakudo 1a468d: OUTPUT«Inf␤»
colomon n: my $o = class {}; say [min] $o, Inf
p6eval niecza v15-4-g1f35f89: OUTPUT«ANON()␤»
colomon \o/
n: my $o = class {}; say [max] $o, Inf 10:03
p6eval niecza v15-4-g1f35f89: OUTPUT«Inf␤»
10:03 kaleem left
masak niecza++ 10:06
moritz (mostly off-topic) I offer my IRC logging service to all open source communities; and my current attempt to gather some stats showed that one of them (#thegeekgroup) talks more than double as much as we do 10:07
but we are the second-most active channel that I log 10:08
10:15 kaleem_ left, scott__ left 10:19 kaleem joined
masak \o/ 10:31
<moritz> felher: a Bool cannot hold all values that an Int can hold 10:39
this is the way derived types work. 10:40
so I don't see how that's an argument at all.
10:40 tarch joined
masak oh, sorear++ also pointed that out, but with Any/Int :) 10:40
by the way, moritz.faui2k3.org/tmp/perl6-lines.svg is fascinating. 10:42
moritz was 2010-07 the month of the first Star release, by any chance?
masak compare it to this graph: moritz.faui2k3.org/tmp/perl6-lines.svg 10:43
er, this one: www.perlfoundation.org/perl6/index....47-0-13730 10:44
yes, the first rakudo star was 2010-07. 10:45
10:45 araujo left, pmurias joined
pmurias fglock: hi 10:45
masak we need a new ambitious goal. something to fire us up as much as Rakudo Star did. the next stage of the rocket.
if it can include Niecza as well as Rakudo, even better. 10:46
pmurias fglock: i'm reading through Perlito5::JavaScripdt::LexicalBlock, do we need something like the visitors from kp6? 10:47
masak the most fascinating bit of the graph is this, I think: the Pugs/golden age has the shape of a falling line, from early 2005 to late 2007. the Rakudo project has the shape of a parabola, spanning from early 2008 to late 2012. 10:48
of course, extrapolating after that is futile.
[Coke]_ is still waiting for masak to relaunch pugs! 10:49
[Coke]_ wonders why his nick is not right.
10:49 [Coke]_ is now known as [Coke]
moritz well, I kinda hope that much of the excitement will move from the compilers to the modules and docs we'll create 10:49
masak whatever Pugs will be, it won't be that kind of DDoS attack of activity again. I'm sorry, I'm not au. :)
moritz that would be a real sign of maturity 10:50
masak puts himself with the sun in his back, holding a banner, shouting "let's build a better tomorrow!"
[Coke] moritz: "oh, real mature."
murhaba, #perl6 10:51
fglock pmurias: I think that's overkill at this point; do you see a specific use?
pmurias if (0) {@bar} creates a package variable bar under "no strict" in perl5 10:52
i'm currently passing back info in a global $Perlito5::PKG_VARS 10:53
but seeing we do a traversal already i thought it migh want to have a visitor 10:54
s/traversal/partial traversal/
10:58 araujo joined
pmurias fglock: and we do 'use statement' expansion? 10:58
fglock about {@bar} - I think we should collect the declarations (and missing declarations) and add them at the end of the lexical block processing (we now do this at the start, but that's more expensive) 11:03
'use' is processed at the wrong place (in perlito5.pl) - this needs some refactoring 11:04
maybe add BEGIN{} first
we have 'do-file'; we can use that to implement 'require' and 'use' 11:05
lunch& 11:06
11:08 pmurias left 11:34 mucker left 11:47 skids left 12:00 lestrrat left 12:01 lestrrat joined 12:04 pernatiy left 12:16 birdwind1pbird joined
fglock pmurias: also related to LexicalBlock - figure out how to implement 'my $x if $z' and 'my $x = $x' 12:18
maybe make $x a stack
so it becomes roughly: push $x, undef if $z; push $x, undef if $x 12:19
same problem in p5-to-p6, I think 12:22
12:24 jaldhar left
fglock sorry, the 2nd one is: push @$x, $x[-1] 12:29
12:32 libertyprime left
moritz .u hbar 12:38
phenny moritz: Sorry, no results for 'hbar'.
12:40 birdwindupbird left
masak .u dash 12:40
phenny U+2013 EN DASH (–)
masak .u em dash
phenny U+2014 EM DASH (—)
moritz .u h bar
phenny U+2015 HORIZONTAL BAR (―)
moritz is looking for hbar really 12:41
masak .u double vertical line
moritz I want a LaTeX-to-Unicode lookup (and the other way round)
phenny U+2016 DOUBLE VERTICAL LINE (‖)
masak .u 210F 12:43
phenny U+210F PLANCK CONSTANT OVER TWO PI (ℏ)
masak that one?
moritz yes 12:44
thanks
masak kowtows
12:47 NamelessTee left 12:52 benabik left
masak .u benzene 13:05
phenny U+232C BENZENE RING (⌬)
masak .u benzene circle
phenny U+23E3 BENZENE RING WITH CIRCLE (⏣)
LoRe .u pile of poo 13:07
phenny LoRe: Sorry, no results for 'pile of poo'.
