»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg camelia perl6: ... | irclog: irc.perl6.org/ | UTF-8 is our friend!
Set by moritz on 3 May 2013.
00:03 census left 00:10 chrisdev joined 00:28 tgt left 00:46 Chillance left 00:48 kurahaupo left, colomon joined 00:53 kurahaupo joined 00:58 raiph joined 01:09 anuby joined 01:26 [particle] joined 01:27 woosley joined 01:30 bonsaikitten is now known as AmazingPudding 01:54 FROGGS_ joined 01:58 FROGGS left 02:02 lustlife joined 02:15 chrisdev left 02:16 chrisdev joined 02:30 adu joined 02:33 jaldhar joined 02:35 chrisdev left 03:24 btyler joined 03:36 eternaleye left 03:40 eternaleye joined 03:58 btyler left 03:59 cognominal__ joined 04:00 preflex_ joined 04:01 preflex left, preflex_ is now known as preflex 04:03 cognominal left 04:09 kaleem joined 04:19 raiph left 04:22 raiph joined 04:25 Psyche^ joined 04:28 Patterner left, Psyche^ is now known as Patterner 04:29 AmazingPudding is now known as bonsaikitten 04:38 adu left 04:39 adu joined 05:00 snoopy joined 05:06 skids left 05:07 SamuraiJack joined 05:25 SamuraiJack_ joined 05:27 SamuraiJack left 05:28 SamuraiJack joined 05:45 dmol joined 06:02 dmol left 06:13 PacoAir left 06:16 kaare__ joined 06:52 arlinius left 07:13 celo joined
dalek ast: fab6f8e | (Brent Laabs)++ | S02-names/bare-sigil.t:
tests for RT #116521, bare sigils as placeholder variables
 07:14
07:20 celo left 07:23 cosimo left, cosimo joined
dagurval r: role B { }; class A does B { multi method new($x) { self.bless(*, { }) } }; A.new 07:24
camelia rakudo 0c1499: ( no output )
dalek kudo/nom: 8d2ec91 | (Brent Laabs)++ | t/spectest.data:
add S02-names/bare-sigil.t to spectest
dagurval Is this a rakudobug right?
It shouldt get the default constructor from role B
r: class A { multi method new($x) { self.bless(*, { }) } }; A.new 07:25
camelia rakudo 0c1499: ( no output )
dagurval oh, it's not the role. It's because of multi
r: class A { method new($x) { self.bless(*, { }) } }; A.new
labster it's a multi method, so it's still inheriting from Any
camelia rakudo 0c1499: OUTPUT«Not enough positional parameters passed; got 1 but expected 2␤ in method new at /tmp/GhrSrsU4H1:1␤ in block at /tmp/GhrSrsU4H1:1␤␤»
dagurval how do I delete the default constructor? 07:27
sorear don't use multi. 07:28
dagurval that's not an answer. I may need multiple constructors, but not an empty one. 07:29
for example, to fix this bug:
r: DateTime.new.say
camelia rakudo 0c1499: OUTPUT«use of uninitialized value of type Int in string context in block at /tmp/JFwMwXXfkV:1␤␤DateTime.new(year => , month => 1, day => 1, hour => 0, minute => 0, second => 0.0)␤»
labster multi new () { fail "must provide arguments to DateTime.new()" }; 07:31
er, multi method new 07:33
dagurval ok, needs to be explicit. I think it's weird behaviour to have a implicit empty new, if there exist others. 07:34
At least from me, used to C++.
s/from/for/
labster Perl: DWIMing except when it doesn't. 07:35
dagurval hehe :) 07:36
labster once Europe wakes up, they may have better suggestions for you.
dagurval is in urope o/ 07:37
but I have a kid that drags me up at 06am, so I guess that doesn't count :)
07:37 FROGGS_ left
labster oh yeah, dagurval++ thanks for fixing my dir() bug, but I totally didn't notice your pull request until I had after I had submitted by own. 07:40
moritz is awake agion
*again
07:41 FROGGS_ joined 07:42 FROGGS_ is now known as FROGGS
dagurval labster: no problem :) labster++ for nice new dir() roast tests 07:44
labster I need to rewrite io-path.t too...
07:51 arlinius joined 07:59 rindolf joined 08:03 adu left 08:04 SamuraiJack__ joined, SamuraiJack left 08:16 kurahaupo left 08:18 xilo left 08:19 xilo joined
lizmat morning #perl6! 08:20
sorear o/ lizmat 08:23
labster Hi lizmat! 08:26
lizmat hi sorear, labster!
08:32 salv0 left 08:44 dakkar joined 08:46 domidumont joined 08:56 fgomez left
masak good antenoon, #perl6 08:57
moritz \o masak, labster, lizmat, * 08:59
lizmat wrt to .hash(…), how can I check if a (named) parameter has been passed with "Any" ? 09:01
r: sub a (:$type) { say "type=$type" }; a; a(type=>Any) 09:02
camelia rakudo 8d2ec9: OUTPUT«use of uninitialized value of type Any in string context in sub a at /tmp/mNmATtWDzJ:1␤␤type=␤use of uninitialized value of type Any in string context in sub a at /tmp/mNmATtWDzJ:1␤␤type=␤»
masak ah, the old "how can I tell if a parameter has been passed" question? :)
lizmat: you set it to an unpassable default, and test against that default ;)
yes, it's a workaround, but it's what we have.
masak digs up the appropriate RT ticket 09:03
sorear
.oO(you set :opt_flag on the next parameter)
lizmat my first guess was using Mu , but that bombs
masak tee hee
lizmat r: sub a (Mu :$type Mu) { say "type=$type" }; a; a(type=>Any)
camelia rakudo 8d2ec9: OUTPUT«===SORRY!===␤Missing block␤at /tmp/oi0RLGu_yL:1␤------> sub a (Mu :$type ⏏Mu) { say "type=$type" }; a; a(type=>Any␤ expecting any of:␤ statement list␤ prefix or term␤ prefix or meta-prefix␤ new nam…
masak lizmat: yeah, that clashes with junctionality, methinks.
lizmat: missing an '=' there
lizmat r: sub a (Mu :$type= Mu) { say "type=$type" }; a; a(type=>Any) 09:04
camelia rakudo 8d2ec9: OUTPUT«use of uninitialized value of type Mu in string context in sub a at /tmp/r3Ad3DmKvq:1␤␤type=␤use of uninitialized value of type Any in string context in sub a at /tmp/r3Ad3DmKvq:1␤␤type=␤»
sorear masak: You can't set an unpassable default 09:05
lizmat except that we wanted Mu as well...
only an improbably default
*improbable
moritz well, there's one way
r: my $a = []; sub f($d = $a) { say $a === $d }; f(); f([]); 09:06
camelia rakudo 8d2ec9: OUTPUT«True␤False␤»
lizmat as long as $a is "secret", that should work...
moritz if you make sure that $a doesn't leak to the outer scope, it's a unique object that a caller can't reproduce
exactly
masak sorear: sure you can. just use a singleton object of an anonymous class.
lizmat moritz++
sorear moritz: isn't there a way to get at the default through the MOP?
masak ah, moritz++ had the same idea but with an array. 09:07
moritz sorear: through introspection, yes
but well, if the caller goes to *those* lengths to fool us, I guess he deserves to get the result of us being fooled :-)
masak .oO( you can always hack the system by bombarding the CPU with alpha particles ) 09:08
sorear r: sub a($x = ($*passed = True)) { my $*passed; say $*passed }; a(2); a() 09:09
camelia rakudo 8d2ec9: OUTPUT«(Any)␤True␤»
sorear er, that logic is backward
moritz s:g/passed/default/
sorear++
masak najs. 09:10
09:10 domidumont left
masak r: sub a($x = (class { method gist { "DEFAULT" } }).new) { say $x }; a(42); a() 09:11
camelia rakudo 8d2ec9: OUTPUT«42␤DEFAULT␤»
masak if you expect to be doing that a lot, but the anonymous class instantiation in a sub.
lizmat hopes for some less convoluted syntactic sugar 09:12
09:13 domidumont joined
moritz the usual approach is just to test .defined, and take an undefined value as a synonym for missing values 09:13
ater all, how often do you want undefined to behave differently than not-supplied?
masak yeah, this is called the "predicate problem" IIRC. the LISP people were already well aware of it early on.
it's one of those problems, it's only really problematic if you stop and think for a while :) 09:14
take JavaScript's notion of "truthy" values, for example.
you check if a function is defined by just checking the truthiness of a slot.
moritz so the solution is not to stop and think about it? :-)
masak yeah, basically.
that works because functions are objects are always truthy by definition.
lizmat already stopped thinking about *that* 09:15
masak now, *theoretically*, someone may have put a 0 or an "" in that slot.
then it's their loss.
moritz but in Perl 6, you'd better not do that (check the truthiness of code objects)
lizmat I'm still not however sure how to introspect type from a given hash
masak ah, en.wikipedia.org/wiki/Semipredicate_problem
lizmat: finally found it: rt.perl.org/rt3/Ticket/Display.html?id=64928
moritz because regexes are also code objects
and if you check their truthiness, they match against $_
masak (RT croaked on me when I searched, but I found it through Gmail) 09:16
moritz which... might not be what you expected
masak moritz: ooh, that's a good point. dangerous.
moritz (which happened to me once inside the backtrace printer; masak++ reported that shortly before the Oslo hackathon)
masak moritz: that's what I mean by "equal and opposite WAT" :)
afk & 09:17
09:23 eternaleye left
sorear allow me to bring up a different perspective 09:25
moritz allows it 09:26
sorear in general, you write functions because you want them to be called
labster o/ moritz, masak
sorear when you're wrapping a function in another function, often you want to pass some subset of arguments straight through
tadzik got a US visa o/
FROGGS morning @all
lizmat tadzik++
sorear if you distinguish "not passed" from "default", you're making it unnecesarilly difficult to make useful wrappers 09:27
lizmat morning tadzik, FROGGS
moritz tadzik: congratulations on you fooling them into believing you're not a terrorist
sorear tadzik: grats!