13:07 mucker joined
moritz phenny doesn't do chars > 0xFFFF, iirc 13:08
.u 10001
phenny moritz: Sorry, no results for '10001'.
masak just discovered en.wikipedia.org/wiki/Grimm's_law 13:14
13:16 shinobicl joined 13:25 pernatiy joined, birdwindupbird joined 13:26 icwiener joined 13:32 pmurias joined, skids joined
pmurias fglock: re my $x = $x, maping them to different javascript variables could work 13:32
fglock: why is my $x if $z special? 13:34
fglock: in p6-to-p5 use $OUTER::x 13:35
13:52 xinming_ joined 13:54 xinming left
masak oh look, this page is so old it uses the 'skip' no-longer-keyword. c2.com/cgi/wiki?GivenWhen 13:56
masak attempts to change it to 'succeed'
wow, that was easy. 13:58
c2++
timotimo so is it the only awesome feature in perl6? :)
masak yeah, we're all doing this twelve-year-and-counting project solely for the awesomeness of given/when :P 13:59
timotimo well, i could see that ... :p
PerlJam obviously Perl 5 has trouble getting given/when right. That Perl 6 does is quite awesome :) 14:01
masak p6: sub fac($_) { when * < 2 { 1 }; $_ * fac $_ - 1 }; say fac 5
14:01 Woodi left
p6eval pugs: OUTPUT«(timeout)» 14:02
..rakudo 1a468d, niecza v15-4-g1f35f89: OUTPUT«120␤»
masak \o/
masak wonders what Pugs thinks that means
moritz it probably doesn't exit the outer block
colomon could it be parsing it as (fac $_) - 1 ? 14:03
p6: sub fac($_) { when * < 2 { 1 }; $_ * fac($_ - 1) }; say fac 5
14:03 Woodi joined
PerlJam pugs didn't know about Whatevers did it? 14:03
p6eval pugs: OUTPUT«(timeout)»
..rakudo 1a468d, niecza v15-4-g1f35f89: OUTPUT«120␤»
PerlJam I think that first * is throwing purs for a loop ;-) 14:04
s/pur/pugs/
masak oh, yes.
moritz p6: sub f($_) { say 'inside when' when 2; say 'outside when' }; f 2
p6eval pugs: OUTPUT«*** ␤ Unexpected "when"␤ expecting operator, ":", "," or "}"␤ at /tmp/eQASgFqNnm line 1, column 31␤»
..rakudo 1a468d, niecza v15-4-g1f35f89: OUTPUT«inside when␤outside when␤»
masak pugs: sub fac($_) { when $_ < 2 { 1 }; $_ * fac $_ - 1 }; say fac 5
p6eval pugs: OUTPUT«120␤»
masak Pugs++
huh, pugs never had statement_mod when? :) 14:05
14:05 icwiener left, icwiener joined
moritz seems like 14:05
p6: sub f($_) { when 2 { say 'inside when' }; say 'outside when' }; f 2
p6eval pugs, rakudo 1a468d, niecza v15-4-g1f35f89: OUTPUT«inside when␤»
geekosaur pugs knows Whatever but it's a very early view thereof 14:06
14:08 ChanServ sets mode: +o moritz
fglock pmurias: my $x if $z will only create the new var if $z is true (at runtime) 14:08
14:08 moritz sets mode: -b *!*@turkeyberry.sgn.cornell.edu, moritz sets mode: -o moritz 14:09 dukeleto joined
dukeleto waves 14:09
fglock dukeleto: o/
pmurias fglock: are you sure about my $x if $z? 14:11
fglock testing
moritz eval: my $x = 5; my $x if 0; $x
buubot_backup moritz: No output.
moritz eval: my $x = 5; my $x if 0; [$x]
buubot_backup moritz: [undef]
masak dukeleto: you were banned last time because your auto-joiner is noisy when it gets in a loop with some other factor that times you out.
eval: $_ = 5; say "OH HAI" when 5 14:12
buubot_backup masak: ERROR: syntax error at (eval 20) line 1, near "say "OH HAI""
masak eval: use 5.016; $_ = 5; say "OH HAI" when 5
buubot_backup masak: ERROR: Perl v5.16.0 required--this is only v5.14.2, stopped at (eval 20) line 1. BEGIN failed--compilation aborted at (eval 20) line 1.
masak eval: use 5.014; $_ = 5; say "OH HAI" when 5
buubot_backup masak: OH HAI ERROR: Can't use when() outside a topicalizer at (eval 20) line 1.
masak eval: use 5.014; given 5 { say "OH HAI" when 5; say "never reached" } 14:13
buubot_backup masak: ERROR: syntax error at (eval 20) line 1, near "given 5" syntax error at (eval 20) line 1, near ""never reached" }"
masak eval: use 5.014; given (5) { say "OH HAI" when 5; say "never reached" }
buubot_backup masak: OH HAI 1
pmurias fglock: paste.scsys.co.uk/188107
dukeleto masak: yes, dircproxy default settings are to autorejoin, so when I would forget to close an irc session on a different machine, they go into a never-ending war
masak p6: given 5 { say "OH HAI" when 5; say "this *is* reached" }
p6eval rakudo 1a468d, niecza v15-4-g1f35f89: OUTPUT«OH HAI␤this *is* reached␤»
..pugs: OUTPUT«*** ␤ Unexpected "when"␤ expecting operator, ":", "," or "}"␤ at /tmp/w3BcEb_tiH line 1, column 24␤»
masak dukeleto: it won't be less distracting next time when I know the underlying cause. :) I'll likely ban you again. so if you find yourself banned again, that was likely why. 14:15
fglock ok - so not a problem :)
shinobicl hi 14:18
rakudo: class B { method new($Z){say "B - $Z";}; }; my $var0 = B.new(100); class C is B { multi method new(:$a!, :$b, :$c, :$x = 0) {say "named. $x"; callwith($x)}; }; my $var2 = C.new(a=>1,b=>2,c=>3)
p6eval rakudo 1a468d: OUTPUT«B - 100␤named. 0␤Too many positional parameters passed; got 2 but expected 1␤ in method new at src/gen/CORE.setting:552␤ in <anon> at src/gen/Metamodel.pm:2579␤ in any call_with_args at src/gen/Metamodel.pm:2576␤ in block <anon> at src/gen/CORE.setting:390␤ in…
masak hi shinobicl
shinobicl i got this problem.. i want to invoke the parent's new function with callwith but i got this error 14:19
moritz uhm
masak submits rakudobug 14:20
moritz that looks like weird interaction of multis, non-multis and callwith
is callwith supposed to stay within the same dispatcher?
masak it redispatches to the default .new, no?
shinobicl i got a class that inherits from Date. So, i want to provide the same constructors as Date has. I'm trying to use something like the code i posted to achieve that 14:21
moritz masak: and I think that's correct
masak I'm not so sure.
it doesn't vibe with my understanding of multi subs.
unless multi methods are vastly different, which they might be, this is wrong too.
moritz masak: because calling .new() without a positional argument narrows down the list of candidates that are available to nextsame
masak hm. 14:22
troo.
moritz r: multi a() { }; multi a($) { callwith() }; a(1)
p6eval rakudo 1a468d: ( no output )
masak yeah, this may be an anti-issue, then.
moritz r: multi a() { say 'here' }; multi a($) { callwith() }; a(1)
p6eval rakudo 1a468d: ( no output )
masak I mean, a non-issue.
moritz shinobicl: any reason not to call self.Date::new(ARGS) directly?
masak dangit, I *still* lack a robust mental model of what nextsame and friends do. :/
moritz it's *not* 'call a routine of the same name as me, but exclude me from the call' 14:23
shinobicl I haven't tried that syntax, but i want my derived class to be a date too. The thing is, my derived class has an additional attribute 14:24
so i'm doing this:
moritz shinobicl: of which you need to take care of anyway, indepently on how you call the parent constructor
shinobicl multi method new(:$year!, :$month, :$day, :$calendar = WorkdayCalendar.new) { my $D = callwith(year=>$year, month=>$month, day=>$day); ^M $D.calendar = $calendar; return $D; } 14:25
masak taking care of an additional attribute sounds more like a job for .BUILD, not .new
Woodi hi 14:26
shinobicl with that 'new', $calendar gets 'attached' to the object. Tyhe problem starts when i start adding more 'new's
masak hi Woodi
moritz shinobicl: try my $D = self.Date::new(...);
shinobicl ok, i'll do, thanks moritz
masak shinobicl: well, the reason for the problem was explained above by moritz++, if you're interested.
moritz shinobicl: oh, and if you have any suggestions on how to make the constructors in Date more friendly to subclassing, I'd be glad to hear them too
Woodi just read Star2012.02 release info and looks nice... especially future part :) 14:27
masak yeah. this is going to crop up again and again until we understand it well :)
Woodi: this month's star release is going to be awesome :)
moritz just hopes that jnthn++ can fix the MiniDBI segfault(s) before the release 14:28
Woodi but just compiled it and got "libparrot.so: warning: strcpy() is almost always misused, please use ..."
shinobicl I think the constructors in Date are well implemented. I just hit a wall with that "callwith" thing.
Woodi is it allow buffer overflow ?
moritz Woodi: not more than before; just the warning is new 14:30
Woodi warning is specific for OpenBSD gcc... 14:32
14:32 leprevost joined, cognominal left 14:33 cognominal joined
shinobicl thanks moritz++, with that change all of my constructors test passed :D 14:33
masak moritz++
14:34 kaleem left
masak (though it feels like defeat (1) to have to spell out the class explicitly, creating a teeny-tiny amount of dis-DRY, (2) to misjudge how the multimethod dispatch will work in a given moderately complex case) 14:34
14:34 Vespakoen joined
moritz agreed 14:37
I can also never remember if the default Mu.new is specced to be a multi method or an only-method 14:38
masak multi, I hope. 14:40
(because I expect a user-defined class to be able to do 'multi method new' and not shadow Mu.new) 14:41
moritz right; shadowing a multi is easier than adding candidates to an only 14:42
bbkr speaking of candidates - I did training about signatures today in my company based on Perl6 and people were really impressed by where{} constraints and things you can do with them :) 14:44
moritz :-)
14:44 thou joined
PerlJam bbkr: your company uses Perl 6? 14:46
:-)
bbkr yes, for utilities mostly. and a lot of things from P6 ported back to P5 such as Regexp::Grammars. 14:47
moritz bbkr: whoa. Where is that company? And are you hiring? 14:48
masak indeed. whoa.
moritz
.oO( "Mr. Lenz, what do you say to those who accuse of blatant job whoring?" -- "hire me!" )
 14:49
*accuse you
bbkr Implix in Poland. main product is getresponse.com - email marketing tool with heavy duty Perl workers sending up to 20_000_000 mails per day.
masak cool.
moritz wow. I can't even imagine the length you need to go to haven them not all flagged as spam immediately 14:50
s/haven/have/ # can't type today
masak .oO( put "*NOT SPAM*" in the subject line :P ) 14:51
pmurias moritz: you mean tricking spam filters ;)
moritz pmurias: I thought more of staying off of DNS blacklists and the likes
14:51 PacoAir joined
bbkr there are mechanisms for killing spammers who wants to use our system. still - maintaining high deliverability for legit users is very complicated :) 14:55
masak I bet.