tadzik that was funny
moritz sorear: I agree
tadzik I had an appointment at 9. I had to wait 20 minutes in the first queue to be let into the second queue. In the second queue there were 60 people before me
FROGGS lizmat: I think I solved my problem from yesterday
lizmat r: my $d=[]; class A { method a( :$type ) { say $type === $d ?? "DEFAULT" !! "SPECIFIED" }}; my $a=A.new; $a.a # seems the === trick is not working?
camelia rakudo 8d2ec9: OUTPUT«SPECIFIED␤»
09:28 spider-mario joined
sorear lizmat: you forgot the = $d 09:28
lizmat duh
r: my $d=[]; class A { method a( :$type = $d) { say $type === $d ?? "DEFAULT" !! "SPECIFIED" }}; my $a=A.new; $a.a # seems the === trick is not working?
camelia rakudo 8d2ec9: OUTPUT«DEFAULT␤»
lizmat sorear++ for the reality check
FROGGS lizmat: I'm augmenting types like Str by adding methods like P5Stringy, and these will be used by the ast produced by Perl5::Actions
labster tadzik: I apologize for my country's visa process. 09:29
lizmat sounds cool
tadzik labster: no need, I don't believe it's your fault
FROGGS lizmat: this way, a variable defined in v6-land will behave fivy in v5, and sixy everywhere else
tadzik the upside is that's it's valid 10 years
FROGGS (and will do the same when declared in v5 land)
tadzik hopefully in 10 years Poland will be considered civilized enough 09:30
sorear 10 years is a long time. 10 years ago my politics were basically the diametric opposite of what they are now.
labster Long enough to make a Perl 6, one would think :/ 09:31
moritz must visit Poland to inspect if it's civilized enough :-)
tadzik please do :) 09:32
labster: or a C99 implementation... ;) 09:33
sorear plenty of C99 implementations exist today. 09:34
tadzik is any of them complete?
at least gcc isn't
moritz I thought most of them were, except for the microsoft compiler 09:35
tadzik gcc.gnu.org/c99status.html has plenty fields Missing
it's more complete than perl6.org compilers matrix, but still :)
09:36 ChanServ sets mode: +o moritz
sorear the missing parts of c99 are the parts nobody uses anyway 09:36
09:36 moritz sets mode: -o moritz
sorear getting ready to kick me? :p 09:36
moritz sorear: no, accidentally op'ped me in the wrong channel 09:37
:-)
moritz wanted to set a /topic elsewhere
labster masak: you mentioned here rt.perl.org/rt3/Ticket/Display.html?id=112276 that slurp to should default to $*IN, but the spec never got changed. Would you like me to do it? 09:38
sorear
.oO( opt )
FROGGS moritz: there are other channels than #perl6?? O.o
that doesn't feel right :P
sorear are you being serious
moritz FROGGS: #ilbot is much more quiet than #perl6, but I created it to be not so much off-topic in here 09:39
labster masak: or was there a reason to keep it with $*ARGFILES
sorear slurp totally should be $*ARGFILES, not $*IN 09:40
that's what P5 does with the <> magic, it works well and there's no reason to change it
FROGGS moritz: does ilbot connect to irc.perl.org too? I'd like to have #sdl logged...
moritz it isn't? 09:41
FROGGS: sure
sorear $*ARGFILES is a relatively recent addition to Rakudo; defaulting to $*IN may be a fossil
moritz FROGGS: do the other inhabitants of #sdl agree with that?
FROGGS moritz++ # thanks, though it won't cost much traffic :o)
moritz: it once was logged, but the server where the bot was running well, got unmaintained... so there is an agreement about that 09:42
sorear it should be prominently announced that the channel is now again being logged 09:43
FROGGS I'll change the topic, and I could write a message to the mailing list
moritz FROGGS: I dimly recall a channel that I once logged, and then deleted all the logs on request from one of the regulars, because he didn't realize *how* public it would be
but I'm not sure if that was #sdl or not
labster I get where you're coming from sorear, but the discussion in the bug report went the opposite way.
09:44 gaussblurinc1 joined
FROGGS moritz: I know the logs where on 'our' server at sdlperl.ath.cx or so 09:44
09:56 eternaleye joined 10:00 tgt joined 10:04 sqirrel joined
lizmat isn't "my %h of Int" supposed to be an alternative to saying "my Int %h" ? 10:04
r: my %h of Int; say %h.of; my Int %i; say %i.of 10:05
camelia rakudo 8d2ec9: OUTPUT«(Mu)␤(Int)␤»
lizmat a case of NYI, I assume? 10:06
10:07 gaussblurinc1 left 10:19 eternaleye left 10:22 eternaleye joined
arnsholt lizmat: Most likely, yeah 10:37
It's implemented for return types on subs, so grepping for trait_mod:<is> in Rakudo should be helpful 10:38
lizmat r: my Int %h{Any}; say %h.WHAT; say %h.of; my %i=%h.clone; say %i.WHAT; say %i.of # .clone doesn't work on typed hashes :-(
camelia rakudo 8d2ec9: OUTPUT«(Hash+{TypedHash})␤(Int)␤(Hash)␤(Mu)␤»
lizmat rakudobug? I think so :-( 10:39
arnsholt Certainly a bug, I'd say. Not sure if it's a NYI-bug or ordinary bug though 10:40
lizmat it's bloody annoying, that's what it is… this cost me the better part of this morning trying to allow .hash :type and :of named parameters 10:41
10:42 anuby left
arnsholt Anyways, it might be due to the TypedHash mixin 10:42
10:42 woosley left
arnsholt There's another annoying bug where Array[Int] !~~ Array[Int], which is due to mixins 10:43
10:43 kaleem left
lizmat I wonder whether the problem isn't really that we cannot have a binding return value. Or can we? 10:51
currently .hash is defined as "method hash { my %h = self }" 10:52
so what happens, is that we create a hash inside the scope of the method, and then clone it on the way back
10:53 sqirrel left
lizmat whereas we might have well bound it: something like "my %i= <a 1 b 2>.hash" would in fact need to do "my %i := METHOD::%h" as it were 10:54
masak, moritz, jnthn: am I making sense here, or is this a noob rambling :-) 10:55
afk for a few hours
masak lizmat: re the NYI: you're trying something no-one's ever really done with Rakudo (cloning a typed hash). so yes, bugs fall out. frustration follows. the best (meta-)workaround I've found for that is to file rakudobugs. it helps, in the sense that lots of them get fixed, sometimes even relatively quickly. 11:00
<lizmat> I wonder whether the problem isn't really that we cannot have a binding return value. Or can we?
I don't understand what you mean here.
and that seems to be the core of your explanation, so maybe rephrase? :) 11:01
lizmat sorry if I seemed annoyed, I was :-) 11:02
dalek ast: b9c3771 | dagurval++ | S32-temporal/DateTime.t:
Added test for empty DateTime constructor
ast: 200549a | dagurval++ | S32-temporal/calendar.t:
remove newline from expected gist output
11:03 gaussblurinc1 joined
lizmat masak: sub a { my %h }; my %i= a # this is in fact %i= %h.clone, I would like it to be %i := %h 11:03
masak ah. now it's clear. 11:06
that actually feels *more* right to me, at first blush. 11:07
hm. I think.
if the patch passes spectest, I'm all for it. I'd apply it.
11:13 sciurius_ left
lizmat is intermittently at keyboard 11:13
one could argue that what I propose is an optimisation
so I will first focus on getting .clone to work on typed hashes 11:14
masak here's how far I can see into the issue: arguments/parameters are bound by default, so it only makes sense that return values be, too.
lizmat I have a patch for .hash( :type(Any), :of(Int ), but it doesn't work because .clone doesn't :-(
masak *but*, there *is* an assignment there, in '%i = a', and that bugs me a bit.
lizmat hmmmm good point 11:15
masak but not too much, since (I think) '%h1 = %h2' also doesn't do a clone, just a reference copy.
r: my %h1; my %h2 = %h1; say %h1 === %h2
lizmat well, whatever, the effect is the same as clone
camelia rakudo 8d2ec9: OUTPUT«False␤»
masak hm.
or not.
lizmat r: my Int %h{Any}; say %h.WHAT; say %h.of; my %i=%h.clone; say %i.WHAT; say %i.of # .clone doesn't work on typed hashes :-(
camelia rakudo 8d2ec9: OUTPUT«(Hash+{TypedHash})␤(Int)␤(Hash)␤(Mu)␤» 11:16
masak I notice that I am confused. I again defer you to the spectest, a better arbriter than I :)
lizmat r: my Int %h{Any}; say %h.WHAT; say %h.of; my %i=%h; say %i.WHAT; say %i.of
camelia rakudo 8d2ec9: OUTPUT«(Hash+{TypedHash})␤(Int)␤(Hash)␤(Mu)␤»
masak I definitely agree that there's something in .clone of typed hashes that should be fixed.
lizmat the reference copy doesn't copy the type or of either, I was assuming it was using .clone under the hood
masak what I meant by "refernce copy" is disjunct from anything having to do with .clone 11:17
after a reference copy, there is still only one hash object in play. not two.
FROGGS nqp: say( 1 // 2 )
camelia nqp: OUTPUT«1␤»
lizmat r: my Int %h{Any}; say %h.WHAT; say %h.of; my %i:=%h; say %i.WHAT; say %i.of 11:18
camelia rakudo 8d2ec9: OUTPUT«(Hash+{TypedHash})␤(Int)␤(Hash+{TypedHash})␤(Int)␤»
lizmat afk again
masak r: class O {}; my $o1 = O.new; my $o2 = $o1; say $o1 === $o2
FROGGS nqp: my $x; $x //= 7; say($x); $x //= 8; say($x);
camelia rakudo 8d2ec9: OUTPUT«True␤»
nqp: OUTPUT«Confused at line 2, near "$x //= 7; "␤current instr.: 'panic' pc 14721 (src/stage2/gen/NQPHLL.pir:5232) (src/stage2/gen/NQPHLL.nqp:279)␤»
masak that's a reference copy.