14:55 Chillance joined 14:58 fgomez joined 14:59 alester joined 15:02 havenn joined 15:03 benabik joined 15:04 geekosaur left
felher In java, a method of a class can change private attributes of all objects of its class. Is this possible in Perl6? i.e. is the following possible in Perl6: class C { has $!priv; method x(C $other) { $other!priv = 3; } }; ? 15:05
bbkr that reminds me we still don't have Email::MIME. I'll try to hack something basic soon.
15:06 araujo left
masak felher: it should be, methinks. I don't know how current implementations conform, though. 15:06
felher masak: nom doesn't :) But i agree that it should be possible :) 15:07
masak felher: private attributes are only visible within the current class block. but I think the visibility isn't limited to the invocant. I might be wrong, though.
otherwise it's really difficult to write some things. 15:08
for example to construct a new instance of an immutable type from some operation on two existing instances.
moritz nom: class A { has $.priv; method a(A $x) { $x!A::priv = 5 ; say $x.priv} }; say A.new.a(A.new)
p6eval rakudo 1a468d: OUTPUT«Private method 'priv' not found on type A␤ in method dispatch:<!> at src/gen/CORE.setting:778␤ in method a at /tmp/Tv1fGPxYEY:1␤ in block <anon> at /tmp/Tv1fGPxYEY:1␤␤»
moritz nom: class A { has $.priv; method !priv() is rw { $!priv }; method a(A $x) { $x!A::priv = 5 ; say $x.priv} }; say A.new.a(A.new) 15:09
p6eval rakudo 1a468d: OUTPUT«Cannot assign to a readonly variable or a value␤ in method a at /tmp/1P2Tpk_SeG:1␤ in block <anon> at /tmp/1P2Tpk_SeG:1␤␤»
felher moritz: maybe i'm wrong but did you really want to try 'has $.priv' istead of 'has $!priv'? 15:10
moritz felher: yes, because I wanted the accessor for reading the result :-)
felher moritz: ah, i see :)
moritz felher: but since the accessor is RO, it doesn't harm here
masak niecza: class A { has $.priv; method a(A $x) { $x!A::priv = 5 ; say $x.priv} }; say A.new.a(A.new) 15:12
p6eval niecza v15-4-g1f35f89: OUTPUT«5␤True␤»
masak so, I guess what I'm saying is I'm with Niecza++ on this one.
15:13 geekosaur joined 15:14 tokuhiro_ joined
felher niecza: class A { has $!priv; method test() { my A $a .= new(); $a!priv = 3; } }; A.test(); 15:14
p6eval niecza v15-4-g1f35f89: ( no output )
felher \o/ :)
15:15 araujo joined, araujo left, araujo joined
moritz nom: role A[::T $] { also is T }; class B { }; class C does A[B] { }; say C ~~ B 15:16
p6eval rakudo 1a468d: OUTPUT«===SORRY!===␤Cannot type check against type variable T␤»
15:16 araujo left
moritz that's... somwhere between LTA and wrong, I'm not sure where exactly it is :-) 15:16
15:18 birdwind1pbird left
masak submits rakudobug 15:19
moritz n: role A[::T $] { also is T }; class B { }; class C does A[B] { }; say C ~~ B
p6eval niecza v15-4-g1f35f89: OUTPUT«===SORRY!===␤␤GLOBAL::T does not name any package at /tmp/8ui7SHbKPA line 1:␤------> role A[::T⏏ $] { also is T }; class B { }; class C ␤␤A type must be provided at /tmp/8ui7SHbKPA line 1:␤------> role A[::T⏏[…
masak n: class A { has $!priv }; sub test() { my A $a .= new(); $a!priv = 3 }; test(); say "alive" 15:20
moritz why does it need to name a package for inheritance?
p6eval niecza v15-4-g1f35f89: OUTPUT«===SORRY!===␤␤Cannot resolve class for private method at /tmp/AeYxRVjBCp line 1:␤------> ; sub test() { my A $a .= new(); $a!priv⏏ = 3 }; test(); say "alive"␤␤Unhandled exception: Unable to resolve method kind in class A…
moritz note that the type constraint allows subclasses too 15:21
which is why it wouldn't be sufficient for resolving private attributes, even if the compiler was sufficiently smart 15:22
15:24 araujo joined
moritz on a completely different topic, S12 says that 'self' is always assumed to be in item context 15:24
why is that the case? that's the role of the $ sigil, which isn't there on 'self'
masak the role of the $ sigil is forcing item context. 15:26
'self', by current spec, has that built-in.
moritz yes, but why? 15:27
masak I don't know. maybe because of some psychological expectation that 'self' is a single thing?
moritz we've extra eliminiated the $ from (|$foo) variables, because it doesn't fit the "$ means item" philosphy 15:28
masak do you have a use case where 'self' being in item context is cumbersome or unfortunate?