11:23 betterwo1ld left 11:26 betterworld joined 11:27 tgt left 11:28 patspam left 11:33 patspam joined 11:38 tgt joined 11:50 kaleem joined, tgt left
dagurval r: Int.min.say; Int.max.say; 11:52
camelia rakudo 8d2ec9: OUTPUT«Inf␤-Inf␤» 11:53
dagurval should it not be the other way around?
masak I can see what set of expectations you're running under that'd make you think that ;) 11:54
r: say ().min; say ().max
camelia rakudo 8d2ec9: OUTPUT«Inf␤-Inf␤»
spider-mario r: Int.min.say
camelia rakudo 8d2ec9: OUTPUT«Inf␤»
masak Inf is the "identity value" of the min operator.
spider-mario oh, ok 11:55
masak -Inf is the "identity value" of the max operator.
just like 0 and 1 are for addition and multiplication, respectively.
11:55 crab2313 joined
masak r: say [+](); say [*]() 11:55
camelia rakudo 8d2ec9: OUTPUT«0␤1␤»
masak r: say [min](); say [max]()
camelia rakudo 8d2ec9: OUTPUT«Inf␤-Inf␤»
masak dagurval: all that said, yes, I can definitely see your expectation being fairly widespread, and *not* honoring it being some kind of breaking Least Surprise. 11:56
11:56 crab2313 left
masak question is, which weighs more heavily here: consistency of .min and .max, or user expectations in this particular case? 11:57
dagurval hehe, indeed
lizmat this actually, to me, feels like an implementation detail leaking out 11:58
dagurval what consistency whould be broken?
I would rather not have .max .min on Int, than to have them behave like that. 11:59
Instead perhaps have limits class like in c++, std::limits<int>::max()
s/limits/numeric_limits 12:00
gfldex r: say int.min; 12:01
camelia rakudo 8d2ec9: OUTPUT«Inf␤»
gfldex that's a lot of bits :)
12:01 jest1 joined
masak dagurval: I think the thing to understand is that type objects are a kind of instances of their type. 12:02
so, if the Dog class has a .bark method, then the type object (also spelled 'Dog') has a .bark method. 12:03
why Int has a .min and .max I don't know offhand, but probably because of the (intentional) confusion between items and lists in Perl 5 and 6.
12:03 pjcj_ left
masak r: class Dog { multi method bark() { say "called on an actual Dog" }; multi method bark(:U:) { say "called on a type object" } }; Dog.bark; Dog.new.bark 12:04
camelia rakudo 8d2ec9: OUTPUT«===SORRY!===␤Malformed parameter␤at /tmp/BUUbnkYxML:1␤------> on an actual Dog" }; multi method bark(:⏏U:) { say "called on a type object" } };␤ expecting any of:␤ postfix␤ statement end␤ statement modifi…
masak r: class Dog { multi method bark() { say "called on an actual Dog" }; multi method bark(Dog:U $self:) { say "called on a type object" } }; Dog.bark; Dog.new.bark 12:05
camelia rakudo 8d2ec9: OUTPUT«Ambiguous call to 'bark'; these signatures all match:␤:(Dog : Mu *%_)␤:(Dog:U $self: Mu *%_)␤ in method bark at src/gen/CORE.setting:440␤ in block at /tmp/pWHCFicdQ9:1␤␤»
lizmat in many contexts, an item is a one element list:
r: my $a=1; say $a[0]
camelia rakudo 8d2ec9: OUTPUT«1␤»
masak hrm.
r: class Dog { multi method bark(Dog:D $self:) { say "called on an actual Dog" }; multi method bark(Dog:U $self:) { say "called on a type object" } }; Dog.bark; Dog.new.bark
camelia rakudo 8d2ec9: OUTPUT«called on a type object␤called on an actual Dog␤»
dagurval so that's what :D and :U are for :) 12:06
masak there are two other letters, with slightly different meanings. they may or may not be more correct here, in fact. I haven't learned them yet. 12:07
12:07 SamuraiJack_ left
masak lizmat: the DWIM end of items-pretend-to-be-lists can be handy sometimes. the WAT end can be correspondingly horrible. 12:07
12:13 census joined 12:20 pjcj joined 12:22 gaussblurinc1 left
lizmat masak: agree, I was just elaborating for dagurval/gfldex 12:26
12:26 muixirt joined
masak aye. and I wasn't arguing against anything, just kvetching out loud :) 12:26
lizmat masak: S12:1616 "Abstract vs Concrete Types" 12:28
12:28 ajr joined 12:29 ajr is now known as Guest64703
masak aye. those are what I was thinking of. 12:29
the need for both :U and :T is another instance of the semipredicate problem popping up ;) 12:31
it's everywhere!
12:31 arlinius left
lizmat It's good that there is no :F, otherwise we could have called if :F:U:D :-) 12:32
masak I always feel sure what kind of smiley :D is, but I'm not really sure what emotions :U and :T represent... 12:35
12:35 jest1 left
lizmat :T is left headdesk? 12:35
or right?
masak .oO( and if you combine an L-headdesk with an R-headdesk, you get a commutative bi-headdesk ) 12:37
lizmat r: sub a ( $a=1, --> Hash ) { say $a; my %h }; say a(2) 12:38
camelia rakudo 8d2ec9: OUTPUT«2␤().hash␤»
lizmat r: sub a ( $a=1 --> Hash ) { say $a; my %h }; say a(2)
camelia rakudo 8d2ec9: OUTPUT«===SORRY!===␤Missing block␤at /tmp/NxD226NWlT:1␤------> sub a ( $a=1 --> Hash ⏏) { say $a; my %h }; say a(2)␤ expecting any of:␤ postfix␤ infix stopper␤ infix or meta-infix␤ prefix or term␤ …
12:38 kaare__ left
lizmat for some reason, if we have a default, we need a comma there 12:38
r: sub a ( $a --> Hash ) { say $a; my %h }; say a(2)
camelia rakudo 8d2ec9: OUTPUT«2␤().hash␤»
lizmat r: sub a ( $a --> Hash is ref ) { say $a; my %h }; say a(2) 12:39
camelia rakudo 8d2ec9: OUTPUT«===SORRY!===␤Missing block␤at /tmp/eL4JExgq_z:1␤------> sub a ( $a --> Hash ⏏is ref ) { say $a; my %h }; say a(2)␤ expecting any of:␤ statement list␤ prefix or term␤ prefix or meta-prefix␤ new name…
lizmat wouldn't that be a nice syntax for binding the return value, rather than cloning?
muixirt hi, while compiling rakudo CORE.setting.pbc (9c5650b) I get a segfault 12:40
lizmat I assume you're using the latest rakudo from HEAD ? 12:41
muixirt lizmat: almost, 1 day old 12:42
gfldex muixirt: what platform?
muixirt linux amd64
masak submits rakudobug for the required comma
lizmat hmmm… fetching last changes to rakudo 12:43
masak++
12:44 MrMeek left
muixirt I noticed that Rakudo got slower, is that temporary? 12:45
gfldex that's the plan
muixirt hmm, didn't segfault the second run 12:46
lizmat muixirt: slower in building, or in execution generally ?
12:47 chrisdev joined
muixirt lizmat: startup time and execution time 12:47
(compared to a Rakudo Star 2013.02.1) 12:48
12:49 tgt joined
[Coke] realizes that the sixperl phone call is still on his favorite contacts list, and rectifies that. 12:50
diakopter heh 12:52
12:58 salv0 joined, btyler joined, Guest64703 is now known as ajr_ 13:01 arlinius joined 13:13 usdmatt joined 13:14 usdmatt left
muixirt much of the slowness came from Parrot version 5.3.0-devel it seems 13:19
oh no, i forgot --optimize
whatever that does in current parrot versions 13:20
sorry for the noise
diakopter :) 13:21
13:22 kivutar joined 13:24 bluescreen10 joined
masak thank you, #perl6, for existing. it makes me happy, all the time :) 13:30
tadzik \o/
masak when I started getting involved in 2004-2005, I think I was suffering from the notion that most things had already been invented. that programming languages, the big ones out there, were basically the last word and there wasn't much left to do. 13:32
but being part of a language design process quickly liberates you from that notion :>
there is *lots* to do. we're only at the *start* of our very *understanding* of what computers and computing and programming *are*. 13:33
PerlJam heh
yoleaux 19 May 2013 03:56Z <[Coke]> PerlJam: about grammar-profiler-simple... the csv.t test has the plan commented out and no actual tests, which is why it's failing.
masak the various sub-guilds, FPistas and reactive programmers and logic programmers, all have pieces of the puzzle. they're building and exploring a better world. the mainstream languages are boring and old. we can do better -- much better. 13:34
PerlJam masak: Just keep in mind that computing is so very young. Look how look it took math (for instance) to get interesting.
s/look/long/
masak yes, that's what I mean. we're all *toddlers* when it comes to programming. there's some variance, but we're basically all setting out. in 20 years or so, we'll know a bunch more, have better tools, better reach, better understanding. 13:35
PerlJam (and still be relative toddlers :)
arnsholt I saw a video talk where the speaker (can't remember who it was, sadly) pointed out that there's a lack of historical knowledge as well 13:36
masak you bet there is. 13:37
PerlJam masak: I wonder if everyone has the same revelation at some point. when I first started using unixen, tools like FTP and archie were important and looked like they'd always been there. I found out later that they had just been invented a couple of years before I first heard of them.
arnsholt There's a lot of really awesome stuff from the Lisp-machine and Smalltalk worlds of the early 80s that's been forgotten
masak arnsholt: it's been forgotten because those languages didn't become mainstream.
arnsholt Or is being rediscovered (like MVC, which goes back to Smalltalk)
ajr_ The lack of historical knowledge is the reason each generation keeps "solving" the same problems.
(And announcing them as a wondrous new step inhuman progress. 13:38
)
PerlJam ajr_: in a way, that's okay because the context around the problems shifts.
arnsholt Also, when you look at the kind of introspection and tweaking that was possible in Smalltalk and Lisp-machine systems you realize that modern OSes aren't all that awesome
masak arnsholt: if you're wiped out by Worse is Better, you'll either spend the rest of your life being bitter about it, or (like Eirich Gamma et al.) you rebuild the tools in Java instead of Smalltalk. now, Eclipse is a weak ghost of what Smalltalk was, but at least it has the original ideals and it's being used a lot. 13:39
13:39 PacoAir joined
PerlJam
.oO( Why are we talking about smalltalk in the past tense? )
 13:39
ajr_ Reading "The Sciences of the Artificial" by Herbert Simon is very instructive
The fundamental problem in computing isn't computers; it's people. 13:40
PerlJam ajr_: truer words were never written :)
arnsholt masak: Yeah, I'm a bit ambivalent about Worse is better. On the one hand I see why it's like that, on the other hand I hate it =)
masak ajr_: that's what I like about this channel. we do cross-generation right. it's happened half a dozen times that I've brought up a problem here, and the people with decades of experience just go "yeah, we did that back in <decade before masak was born>. it's called the lagging variable pattern."
ajr_ We can make computers faster and give them more peripherals, but the Mark I Cerebellum doesn't get updated. 13:41
masak arnsholt: here in #perl6, we're in a Worse-Is-Better-free zone :P
arnsholt Second system done right, indeed =D
ajr_ One of the most interesting consequences of computer development has been the revelation of how little we know about thinking processes. 13:42
The history of AI is particularly valuable.
arnsholt Definitely. I made that very point this semester
Moravec's paradox is very relevant 13:43
And the summary "Hard problems are easy and easy problems are hard" is very pithy (and with chiasmus, which I like) 13:44
masak ajr_: www.ageofsignificance.org/ looks interesting and reminds me of what you're saying about thinking processes. I keep telling myself to read what's there already, but I always seem to get distracted. it's interesting, though.
ajr_ "Computers can do easily what PhD.s do, but not what toddlers do".
13:45 pjcj left
arnsholt Perhaps an even better summary =) 13:45
masak arnsholt: doesn't Moravec's paradox simply mean that (since brains don't constantly blow a fuse when we're lifing a carton of mild) a lot of assumptions of the real world are hard-coded into a healthy adult brain?
arnsholt: so we're not general computing devices; evolution has cut corners and cheated a lot on the way. 13:46
milk*
arnsholt I dunno. I see it more as a reflection of our lack of understanding of how certain fundamental tasks actually work
ajr_ "Mild" works too, but too much of it will disrupt the computation
arnsholt The example I used with my students was walking. It's something we do without really thinking about it, yet making a walking robot is actually very hard 13:47
ajr_ (Mild is a kind of English beer.)
13:48 pjcj joined
masak arnsholt: bootstrapping the knowhow to walk takes O(year) for a "clean-slate" human. 13:49
13:49 pjcj left
arnsholt Yeah, which illustrates that it's really hard 13:49
masak so arguably, it's not easy for us either ;) we just forget how hard it is.
arnsholt Yeah 13:50
I segued into riding a bike after talking about walking
It's something we learn later in life, so we remember how hard it was to learn, while also doing it sub-conciously
13:50 pjcj joined
arnsholt People have also mentioned driving a car, but I don't have a licence, so I don't know =) 13:51
13:51 pjcj left
masak arnsholt: hilarious scene from some SF novel: friendly alien is an 8-legged spider. it watches its human friend stop and remove a stone from the sole of his shoe, balancing on one leg as he does so. alien almost faints out of surprise at this feat. 13:52
13:52 kaleem left
arnsholt Nice one! 13:52
masak :) 13:53
13:53 bluescreen10 left, pjcj joined
ajr_ One vital lesson from the learning-to-walk process; you'll never make progress if you're afraid to make mistakes. 13:53
13:54 risou_awy joined
arnsholt Also an important theorem of machine learning theory: Most interesting classes of problems can't be learned accurately solely from positive examples 13:54
lizmat and one wonders how much mind you need to learn to walk:
www.extremetech.com/extreme/101525-...wer-source
ajr_ I used to use that as a metaphor at the start of training courses.
13:54 skids joined
ajr_ Adults are afraid of making mistakes, which inhibits their learning processes. 13:55
13:56 risou_awy is now known as risou
masak yeah. adults are weird that way. very risk-averse, at least when it comes to being wrong. 13:58
masak .oO( adults are obsolete children )
lizmat aren't we all pretending to be adults ? 14:01
14:05 ggoebel left
moritz lizmat: part of me pretends to be adult, parts are (and some parts don't even try to pretend :-) 14:10
14:10 xilo left 14:11 ggoebel joined
lizmat so, investigating where Hash.clone lives, I've come to the conclusion that that is just really nqp::clone in disguise 14:12
jnthn: am I on the good track there? 14:13
14:13 pjcj left 14:15 pjcj joined
moritz lizmat: sounds plausible 14:15
lizmat which in turn seems to be just a wrapper around pir::clone__PP, at least in the parrot version of rakudo 14:16
14:17 bluescreen10 joined
arnsholt Sounds plausible as well 14:20
lizmat because of this behaviour of clone, we cannot create any classes that return typed Hashes 14:22
unless you tell the users to bind to the return value of .new, rather than just assign
arnsholt Some limitation in what pir::clone does?
lizmat I assume that nqp::clone doesn't know about typed hashes, that they only live in Perl 6 land proper 14:23
TimToady dagurval: to hide parental multis you can just define your own proto 14:27
lizmat hmmm… I thought I could write a method clone for typed hashes, assuming that %i = %h would be syntactic sugar for %i = %h.clone 14:28
TimToady we need to find a better name than "typed hashes", because they're all typed, both in key and value; it's just that some are more typed than others 14:29
lizmat but it isn't: I guess it compiles into a direct call to nqp::clone for performance reasone
KeyConstraint and ValueConstraint ?
TimToady well, that's a bit of a mouthful, but I don't have a better idea yet 14:30
rjbs Type-Refined Hashes, or "trashes"
14:31 xilo joined
TimToady rjbs: again, that also applies to my Thing %hash; 14:31
lizmat TimToady: is there a way for a sub to specify that the "return" values should be bound, rather than returned?
I was suggesting earlier something like "sub a ( $a --> Foo is ref ) { my %h } 14:32
TimToady that's what "Parcel" means
lizmat "my %i= a" would then actually be "my %i := a"
TimToady except I don't think we should do that to assignment 14:33
lizmat oooh, wow 14:34
TimToady but my arguer hasn't woke up yet...
masak lizmat: re being an adult: twitter.com/dresdencodak/status/33...7589962754 14:35
14:35 MrMeek joined
lizmat alas, for a moment I thought that (my %h,) would be a fix for binding 14:37
14:37 spider-mario left
lizmat would be a way to bind the %h to the outer %i 14:37
14:38 crab2313 joined
GlitchMr .u ẗ 14:38
yoleaux U+1E97 LATIN SMALL LETTER T WITH DIAERESIS [Ll] (ẗ)
GlitchMr .u n̈
yoleaux U+006E LATIN SMALL LETTER N [Ll] (n)
U+0308 COMBINING DIAERESIS [Mn] (◌̈)
diakopter .u .u
yoleaux U+002E FULL STOP [Po] (.)
U+0075 LATIN SMALL LETTER U [Ll] (u) 14:39
diakopter .u yoleaux
yoleaux No characters found
masak a good read: www.yosefk.com/blog/what-worse-is-b...about.html -- speaking of "Worse Is Better". 14:43
relating that to Perl 6, I see the design process of Perl 6 as starting out in a The Right Thing kind of place, and then trying to mix as much evolution into the thing as possible at each stage. 14:44
and maybe Pugs partly taught us that, that you need the running implementation and the spectests and the feedback. I'm actually not sure, because I wasn't active enough before Pugs. 14:45
TimToady I believe in Intelligent Design, except I happen to think that evolution is the most intelligent way to do that :) 14:46
14:46 tgt left 14:47 gaussblurinc joined 14:48 gaussblurinc1 joined
masak that sentence ended way better than it started. :P 14:49
14:51 gaussblurinc left
TimToady well, rhetorically speaking, it was designed to jerk you around, so I guess it worked. 14:52
For all its faults, I like how the universe is put together, and many of the "improvements" people would make are not, in fact, better than what we have. 14:55
gtodd evo still provokes awe and wonder for me ... ever since I ate that apple in that garden ...
tadzik according to evolution a monkey ate that apple, no? ;) 14:56
TimToady well, an ape 14:57
gtodd recalls it was a pomme de terre or ground apple ... they'll grow anywhere feed/free the poor ... they're like manna from heaven :P
TimToady and we're still aping them
tadzik . o O ( not enough manna )
TimToady I like the theory that it was a watermelon
but regardless of how literally/figuratively you take that story, it expresses something profound: when the ape became infected with memes ("the knowledge of good and evil"), we could no longer remain innocent like animals, but are cursed to keep thinking about our problems in memespace regardless of their immediacy. 15:05
animals can live in the moment; when people do this, we call them "animals" 15:06
well, that's a bit misstated; we can live in the moment too, but we also transcend that mode of operation 15:07
occasionally
masak aye. though we're still busy creating more problems (environmentally and ecologically) than we're solving, as we're cursed to do according to that story. 15:09
15:09 kaare__ joined
masak overall, humanity hasn't been a net good for Life on earth so far. 15:09
TimToady well, but we might transport some portion of it to other planets eventually, which will be good in another way 15:10
15:10 fgomez joined
TimToady assuming there isn't native flora and fauna to wipe out 15:10
gfldex
.oO(if animals are innocent, why do cats avoid going to hell 7 times?)