can you *guarantee* that a lot of existing Perl 6 code (in CORE, for example) doesn't rely on the assumption that 'self' is in item context? :)
moritz r: my @a = 1, 2; .say for @a.push: 3 15:29
p6eval rakudo 1a468d: OUTPUT«1␤2␤3␤»
moritz push simply returns self. Very convenient, and only possible because rakudo doesn't enforce the contnext 15:30
masak by the way, I wrote a blog post last week that was published today. it has vibes of #perl6 in it. some of you might find it pleasant reading: blog.edument.se/2012/03/21/how-to-t...effective/
moritz: so you're saying that Rakudo isn't to spec in that regard?
moritz masak: correct 15:31
masak submits rakudobug
masak grins
15:31 havenn left
moritz nom: class A is Array { method identity { self } }; my @a := A.new; @a = 3, 4, 5; .say for @a.identity 15:31
p6eval rakudo 1a468d: OUTPUT«3␤4␤5␤»
benabik r: my @a = 1,2; .say for $(@a.push: 3)
p6eval rakudo 1a468d: OUTPUT«1 2 3␤»
moritz there you go, without reference to any core thing
masak moritz: does Niecza mandate item context for 'self'? if it doesn't, I say excise the wording from the spec. if it does, what spectests does Niecza fail if the itemization logic is removed? 15:36
moritz n: class A is Array { method identity { self } }; my @a := A.new; @a = 3, 4, 5; .say for @a.identity 15:37
p6eval niecza v15-4-g1f35f89: OUTPUT«3 4 5␤»
moritz it does itemize
15:43 mj41_nb left 15:46 tokuhiro_ left, Psyche^ joined 15:47 tokuhiro_ joined, mj41 joined 15:50 Patterner left, Psyche^ is now known as Patterner, birdwind1pbird joined 15:51 tokuhiro_ left 15:55 fgomez left 15:57 birdwind1pbird left
sorear good * #perl6 16:05
masak sorear! \o/ 16:06
16:07 fgomez joined
sorear niecza: class A is Array { method identity { self } }; my @a := A.new; @a = 3,4,5; .say for @a.identity; my $x = $a; .say for $x.identity 16:08
p6eval niecza v15-4-g1f35f89: OUTPUT«===SORRY!===␤␤Variable $a is not predeclared (did you mean @a?) at /tmp/f6713e9Jxn line 1:␤------> = 3,4,5; .say for @a.identity; my $x = ⏏$a; .say for $x.identity␤␤Unhandled exception: Check failed␤␤ at /home/p6eval/n…
sorear niecza: class A is Array { method identity(\:) { self } }; my @a := A.new; @a = 3,4,5; .say for @a.identity; my $x = @a; .say for $x.identity 16:09
p6eval niecza v15-4-g1f35f89: OUTPUT«3␤4␤5␤3 4 5␤»
sorear masak, is that what you really want?
16:10 plobsing left
masak I think so, yes. moritz++ managed to sway me. 16:11
at least until I see a good use case for "self forces item context".
16:15 rhr left
fglock pmurias: t/base/lex.t parses now 16:18
16:19 araujo left 16:20 rhr joined 16:29 pyrimidine1 is now known as pyrimidine 16:33 fglock left, pyrimidine left 16:34 pyrimidine joined 16:40 havenn joined 16:42 araujo joined, araujo left, araujo joined 16:48 colomon left 16:51 ponbiki joined, araujo left, shinobicl_ joined 16:52 araujo joined, araujo left, araujo joined 16:53 daxim left, shinobicl left 16:54 envi_ left, envi_ joined
masak decommutes 17:02
17:03 fglock joined 17:11 mj41 left 17:12 birdwindupbird left 17:13 havenn left
felher Can i tell a hash to throw a exception if i try to read a element that he doesn't have? So that 'my %hash; say %hash<x>;' throws an exception? 17:14
17:15 NamelessTee joined 17:18 havenn joined 17:19 ponbiki left
sorear felher: no. 17:20
especially because reading and writing are the same.
benabik I wonder if you could get it to return a Failure on FETCH. 17:21
moritz that's probably doable 17:22
though of course it doesn't mix with autovivifcation
if you do something like %h<a>[1] = 0; , it will call method postcircumfix:<[ ]> on the return value from %h<a>
and if that's a Failure -> BOOM
benabik Point. 17:23
17:23 MayDaniel joined
felher sorear: reading and writing is the same, because %hash<x> returns and value that i can assign something in every case. So 'say %hash<a>' is just a special case, where i don't use the value for assignment? 17:26
Maybe 'returns a container' is better that 'returns a value'
moritz well, the container contains the value 17:27
benabik Isn't there a way to specify a default value for a hash? You could default it to Failure.new(:message => 'Item not in hash'), no?
moritz and say() is only interested in the value, not the container
benabik (Yes this wouldn't work with auto-viv, but you're asking for it.)
sorear returning a Failure is much less useful than throwing an exception 17:28
esp. since, by my reading of S04, most of the magic is in &fail 17:29
benabik Seemed like the easiest way to get reads to throw and writes not to.