 15:11
TimToady anyway, doing the Right Thing only works in memespace, and computer languages also exist in their own kind of meatspace, so both design and evolution are needful 15:12
moritz gfldex: because they can :-)
masak here's to populating other planets. or other places in general. putting all the humans in the same gravity well is a frightfully irresponsible case of monoculture. 15:13
gtodd the ape was just a $variable and 'lo it became a %HoH
moritz masak: ..or the only responsible thing to do, depending on how you see it :-)
gtodd perl6 ... ready for mars 15:14
moritz compiling rakudo probably requires more computing powers than the first space missions had available 15:15
Ulti masak: don't worry give it enough time and humanity will fragment into myriad species populating all the niches
TimToady well, depends on whether you include ground systems, I suppose, but even then, it depends on how you define "first"
gfldex moritz: Sputnik don't got a computer nor did the folk who build it. So yes, you are right. :) 15:17
TimToady is it a "mission" without people on it?
if one were trying to keep the mission metaphor viable, it's not really a mission till you go far away, so arguable the first mission was to the moon, since LEO is so...well, low... 15:19
15:19 tgt joined
moritz OK, include the spectest run 15:19
TimToady or maybe it's not a mission till we get out of the earth's gravity well, or maybe the sun's...
moritz then rakudo needs more computing power than the first manned earth mission, or so :-)
TimToady I suppose it depends on whether you include the compute power of wetware 15:20
moritz I don't.
gfldex it's going to be tricky to have a mission 13.7Bly away. Esp. since the gravity well of the sun grows with the speed of light. technically 15:21
TimToady well, there is that
and if you define it as "sphere of influence", we already got to the moon's gravity well 15:22
15:22 btyler left 15:23 kbenson joined, kbenson left
masak surely there must be some decent definition of "gettin gout of something's gravity well" that isn't limited by the fact that gravitation gravitational influence never really goes to 0. 15:24
TimToady well, that's what "escape velocity" is going after, really
gfldex apollo computer: "The computer had 2048 words of erasable magnetic core memory and 36 kilowords of read-only core rope memory. Both had cycle times of 11.72 micro-seconds. The memory word length was 16 bits: 15 bits of data and 1 odd-parity bit. The CPU-internal 16-bit word format was 14 bits of data, 1 overflow bit, and 1 sign bit (ones' complement representation)." 15:25
TimToady and in the absence of escape velocity, an adequate means of acceleration 15:26
15:26 btyler joined
TimToady gfldex: most of the compute power was in ground systems, sure :) 15:26
that's still the case
gfldex i have seen laptops in the hand of kosmonauts on the ISS 15:27
masak I mean, we can easily talk about the "size" of atoms, even though it's all really clouds of probability. stars and planets are more concrete than that.
TimToady in fact, that case paid for much of the early development of Perl
gfldex and singing, but no dancinge, likely due to a lack of a floor
TimToady masak: we can easily talk about a lot of things we don't really understand :) 15:28
gfldex hence this channel :-> 15:29
masak heh.
TimToady That's the lovely thing about memes: there's so many to choose from.
15:30 kaare__ left
masak well, in some sense, that's what "separation of concerns" and "encapsulation" are all about. being able to focus on one thing without the interference of other things. 15:30
TimToady has so far been unsuccessful separating the concern of reading the backlog this morning, unless you count temporal separation... 15:31
so stop saying interesting memes for a bit... :)
15:33 kaare_ joined
lizmat away for cycling and dinner 15:33
jnthn evening, #perl6 15:37
colomon \o 15:40
moritz \o jnthn, colomon 15:42
colomon o/ moritz
masak jnthn! \o/
15:44 kaare_ left 15:45 kaare_ joined
tadzik hello hello 15:48
masak tadzik! \o/ 15:53
TimToady okay, you may talk about interesting things again 15:54
tadzik what a day
GlitchMr Just wondering, why forward gotos need to be more verbose than backward gotos? Backwards gotos are just bad, but sometimes forward gotos are good. 15:55
When I have forward goto, I have to type goto "label".
I don't have to do this with backward goto.
TimToady because for backward goto the label is already declared as a term 15:56
std: foo: say foo 42; sub foo {} 15:57
camelia std 9906f18: OUTPUT«===SORRY!===␤Two terms in a row (listop with args requires whitespace or parens) at /tmp/AeEN6WHYI6 line 1:␤------> foo: say foo ⏏42; sub foo {}␤ expecting any of:␤ infix or meta-infix␤ infixed function␤ statement modifier
..loop…
GlitchMr Still, it seems backwards. Forward gotos are way more useful than backward gotos (where while {} could be easily used).
TimToady that error is LTA
jnthn Rakudo solves the whole thing by not implementing goto in any direction. :P 15:58
TimToady are you suggesting goto should be a reserved word, and not just a function? :)
jnthn We really should do the loop labels though. Just...tuits...
TimToady std: constant foo = 42; say foo 43; 15:59
camelia std 9906f18: OUTPUT«===SORRY!===␤Two terms in a row (listop with args requires whitespace or parens) at /tmp/_TcTUDpbqY line 1:␤------> constant foo = 42; say foo ⏏43;␤ expecting any of:␤ infix or meta-infix␤ infixed function␤ statement modifier
..l…
nwc10 Rakudo - purer than PHP
TimToady same LTAness
it certainly shouldn't recommend whitespace when there *is* whitespace
but the error should really know that foo is a term 16:00
nr: constant foo = 42; say foo 43; 16:03
camelia rakudo 8d2ec9: OUTPUT«===SORRY!===␤Two terms in a row␤at /tmp/Cl4fp2919d:1␤------> constant foo = 42; say foo ⏏43;␤ expecting any of:␤ postfix␤ infix stopper␤ infix or meta-infix␤ statement end␤ statement modifie…
..niecza v24-51-g009f999: OUTPUT«===SORRY!===␤␤Two terms in a row (listop with args requires whitespace or parens) at /tmp/2vJECKeAWy line 1:␤------> constant foo = 42; say foo ⏏43;␤␤Parse failed␤␤»
TimToady something is setting a bad listop memo for the error message
but why would something parsed as a term be calling <args>? and that's the only place (other than reduce) that sets it 16:04
std: rand 42; 16:06
camelia std 9906f18: OUTPUT«===SORRY!===␤Unsupported use of rand(N); in Perl 6 please use N.rand or (1..N).pick at /tmp/MayKfot5GK line 1:␤------> rand⏏ 42;␤Parse failed␤FAILED 00:00 41m␤»
TimToady std: pi 42; 16:09
camelia std 9906f18: OUTPUT«===SORRY!===␤Two terms in a row (listop with args requires whitespace or parens) at /tmp/6QDYFdmcAQ line 1:␤------> pi ⏏42;␤ expecting any of:␤ infix or meta-infix␤ infixed function␤ statement modifier loop␤Parse failed␤FAILED
..…
TimToady same issue
16:15 gdey left
dalek d: 52b6353 | larry++ | STD.pm6:
don't set listop memo on terms
 16:17
16:26 ajr_ left 16:29 woosley1 joined, woosley1 left 16:42 tgt left 16:44 SamuraiJack__ left 16:45 SamuraiJack__ joined 16:46 census left 16:48 gdey joined
dagurval TimToady: like this? 17:12
r: class A { multi method new($x) { self.bless(*, { }); proto method new() { * } } }; A.new
camelia rakudo 8d2ec9: OUTPUT«Cannot call 'new'; none of these signatures match:␤:(A : $x, Mu *%_)␤ in method new at /tmp/5TCx8vwhba:1␤ in block at /tmp/5TCx8vwhba:1␤␤»
17:12 berekuk left 17:13 berekuk joined
jnthn Just like that. 17:15
dagurval nevermind, that was wrong :). 17:16
jnthn It did what I expected :P
17:17 dakkar left
jnthn dinner, bbl 17:17
dagurval r: class A { multi method new($x) { self.bless(*, { }); }; proto method new() {*} }; A.new
camelia rakudo 8d2ec9: OUTPUT«Cannot call 'new'; none of these signatures match:␤:(A : $x, Mu *%_)␤ in method new at /tmp/YSF8GJ5jr_:1␤ in block at /tmp/YSF8GJ5jr_:1␤␤»
dagurval I'm trying to figure out this: 16:27 < TimToady> dagurval: to hide parental multis you can just define your own proto
:) 17:18
masak dagurval: visually at least, it's weird to me that you'd introduce a proto after the multi. 17:25
dagurval: because the proto sort of pulls all its multis together, or contains them, or whatever. 17:26
actually I'm surprised that's not a compile error.
dagurval: also, as a rule, all the signatures of the multis should match (be a subset of) the signature of the proto. 17:27
r: class A { proto method new($) {*}; multi method new($x) { say "OH HAI"; self.bless(*) } }; say A.new ~~ A 17:28
camelia rakudo 8d2ec9: OUTPUT«Not enough positional parameters passed; got 1 but expected 2␤ in method new at /tmp/oFf5yAkYF_:1␤ in block at /tmp/oFf5yAkYF_:1␤␤»
masak r: class A { proto method new($) {*}; multi method new($x) { say "OH HAI $x"; self.bless(*) } }; say A.new("dagurval") ~~ A
camelia rakudo 8d2ec9: OUTPUT«OH HAI dagurval␤True␤»
masak \o/
dagurval yay! \o/
17:29 dmol joined
masak so the proto says "I handle routines of this name, and they all have this kind of signature". multis say "I answer to calls to the routine when the signature looks like... this!" 17:29
PerlJam r: class C { proto method foo {}; proto method foo {}; };
camelia rakudo 8d2ec9: OUTPUT«===SORRY!===␤Package 'C' already has a method 'foo' (did you mean to declare a multi-method?)␤at /tmp/my1wqf2Itd:1␤------> ␤»
PerlJam r: class C { multi method foo {}; proto method foo []; } 17:30
camelia rakudo 8d2ec9: OUTPUT«===SORRY!===␤Missing block␤at /tmp/EBbo1U_SXS:1␤------> { multi method foo {}; proto method foo ⏏[]; }␤ expecting any of:␤ statement list␤ horizontal whitespace␤ postfix␤ statement end␤ stat…
PerlJam r: class C { multi method foo {}; proto method foo {}; }
camelia rakudo 8d2ec9: ( no output )
PerlJam Hmm. that is weird.
masak yeah, protos and onlys are the same "kind" of thing. and multis are subordinate to their proto, whether the proto is explicit or not.
dagurval r: class A is DateTime { proto method new($) {*} }; A.new 17:31
camelia rakudo 8d2ec9: OUTPUT«Not enough positional parameters passed; got 1 but expected 2␤ in method new at /tmp/9tdoeoTLEh:1␤ in block at /tmp/9tdoeoTLEh:1␤␤»
dagurval r: class A is DateTime { proto method new($) {*} }; DateTime.new
camelia rakudo 8d2ec9: ( no output )
dagurval sweet
r: class A is DateTime { proto method new($) {*} }; A.new(now)
camelia rakudo 8d2ec9: OUTPUT«Null PMC access in get_pmc_keyed()␤ in method new at /tmp/unWQf3YSyy:1␤ in block at /tmp/unWQf3YSyy:1␤␤»
dagurval ok, I still don't completely get it, but thanks masak. I'll wrap my head around it. 17:32
that looks like implementation leak though
gfldex Null PMC is parroty way to say segfault 17:35
masak it's basically always wrong, unless you're doing weird MOP stuff.