sorear dumbly &return-ing a thing which happens to be a Failure will not do anything useful in most cases
moritz well, &fail just constructs a Failure object and returns the current routine 17:30
sorear reads won't throw at all
moritz you could, in theory, return a proxy that dies on FETCH
benabik It would throw when you attempt to use the value, no? That's pretty close to 'throw on read', I think.. 17:31
17:38 havenn left, kaare_ joined
felher benabik: the problem is: the function that accesses the hash only assigns it to variables. I will then die later in another part of the code, which probably doesn't have the knowledge to give a good error message. 17:40
I could write a ' || die ' at the end of every assignment, though... or derive from hash and let the new class throw the exception. 17:41
sorear // die better
benabik I've seen C++ comments like that. 17:42
moritz if you do it inside the hash (or derived class), you can get nice X::Hash::NoSuchKey(key => $yourkey) errors out of it
felher benabik: comments like what? And which part is it where you think i'm wrong? 17:43
benabik felher: // die better 17:44
17:44 havenn joined
felher benabik: ah, okay :) 17:44
sorear: yeah, right '//' is bettern than '||' :) 17:45
moritz: tempting :) 17:47
sorear commute& 17:49
17:54 Ulti left
felher I have a constructor in C++, e.g. "Class(int 3)", and function that needs an argument of type "Class", e.g. : "printClass(Class a) {...}". Then i can write "printClass(3)", because C++ knows that it can convert 3 to Class using the "Class(int e)"-Construcutor. Is that possible in Perl6, too? 17:58
Because then i could write: "method doSomethingWithTheHash(AutodieHash %hash)" and put normal %hahes in it... :) 17:59
18:00 pernatiy left
felher Maybe that should be: "is something like that possible in Perl6", since the default constructor concept ist quite different from C++'s :) 18:00
moritz felher: gist.github.com/2150358
felher moritz: omg, that is so cute :) 18:01
moritz++, thnx :)
moritz well, there's the coercion syntax 18:02
felher moritz: but regarding "%h does safety". I thoug "but" is for runtime-mixin of roles into objects?
benabik IIRC, but returns a new object and does mixes into the existing one.
moritz felher: 'does' is like 'but', but in-place
felher moritz: oh, sweet :)
benabik does is but=
moritz benabik++ is right
felher benabik: thanks :) 18:03
moritz benabik: under the hood there are fine differences
like, 'does' doesn't do any assignment
benabik moritz: True. Does but= even work?
moritz no idea 18:04
benabik r: role Foo {}; my @a; @a but= Foo
p6eval rakudo 1a468d: ( no output )
benabik r: role Foo { method a() { say 'yes' } }; my @a; @a but= Foo; @a.a
p6eval rakudo 1a468d: OUTPUT«Method 'a' not found for invocant of class 'Array'␤ in block <anon> at /tmp/LTY5dZWRk5:1␤␤»
moritz felher: you do so something like class SafeHash is Hash does safety { method &.(*@a) { coercion here } }
benabik Oh. It probably does @a = @a but Foo, which just re-stores the contents.
18:05 dakkar left
moritz felher: then you can write SafeHash(%other) and get a SafeHash back 18:05
18:05 tyatpi joined
moritz benabik: though in rakudo, 'does' is the more primitive operation, and 'but' is basically $lhs.clone does $rhs 18:05
18:06 am0c left
felher moritz: nice, thanks. I guess S13 is the right address, if want to know more about that? :) 18:06
moritz felher: and S02 maybe
anyway, I was quite surprised how that 'safety' role was
it worked on first try
though notice that it's expensive, it creates a new object for every access 18:07
felher moritz: couldn't one write something like: "if !self.exists($key) { die "blubb" } else { return nextsame($key) }" or something like that? Like delegating to the original at_key-method? 18:11
Juerd sorear: Anyone can get an account on feather1, and it's not very secured. As a result, it was not very practical to host production websites (and, back then, an svn server) on it. It's easy to over-load feather1, and that would also bring down sites like perl6.org 18:12
18:12 mucker left
Juerd sorear: The virtual machine feather3 is for evalbots, which are very insecure in nature as they allow arbitrary code execution to unauthenticated users by design. 18:12
moritz Juerd: I guess I should try to set up the build environment on feather1, and run evalbot on feather3 18:13
Juerd: if I run into any limitations, I'll tell you
Juerd moritz: That solution definitely has my preference and fits the separation's intent.
felher moritz: ah, obvisly not... assignment would never work, then... 18:14
Juerd I guess rsync over ssh will probably be an easy way to copy the binaries to feather3
But use any solution you like
tadzik good evening 18:16
felher o/
moritz Juerd: yes, it just requires me to put a passwordless ssh key on feather1. Not a great solution, but well, feather3 isn't secure anyway 18:19
Juerd moritz: Feel free to create a separate user account for this, on feather1, 3, or both 18:20
moritz both, for sure
18:28 tarch left 18:39 Vespakoen left, shinobicl joined, Vespakoen joined 18:41 shinobicl_ left 18:44 litejk joined, litejk left
TimToady felher: if a failure propagates to a different part of the code and dies, it's supposed to have enough internal info to tell you what went wrong originally 18:46
18:47 wolfman2000 joined
felher TimToady: I see, but ... i really have to catch the bus... :) 18:51
felher off 'til home
18:52 mj41 joined
TimToady sorear: exceptions are not better than failure under a dataflow paradigm 18:54
18:55 MayDaniel left, colomon joined
TimToady exceptions are a good example of hanging semantics on the wrong peg at least some of the time 18:56
I think overuse of exceptions is one of the traps of the imperative paradigm 18:58
control flow tends to make dataflow unanalyzable 18:59
19:00 birdwindupbird joined 19:03 libertyprime joined 19:06 tyatpi left 19:17 GlitchMr joined, envi_ left 19:19 tyatpi joined
TimToady I agree with the feeling that self should behave like a \self binding on the invocant and not itemize 19:20
moritz \o/ 19:21
TimToady: do you have any wisdom to share about what happens under the hood when 'use Module <argumentlist>;' is executed? 19:22
TimToady: ie, which part of the module gets the argumentlist?