masak submits rakudobug 17:36
17:36 gaussblurinc1 left
arnsholt I'd go so far as saying a Null PMC error is always a bug, unless you happen to know it's not =) 17:37
dagurval r: use Test; class A { multi method new($x) { } }; class B { method new($x) { } }; dies_ok { A.new() }; dies_ok { B.new(); } 17:38
camelia rakudo 8d2ec9: OUTPUT«not ok 1 - ␤ok 2 - ␤»
dagurval all I want is for A to not have implicit new()
(but still have multi) 17:39
17:39 spider-mario joined
masak r: class A is Any { proto method new($) {*} }; A.new(now) 17:39
camelia rakudo 8d2ec9: OUTPUT«Null PMC access in get_pmc_keyed()␤ in method new at /tmp/6ggP2nQoNi:1␤ in block at /tmp/6ggP2nQoNi:1␤␤»
masak ok, not related to DateTime. phew. 17:40
dagurval :)
masak rt.perl.org/rt3/Ticket/Display.html?id=118067 17:42
17:44 Chillance joined
dagurval r: proto sub a($) {*}; sub a(Int $x) { "yay" } 17:44
camelia rakudo 8d2ec9: OUTPUT«===SORRY!===␤Redeclaration of routine a␤at /tmp/be1qqc_4fW:1␤------> to sub a($) {*}; sub a(Int $x) { "yay" }⏏<EOL>␤ expecting any of:␤ postfix␤ statement end␤ statement modifier␤ statement modifie…
dagurval shouldn't that work? from syn: ". A proto also adds an implicit multi to all routines of the same short name within its scope, unless they have an explicit modifier" 17:45
masak submits rakudobug for that 17:51
moritz I think that's already (long) known 17:52
masak yes, but I don't find it in RT. 17:53
submitting Justin Case, and then we can merge if someone finds it.
17:55 btyler left, risou is now known as risou_awy, risou_awy is now known as risou
dagurval masak++ 17:57
dalek rl6-roast-data: acd6efa | coke++ | / (4 files):
today (automated commit)
 17:58
17:58 cognominal__ left 17:59 cognominal__ joined
[Coke] rakudo still exploding. 17:59
18:00 sciurius joined
dagurval masak: I saw your talk(s) about perl6 on this years FOSDEM. They were inspiring 18:00
[Coke] lizmat++ # only the exploding test is failing now.
18:02 SmokeMachine left, tgt joined 18:03 japhb_ joined, SmokeMachine joined
GlitchMr planetsix.perl.org/ 18:08
oh, wow, I just have noticed issue with my RSS
but... huh... it wasn't broken before 18:09
Let me guess, the guys at GitHub have broken GitHub Pages again... 18:10
...
18:12 btyler joined 18:13 risou is now known as risou_awy
GlitchMr Jekyll v1.0.0, I see... 18:14
18:26 crab2313 left
jnthn Note that the order of proto method and multi doesn't matter so much since it doesn't put all the pieces together until compose time. 18:32
18:33 REPLeffect joined
PerlJam jnthn: except that the language in the specs refers to the order. "A C<proto> always functions as a dispatcher around any C<multi>s declared after it in the same scope" (for instance) 18:33
18:34 SamuraiJack joined, census joined, SamuraiJack__ left
jnthn I suspect that language is in S06 and primarily thinking about subs, since "scope" tends to be about lexical scope and that's not really relevant for methods. 18:35
PerlJam Sorry, that was quoted from S12:1093
jnthn The reason we don't put all the bits together until compose time is 'cus you don't know what you're gonna be getting from a class.
Ah. 18:36
Then I'm gonna declare it over-specified. :P
PerlJam wfm.
jnthn Thing is that the proto is often auto-generated.
But we don't know if we need to do that until we've seen what the roles want to bring in. 18:37
And role comp certainly happens at compose time.
Hm, I guess I can make it an error to write a proto after you wrote a multi in a class... 18:38
...by seeing if there's a non-multi already sat in the method table.
But it'd be an error for the skae of wanting it to be one rather than out if needing to be one. 18:39
*sake
(Whereas with lexical multis it kinda has to be one.) 18:40
timotimo hm, how does one properly make a custom short-circuiting operator? will the S metaop help? otherwise i suppose one would build a macro for it 18:44
moritz macros, if it works 18:45
S won't work
pmichaud good morning, #perl6 18:46
timotimo i don't really understand how it'd achieve short-circuiting, but in the spec it says »The 'S' can be thought of as standing for Sequential, Serial, Synchronous, Short-circuit, Single-thread, and Safe.«
moritz \o pmichaud
PerlJam r: role R { multi foo { say "R-foo" } }; class C does R { method foo { say "C-foo" } }; C.new.foo;
camelia rakudo 8d2ec9: OUTPUT«C-foo␤»
pmichaud modelscoutrobotics.org/naopen-win/ # a successful weekend :)
PerlJam r: role R { multi method foo { say "R-foo" } }; class C does R { method foo { say "C-foo" } }; C.new.foo; 18:47
camelia rakudo 8d2ec9: OUTPUT«===SORRY!===␤Cannot have a multi candidate for 'foo' when an only method is also in the package 'C'␤»
moritz pmichaud: congratulations
masak pmichaud++ pmichaud.kids++ 18:49
PerlJam moritz: But the kids did all the work! :)
pmichaud they did work hard, yes. :)
timotimo masak: shouldn't it be pmichaud.kids>>++ or something?
;)
jnthn pmichaud: Congrats! 18:50
masak timotimo: oh, absolutely, unless .kids have a special postfix:<++> defined on them.
which they probably don't.
18:50 zby_home_ joined, kivutar left
jnthn I dunno, they seem pretty successorful... 18:51
moritz :-)
masak jnthn: that pun .succ
18:51 s1n left
moritz 's 18:51
tadzik :D
timotimo m)
masak moritz: I'm torn. "succ" is short for "successor", so at least in the original context, the second "c" already has the "s" sound.
moritz: but yeah, I did consider adding that "s". 18:52
lizmat .hash( :type(Any, :of(Int) ) now for perusal at github.com/rakudo/rakudo/pull/155
masak lizmat.rock's :P
lizmat it needed string eval to work :-(
masak o.O
sacrilege! 18:53
lizmat eh, if life serves you lemons, you better make lemonade :-)
masak looks for the torches and pitchforks...
pmichaud can I make a lemon meringue pie, istead?
*instead? 18:54
18:54 ggoebel left
lizmat needs additional ingredients 18:54
but yes, sure
masak lizmat: looks good, apart from the string eval ;) 18:55
lizmat seriously though, there is to my knowledge currently no way of finding out what "type" a hash is
masak anyone mind if I merge that?
lizmat: I think we should take this pull request, and then find a better (spec'd) way to do it ASAP.
lizmat fine by me… :-)
jnthn could add a way to give the type... :)
lizmat I can do that, I think myself... 18:56
18:56 kivutar joined
lizmat if I knew how to get at the info, somewhere hidden in NQP's bowel, afaics 18:56
but jnthn, don't let me stop you! 18:57
masak lizmat: you've already spectust, of course?
lizmat yes
jnthn lizmat: Isn't it just the type arg?
lizmat I have
dalek kudo/nom: 61db0e0 | (Elizabeth Mattijsen)++ | src/core/Any.pm:
.hash now accepts optional :type and :of named parameters

So that we potentially can roundtrip .perl output, provided it can be generated properly
kudo/nom: 14e7d74 | (Carl Mäsak)++ | src/core/Any.pm:
Merge pull request #155 from lizmat/hashwithtypeandof

  .hash now accepts optional :type and :of named parameters
lizmat jnthn: wat?
jnthn method ofkey() { TKey } or so in the TypedHash role?
masak 'ofkey'! 18:58
18:58 ggoebel joined
masak that's what it's called! 18:58
jnthn heh, well, the value one is .of :)
masak right.
jnthn I just strappolated :)
masak 'ofcourse' :P
lizmat $ ack -a ofkey
lizypro:myrakudo liz$
that string is not anywhere in the repo 18:59
jnthn lizmat: Yes, I was suggesting adding it :)
masak .oO( the music your Hash produces sounds a little 'ofkey' )
lizmat ah, ok
masak jnthn: oh!
yeah, I didn't get that either.
pmichaud ofkey sounds reasonable to me
PerlJam lizmat: looks like you need to upgrade your ack too ;)
jnthn Sorry, I'm trying to wrajt slajds :P
masak that's OK.
lizmat personally, I think they "keyof" and "valueof" feel better 19:00
or "valof"
19:01 rking1 is now known as rking
jnthn Well, .of already exists 19:01
r: my Int %h; say %h.of
camelia rakudo 8d2ec9: OUTPUT«(Int)␤»
lizmat yes, I know :-)
jnthn And matches the trait name. 19:02
pmichaud and we also use "atkey" and "atpos"
lizmat ok, you got me there :-)
pmichaud so "ofkey" seems to match the pattern better.
timotimo hmm. i wonder how hard it could be to implement the ; operator for lists (lists of lists, more precisely?)
19:03 rking left
lizmat fwiw, I'm wondering why typed hashes aren't implemented as a subclass, rather than as a role 19:03
PerlJam It's too bad you can't ask the hash to give you a prototypical key that you can just call .of on.
jnthn lizmat: 'cus only roles are type parametric.
lizmat you mean, they can MMD ? 19:04
jnthn No, I mean they can have type parameters.
role Foo[::T] { }
You can't have a class Foo[::T] { }
lizmat I guess I need to brush up on roles 19:05
19:05 s1n joined
jnthn I probably wrote a bunsj of slides about them at some point :) 19:05
Talking of which...
jnthn hides IRC so he can try to get his $dayjob stuff rescued 19:06
lizmat masak: and c.s. but just for the record, if I .clone a Typed hash, the clone should ofkeyed and ofed in the same way, right? 19:07
masak lizmat: right.
lizmat well, it isn't right now, so that's a rakudobug, I guess
masak thought you filed that before. 19:08
also, "c.s."?
lizmat cum suis
masak ah.
pmichaud I'm afk, preparing to board flight
bbt
masak lizmat: hah, a ddg search for the concept turns up a Dutch Wikipedia page: "Lijst van Latijnse spreekwoorden en uitdrukkingen" :P 19:09
so I guess it's something used mostly in the Netherlands.
lizmat r: my Int %h{Any}; say %h.of; say %h.WHAT; my %i= %h; say %i.of; say %i.WHAT # should this also copy .ofkey and .of? 19:10
camelia rakudo 8d2ec9: OUTPUT«(Int)␤(Hash+{TypedHash})␤(Mu)␤(Hash)␤»
masak (also, ddg made the initial search for just "suis" and asked if I wanted to turn off safe search. gee, computers are so stupid.)
arnsholt "cum suis" would be something along the lines of "with its/their own" 19:11
19:11 kivutar left
lizmat in my mind, "cum suis" means "and the ones with you" 19:12
translated to english, from the Dutch explanation: with supporters, the people who usually with him or her 19:13
masak it's the same "suis" as in "sui generis" -- "of their own kind".
PerlJam (or you could have just said exactly what you meant in english and avoided the length explanation of the latin ;-) 19:14
lizmat probably, my latin is not *that* good. It's a standard Dutch idiom, for some reason
huffman: cs is shorter
:-)
arnsholt suus is very generic, which makes it hard to translate in a fragment like this 19:15
It can be either be the third person possessive (singular or plural) or reflexive
lizmat r: my Int @a; say @a.of; my @b=@a; say @b.of # I guess just copying also loses .of 19:16
camelia rakudo 14e7d7: OUTPUT«(Int)␤(Mu)␤»
masak arnsholt: oh, just like 'sig' in Swedish (and I guess Norwegian too).
probably 'sej' or something in Norwegian.
lizmat r: my Int @a; say @a.of; my @b=@a.clone; say @b.of # I guess just copying also loses .of
camelia rakudo 14e7d7: OUTPUT«(Int)␤(Mu)␤»
arnsholt "seg", actually =) 19:17
masak close enough :P
lizmat even cloning loses the .of. Is that correct or a rakudobug? feels wrong to me
arnsholt But yeah, same etymology
masak .oO( what, do I look like a Norwegian? ) :P
arnsholt sva in Sanskrit =)
masak najs.
PerlJam lizmat: looks like a bug to me.
lizmat what me, German? (about 50% actually :-) 19:18
masak Sanskrit always has the last laugh. (or first, actually)
Sanskrit is the Lisp of spoken languages. they always did everything first.
19:18 domidumont left
moritz are they also smug about it? :-) 19:18
masak exceedingly.
sorear so much of linguistic history has been lost though
masak "A written grammar? Yawn, we had that 3000 years ago." 19:19
lizmat hasn't seen any Sanskritians lately
masak maybe not quite that long ago.
arnsholt masak: Just 2000 and a bit, actually, but the point still stands =)
Not only a written grammar, but also a grammar that is a lot more formal than the Greco-Roman grammarians
masak 8th century BC, according to Wikipedia.
right, so the Lisp analogy is quite apt. 19:20
"we were there before you, and we did it better, blogdangit!"
arnsholt Oh, right. I keep forgetting that there's a grammatical tradition prior to Panini
masak the guy who made flat breads with salad in them?
arnsholt Exactly! =D 19:21
masak .oO( nono, Panini the violinist )
lizmat wasn't Panini the guy who invented the single pan dish ? 19:22
timotimo that was Panuno.
huf guy? it's just a config file
masak you're getting the hang of it now :) 19:23
19:23 SamuraiJack_ joined, SamuraiJack_ left
masak "no no, 'Panini' means "a small pain", like in the lower back" 19:23
19:25 SamuraiJack_ joined
lizmat wasn't "pa" sanskrit for "knight" ? 19:26
19:26 SmokeMachine left
arnsholt Not AFAIK =) 19:26
masak lizmat: hahaha 19:27
lizmat wonders how many people got the Monty Python reference
masak
arnsholt Not me, sadly =)
tadzik ni!
arnsholt Oh, right. I'm bad at puns =) 19:28
masak I got it on the second read. felt a bit like being SQL injected :P 19:29
PerlJam masak: did you suffer a flesh wound in the process?
lizmat for the MP deprived: en.wikipedia.org/wiki/Knights_who_say_Ni
masak should re-watch "The Holy Grail" 19:31
19:34 SamuraiJack left 19:35 SamuraiJack joined
lizmat after the latest mail on the YAPC mailing list, I'm seriously considering not attending the conference, but instead hack on Perl 6 while in Austin 19:40
*sigh* 19:41
lizmat hopes coc mail was a fake 19:43
PerlJam lizmat: why exactly? Because you might run into rurban?
arnsholt I intend to hack on Perl 6 while attending the conference
lizmat I know Reini personally, I'm not afraid to run into him
PerlJam lizmat: then what do you save yourself by not attending? 19:44
arnsholt Find a seat next to me and I can provide you with an infinite supply of useless historical linguistics trivia as well
lizmat Reini has his faults, and actually many, but to get thrown from the mailing list like that for that
arnsholt: I just might :-)
PerlJam I'd guess his removal is probably temporary 19:45
lizmat it's the principle: the YAPC mailing list is *not* just for the YAPC::NA afaik 19:46
19:46 SmokeMachine joined
lizmat same as the #yapc channel 19:46
and the conference hasn't even started yet
19:47 bluescreen100 joined
masak I also want to publicly announce that my strategy at YAPC::NA will be to explicitly *avoid* confrontations of any kind. I'm not interested in drama. I will leave a situation rather than be a participant or spectator of it. 19:47
if I can, I will find lizmat and arnsholt and discuss linguistics with them ;)
lizmat masak: good plan, let's hack!
PerlJam masak: so ... no hugs? ;)
masak and hack, of course.
PerlJam: no unannounced hugs.
lizmat only with mutual consent!!!!! 19:48
PerlJam <masak> I wish to hug you, do you consent to this hug?
lizmat hugme: hug masak
hugme hugs masak
lizmat oops, that was without consent 19:49
the motto of a good friend of mine; "arms are made
for hugging"
19:50 bluescreen10 left
PerlJam "Breaking news ... pandemonium broke out in Austin today at a conference for Perl geeks when a series of unannounced hugs led to a large hug-pile that sent 4 people to the hospital." 19:50
.oO( panda-monium has some sixiness )
 19:51
lizmat "a spokespersons for the huggies said that all hugs were given with consent" 19:52
masak lizmat: following Postel's Law, I accept hugs liberally, but emit hugs conservatively ;)
19:54 sftp left
diakopter yeah my mail was rejected 19:54
removing him didn't need invoking the coc at all... just normal mailing list administration. also, the coc doesn't mention anything about removal from mailing lists as a consequence.
19:54 sftp joined
diakopter "" 19:54
PerlJam diakopter: um ... "any action they deem appropriate" is fairly broad. 19:56
diakopter oh
lizmat this just in: Perl will have a booth and a devroom at the Froscon!
www.froscon.de/en/home/ 19:57
dalek kudo-star-daily: ccb6549 | coke++ | log/ (5 files):
today (automated commit)
[Coke] only real star module change is that panda dropped half its tests. 19:58
(but still passes them all)
tadzik *relief* 20:00
[Coke]++ for those tests
labster good postnoon, #perl6 20:01
The drama on the yapc list looks pretty minor compared to what I've experienced in student government conferences. 20:02
lizmat hugme: hug labster
hugme hugs labster
labster is hugged
PerlJam lizmat: A liberal reading of #2 in www.yapcna.org/yn2013/code-of-conduct.html could consider #yapc and the yapc mailing list be "conference-related social events"
20:03 adu joined
labster hi lizmat 20:03
lizmat hi labster! 20:04
PerlJam: topic in #yapc is "Topic is UNOFFICIAL channel for Various YAPCs"
20:05 prevost joined
lizmat PerlJam: that is *very* liberal 20:06
PerlJam aye. the "UNOFFICIAL" was added in direct response to that other CoC fiasco. Still doesn't mean that someone won't consider #yapc a "conference-related social event"
labster r: my ($a, $c); ($a, $, $c) = 1..3; 20:09
camelia rakudo 14e7d7: OUTPUT«===SORRY!===␤Unsupported use of $, variable; in Perl 6 please use .join() method␤at /tmp/OlBWFcsobt:1␤------> my ($a, $c); ($a, $,⏏ $c) = 1..3;␤»
20:09 SamuraiJack left
labster perlcabal.org/syn/S02.html#Names says that line 'will result in the message, "Anonymous variable requires declarator".' 20:09
TimToady std: foo: foo 42 20:10
camelia std 52b6353: OUTPUT«===SORRY!===␤Two terms in a row at /tmp/TNTRd1GJTA line 1:␤------> foo: foo ⏏42␤ expecting any of:␤ infix or meta-infix␤ infixed function␤ statement modifier loop␤Parse failed␤FAILED 00:00 42m␤»
labster r: my ($a, $c); ($a, $ , $c) = 1..3; #added space
camelia rakudo 14e7d7: OUTPUT«===SORRY!===␤Non-declarative sigil is missing its name␤at /tmp/eFX9NIpWzs:1␤------> my ($a, $c); ($a, $⏏ , $c) = 1..3; #added space␤ expecting any of:␤ postfix␤ infix stopper␤ infix or meta-infix␤ …
masak labster: seems the error message is shadowed by rakudo thinking it's the $, variable
...or another error message is. :) 20:11
labster so, theoretically a rakudobug, but do we really want exact wording for error messages defined by the spec?