sorry for fast context switching :-)
TimToady well, 'use' breaks down into a 'need' and an 'import' 19:25
STD currently treats those as primitives that look for a module's export declarations in an EXPORT package 19:28
there's no call into the module, per se
I think rakudo handles this with the metaobject somehow 19:29
19:29 havenn left
moritz if we want use lib $path; to work, we need some kind of call into the module 19:29
TimToady if there's actually going to be code run at import time, I think we should treat use/import as a funny kind of macro call 19:30
that can work either hygienically or not (or as a hybrid, partly text and partly AST) 19:31
one could imagine doing import with a textual 'my $foo := Their::Export::<$foo>' of some sort, in the limiting case 19:32
*EXPORT 19:33
moritz that would also explain how to do export a CATCH or CONTROL block, for example
TimToady (though I think EXPORT is lexically scoped, so you'd need a magical accessor)
you can basically do anything with a macro-view of importation 19:34
19:34 GlitchMr42 joined
TimToady it seems like a good unification to not reinvent the macro wheel badly 19:34
19:35 GlitchMr left
TimToady of course, the defaults should still encourage people to be as hygienic as possible 19:35
moritz that sounds surprisingly sane 19:36
I'd like to hear what masak++ has to say about it
TimToady so from the viewpoint of the module-macro, it's more like COMPILING:<$foo> := $foo
*:: 19:37
19:37 lichtkind joined
TimToady I've been thinking of 'use Stuff <a b c>' as a funny-looking macro call for a year or two or three now 19:38
19:38 GlitchMr42 is now known as GlitchMr
TimToady nap & 19:40
lichtkind good night :) 19:41
are the planned actors in perl 6?
shii found incredible stupid typos
19:41 fgomez_ joined, fgomez_ left, fgomez left 19:45 fgomez joined 19:47 ksi joined 19:49 wolfman2000 left
fglock 149 tests pass # perlito5 running perl5 test suite; this is not much, but these tests are hard :P 19:53
20:06 leprevost left, bacek left 20:07 Chillance left 20:08 Chillance joined
ashleydev fglock++ 20:09
lichtkind fglock++ 20:12
fglock pmurias++ # added node.js support 20:14
20:17 birdwindupbird left 20:22 bacek joined
masak what do I think of it? I think it sounds sane, but that it's hard to tell with just the skeleton of an idea. 20:23
most uses/examples of Perl 6 macros out there are a combination of fantasy and misunderstanding :)
outline the mechanism in as much detail as you can, and then we can talk about it. 20:24
I'm not sure 'use Foo;' could be an AST macro, because it contains an undeclared bareword Foo. 20:25
masak looks forward to the coming of the age when "it's a macro" stops meaning something abstract and starts meaning something concrete 20:26
20:27 pernatiy joined
moritz it might need to be "is parsed" 20:28
masak yeah, I'm not going near that.
maybe in another grant, but I suspect beer isn't gonna be convincing enough for that... 20:29
the thing with "is parsed" macros is that *anything* is potentially an "is parsed" macro. 20:30
moritz if!
for!
masak those two are innocent enough to be AST macros, even. 20:31
moritz uhm
masak hm, almost :)
moritz they have TTIAR
masak yeah, just (re-)noticed :)
apart from that :)
there could even be an intermediate form, something like a "mostly AST"/"only slightly 'is parsed'" macros, that let you write stuff like TTIAR and the parser will turn a blind eye because it's already seen the macro name. 20:33
20:33 pernatiy left
moritz niecza: sub term:<--->() { say 'foo' }; --- 20:33
p6eval niecza v15-4-g1f35f89: OUTPUT«foo␤»
moritz niecza: sub term:<--->($x) { say 'foo' }; --- 3 20:34
p6eval niecza v15-4-g1f35f89: OUTPUT«===SORRY!===␤␤Two terms in a row at /tmp/OrpZNtH7Pb line 1:␤------> sub term:<--->($x) { say 'foo' }; --- ⏏3␤Other potential difficulties:␤ $x is declared but not used at /tmp/OrpZNtH7Pb line 1:␤------> sub term:<-…
moritz ok, thought so
20:35 ponbiki joined 20:38 ponbiki left 20:44 fsergot joined
fsergot hi o/ 20:44
phenny fsergot: 19 Mar 16:56Z <sjn> tell fsergot gist.github.com/1711730 for your register-joyment :)
tadzik hi fsergot 20:45
masak fsergot! \o/ 20:46
I mean, dobry wiecz?r, o fsergotu. 20:47
fsergot nice masak. \o/ 20:48
:-)
masak tadzik-san taught me ;) 20:49
tadzik masak: I think it'd be like "o fsergocie" :)
masak oh, ok :/
tadzik but given the fact that "fsergot" is not a Polish word known to me, you can choose whatever you want :) 20:50
masak .oO( I must've fsergocie the rules... )
tadzik -u works for Królu, but I don't think that's regular at all, actually
O Tadeuszu, but O Filipie 20:51
probably there's some rule for this, but I never had a head for grammar rules
masak in other news, the strained multilingual puns will be delivered by me for the duration of jnthn's absence, to the best of my ability, which isn't much... :P
tadzik I consider myself well entertained :)
masak it's grammar-tainment hour! 20:52
tadzik: I felt I "got" grammar when I realized that it's just a description framework for how language behaves, just as math is "just" a description framework for number-like things behave. 20:54
20:54 y3llow_ joined, bruges_ joined 20:56 y3llow left 20:57 bruges left, y3llow_ is now known as y3llow
tadzik I guess so. It's just that it was always easier for me to just use it long enough to grok it instinctiv...something 20:58
20:59 skids left
masak to me, that's like saying "I never bothered about the principles of addition and multiplication, I just looked at enough examples until it fell into place". 20:59
tadzik heh
well, I learned Polish the very same way :) 21:00
masak but then, I'm fascinated by grammar as a phenomenon because I find it cool that such a thing even exists.