20:12 kaare_ left
masak I think it's a rakudobug that it considers ($a, $, $c) as being the '$,' variable. 20:15
PerlJam labster: The message in the spec at least points them in the direction that anonymous vars can only be used in declarators
masak but I don't know what to do about it, really.
labster Well really, the only place you should ever write $, is in a subroutine signature. 20:18
lizmat just got caught for the nth time doing a ternary with || rather than !! 20:19
PerlJam labster: You don't think "my ($a,$,$c) = $string.split(/\t/);" would ever happen? 20:20
20:20 sftp_ joined
labster oh... right 20:20
I'm not thinking straight.
lizmat r:my ( $a, Mu, $c )= <a b c>; say $a; say $c 20:21
r: my ( $a, Mu, $c )= <a b c>; say $a; say $c
camelia rakudo 14e7d7: OUTPUT«a␤c␤»
20:21 kivutar joined, sftp left
labster r: my ($a,$,$c) = "a b c".split(/\s/); 20:21
camelia rakudo 14e7d7: ( no output )
labster hm 20:22
dalek d: 8850393 | larry++ | STD.pm6:
only carp about P5 $, variable where it's obvious
 20:27
masak TimToady++ 20:34
labster okay, shall i submit a proper rakudobug then? 20:35
PerlJam labster: can't hurt
jnthn plz reference the STD commit TimToady++ just made too, so the fixer knows where to steal a fix :) 20:36
lizmat is feeling dense 20:37
given a hash %h, I want to pass that to a sub and have the contents of the hash be interpreted as named parameters 20:38
:%h is close, but not really what I want :-)
timotimo r: my %h = a=>1, b=>2; sub foo(:$a, :$b) { say "$a, $b" }; foo(\%h); 20:39
camelia rakudo 14e7d7: OUTPUT«Too many positional parameters passed; got 1 but expected 0␤ in sub foo at /tmp/j2LMZcc6XS:1␤ in block at /tmp/j2LMZcc6XS:1␤␤»
masak r: sub foo(:$named1, :$named2) { say "$named1 $named2" }; my %h = named1 => "OH", named2 => "HAI"; foo(|%h)
camelia rakudo 14e7d7: OUTPUT«OH HAI␤»
labster submits rakudobug
lizmat | duh!
masak++
masak lizmat: prefix:<|> flattens arrays into positionals and hashes into nameds.
lizmat tests 2nd attempt at Hash.new accepting :type and :of parameters 20:40
TimToady: should something like .ofkey be specced ? 20:45
20:46 chrisdev_ joined
lizmat If we're going to use .ofkey as the method to get the type of the typed hash, then maybe the :type parameter to .hash() should me also called ".ofkey" ? 20:48
20:48 bruges left
lizmat *maybe 20:48
masak pending TimToady's answer to "should it be spec'd", my guess is on "yes" :)
TimToady that's a misuse of "of", I think 20:49
druther have something like .keytype
20:49 chrisdev left
TimToady or if "of" is what you get from a hash, the other is the "for" type or some such 20:49
masak .keytype is the best I've heard so far. 20:50
.for is cute, but less clear.
20:50 bruges joined, SamuraiJack_ left
TimToady but I think the original idea was that you'd get these types by instantiating a role, and that role just instantiates to Str,Any by default 20:50
doing it by attaching info to the constructor means you're always forced to do introspection to find out if two types are equal, and introspection is kinda smelly 20:51
with classes you can use named equivalence sometimes 20:52
lizmat so effectively you're saying they're *all* typed hashes 20:54
the fact that the underlying implementation is Str,Any with other keytypes and .ofs bolted on top
is now leaking out 20:55
TimToady I said they were all typed last night, iirc
when I carped about calling these "typed hashes" :) 20:56
unlike in P5, where both computer and user are confused about the types, in P6 we only want the user confused :)
lizmat hehe, mission accomplished ! 20:57
masak helpfully links to irclog.perlgeek.de/perl6/2013-05-20#i_7092388
FROGGS good evening 20:59
lizmat evening FROGGS!
masak FROGGS! \o/ 21:00
FROGGS :o) 21:01
21:03 zby_home_ left 21:05 skids left
labster yo FROGGS 21:05
FROGGS o/ 21:06
there are nice bur reports in my inbox
bug*
timotimo hey froggs :) 21:07
do you have an idea how to finish the ENTER/FIRST support?
FROGGS timotimo: no 21:08
timotimo: I guess my approach isnt wrong at all, I'm just not sure why it is failing
maybe that indirect lookup does something wrong or the variable is declared in the wrong place 21:09
timotimo what's the failure mode? i haven't tried it tbh
21:09 alester joined
FROGGS timotimo: there is a comment attached which shows the error 21:09
21:10 spider-mario left
timotimo ah, ok, let me see 21:12
was it a secret gist?
ah, here it is
lizmat TimToady: isn't S04:1639 contradicting S06:3298 ?? The reason I'm asking is that I want to change "sub hash(*@a, *%h) { my % = @a, %h }" to "sub hash(*@a, *%h) { @a.hash(|%h) }" to allow for setting type and .of 21:27
and the example at S04:1639 seems to make that impossible ( even though it is now implemented that way)
lizmat now running spectest with that change
colomon lizmat++ 21:29
21:30 adu left 21:32 kurahaupo joined 21:38 kurahaupo left 21:41 muixirt left, erkan left
lizmat r: sub a { say "Positional:"; say @_.perl; say "Named:"; say %_.perl }; a( "a" => 1 ); say "==="; a( a=>1 ) # un-autoquoted pairs become positional? 21:50
camelia rakudo 14e7d7: OUTPUT«Positional:␤Array.new("a" => 1)␤Named:␤().hash␤===␤Positional:␤Array.new()␤Named:␤("a" => 1).hash␤»
21:51 kurahaupo joined
timotimo rosettacode.org/wiki/Galton_box_animation#Perl_6 - i RosettaCode'd :) 21:52
jnthn lizmat: Yes.
lizmat ok, well I was surprised to see t/spec/S04-statement-parsing/hash.t test 4+5 succeed, when I had expected them to fail 21:53
which would make S04:1639 *not* contradicting S06:3298, as "a" => 1 is not a pair when specified as a parameter 21:54
which means that tests 6+7 in t/spec/S04-statement-parsing/hash.t are wrong according to spec 21:55
timotimo oh, that's curious
lizmat but right in the current implementation of hash()
timotimo is that to do with only allowing identifiers to be passed as nameds or something? 21:56
lizmat I have no idea… I'm just the messenger in this
:-)
21:56 xilo left 21:57 adu joined 22:03 lustlife left 22:07 dmol left 22:08 prevost left, bluescreen100 left
lizmat spectest complete, only t/spec/S04-statement-parsing/hash.t test 6+7 failed, as expected 22:09
Hash.new() and hash() accept the same typed parameters as .hash() , see github.com/rakudo/rakudo/pull/156 for your perusal 22:15
jnthn Does this break hash(a => 1, b => 2) ? 22:16
lizmat yes
jnthn I. Um. Don't think that'll fly...
lizmat but that is not according to spec
jnthn Hmm.
I'm too tired to go look.
lizmat it does not break hash( "a" => 1, "b" => 2 ) 22:17
jnthn It just feels wrong/surprising.
lizmat well, to an extent I agree with you
22:17 xilo joined
timotimo r: hash( :foo<bar> ).perl.say 22:17
camelia rakudo 14e7d7: OUTPUT«("foo" => "bar").hash␤»
timotimo will this work?
lizmat but that is the same as with Hash.new( a=>1 ) not working either
timotimo: no 22:18
timotimo mhm
lizmat r: Hash.new( :foo<bar> ).perl.say
camelia rakudo 14e7d7: OUTPUT«().hash␤»
lizmat it would make it consistent with Hash.new
timotimo sounds kind of sensible so far 22:19
lizmat in the current implementation, Hash.new and hash() are not doing the same thing
timotimo is that after your changes or before? 22:20
lizmat as far as I'm concerned, this could go either way
*before* my changes
if this pull request is merged, they would do the same
(as they both just call .hash)
timotimo mhm 22:22
i wonder how much existing code would be b0rked by the change?
i think it would be kind of surprising that autoquoted and non-autoquoted would be treated differently, but at the same time treating "of" and "key" or so differently would be bad, too 22:23
lizmat well, I was surprised by Hash.new( a => 1 ) not working, for sure
the argument in that case was, that named parameters should be passable so alter the behaviour of Hash.new 22:24
*to
timotimo right 22:25
lizmat and I think hash() is just short for Hash.new
timotimo the form (...).hash seems most sensible, tbh
lizmat well, I'll check further comments in the backlog tomorrow 22:26
goodnight #perl6!
timotimo good night :)
22:28 REPLeffect left
jnthn sleep & 22:30
22:31 skids joined
timotimo rosettacode.org/wiki/Galton_box_animation#Perl_6 - now with animated gif 22:50
masak nice. 22:51
22:51 kivutar left
masak 'night, #perl6 22:54
colomon \o
timotimo why is the stray frame still in there? :(
22:57 adu left
lue didn't notice a stray frame, even when knowing it was there 22:57
timotimo it has a different background color and a different placement of the pegs and such 22:58
i'm thoroughly confused
the file i selected on disk doesn't have the frame, then i try to upload it and i end up with the version that has the frame. i can even wget it back down and it will have the stray first frame in there 23:01
lue how different is the background color? (it's black the whole time for me) 23:05
timotimo #000000 vs #161819 23:06
also, the pegs are in different places
rosettacode.org/wiki/File:Galton_Box_Perl_6.png - this frame ended up in the beginning by accident 23:07
23:08 census left
lue I'm afraid I don't see that stray frame :/ 23:11
timotimo can you reproduce the stray frame in the first version of the image? 23:12
23:12 shinobicl joined
lue Ah, it's in the first uploaded picture, easy to spot :) [didn't notice it in the second or the current one] 23:14
timotimo what, weird!
when i point wget at the url, i definitely get the stray frame
ah, whatever, i'll just ignore it
lue fwiw I didn't use wget to download the picture from rosettacode 23:15
timotimo don't need to
lue (the stray frame appeared neither in the download of the current one nor in viewing the second or last version in browser)
timotimo thanks. my setup must be severely disturbed then 23:17
lue But it was there at some point, at least your setup's not *that* disturbed :)
23:29 btyler left 23:40 ehouse joined
ehouse does perl6 have to be installed in your homedir for panda to work? 23:46
I have it installed system wide and panda can't find .work 23:49
23:53 prevost joined 23:54 shinobicl left 23:58 PacoAir left

Powered by Cro and the Raku® Programming Language.

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

Thank you!