in the worst case, we might've been stranded with having to haphazardly stringing words together, trying to convery some nebulous "meaning" through juxtaposition and sheer wearing down of the listener. 21:01
tadzik yeah, it's interesting. Like, the languages probably evolve mostly on their own, and still you're able to write a strangely consistent spec for them
21:02 ksi left
masak instead, we have a framework with which to structure utterances according to fixed, explainable rules. and people who grok the rules are richer, more successful, and more beautiful than those who don't. 21:02
or so I've heard.
but srsly, language is one of the chief assets of a programmer. and there are probably few better forums to express that idea than #perl6 :) 21:03
21:04 icwiener left
masak sometimes I think that if it was Perl 5's mission to export effective "text munging" in its various forms to the world, it's Perl 6's mission to export effective "language munging" to the world. 21:04
21:05 GlitchMr left
masak ooh, nice p6l thread. 21:07
felher nom: my @array; say @array[4];
masak probably belonged to p6u, but who cares.
p6eval rakudo 1a468d: OUTPUT«Any()␤»
felher nom: my @array; say @array[*-1];
p6eval rakudo 1a468d: OUTPUT«Failure.new(exception => X::AdHoc.new(payload => "Cannot use negative index -1 on Array"))␤»
felher hoped for 'Any', too 21:08
masak probably it's * getting expanded as 0, and then the failure triggers (wrongly) 21:10
it could be argued that it should never trigger for whatever-y expressions.
masak submits rakudobug 21:11
felher \o/ 21:13
colomon perl6: say Inf.WHAT 21:19
jnthn Hm, you indexed into it with -1...isn't that a reasonable error? :)
p6eval rakudo 1a468d, niecza v15-4-g1f35f89: OUTPUT«Num()␤»
..pugs: OUTPUT«Num␤»
masak jnthn: no, the thing that is forbidden is writing .[-1] -- we want to catch people who *don't know that it's deprecated syntax* 21:20
jnthn: people who write .[*-1] clearly got the memo.
jnthn nom: my @array; say @array[*-2] 21:21
p6eval rakudo 1a468d: OUTPUT«Failure.new(exception => X::AdHoc.new(payload => "Cannot use negative index -2 on Array"))␤»
jnthn I fear [*-x] may do the computation and delegate to .[result] 21:22
21:22 fglock left
jnthn And the latter throws the error 21:23
masak that would explain Rakudo's less-than-correct behaviour, for sure. :)
jnthn Anyway, it's all written in Perl 6 in the setting for anybody who wants to twiddle :)
masak: I'm rather dubious about "less-than-correct" or about returning Any for something that is almost certainly an out of bounds bug, rather than a Failure that definedness tests in the right way but flags up the error if you expect it to be a value. 21:25
I think Rakudo's behavior could be argued to be helpful.
masak aye. I'm playing Debby's avocado here. 21:26
jnthn I'm not going to argue strongly either way. I just don't find it obvious that Rakudo's behavior is less than correct. :)_ 21:27
Unless you have a handy spec reference. :)
masak well, S09 already allows arrays with negative indices, so it's not forbidden per se. 21:29
jnthn Right, but in that case you'd have declared it
Also, we won't be expecting arrays that are fixed length to just return Any when you index outside of that bound, surely. 21:30
masak no, that probably throws. 21:31
so you're basically arguing that ordinary arrays are "semi-fixed" in length, in that they implicitly declare themselves never to have negative indices? 21:32
if so, fair enough. that's a consistent argument.
jnthn masak: Yes, that's my argument 21:33
21:36 pmurias left 21:47 havenn joined 21:53 xenu left 22:02 cogno joined 22:09 shinobicl left 22:12 benabik left 22:13 cogno left 22:18 cogno joined 22:20 plobsing joined 22:21 NamelessTee left 22:24 REPLeffect joined
sorear good * #perl6 22:28
masak sorear! \o/ 22:29
22:30 tyatpi left
masak 'night, #perl6 22:30
sorear I agree that exceptions aren't so hot for dataflow, but I'm still not convinced Failure can be made better :|
night masak. 22:31
22:36 snearch joined 22:37 M_o_C joined 22:38 cogno left 22:39 M_o_C left
colomon \o 22:40
sorear o/ colomon
22:41 sftp left 22:42 skids joined 22:48 sftp joined 22:52 kaare_ left 22:53 mj41 left 23:03 xenu joined 23:04 xenu is now known as Guest18260 23:15 nebuchadnezzar left, havenn left 23:21 PacoAir left 23:27 lichtkind left 23:28 thou left 23:29 mdupont left 23:30 thou joined 23:32 thou left 23:42 snearch left 23:44 whiteknight joined 23:53 shinobicl___ joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!