»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or rakudo:, std:, or /msg camelia p6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by masak on 12 May 2015.
00:01 HoppingMadMan joined 00:04 shinobi-cl_ joined
shinobi-cl_ Hi all... 00:04
smls_ Hi
00:04 smls_ is now known as smls
shinobi-cl_ how can i create an array filled with the same value over and over? example, an array with 100 Int with value zero 00:05
00:05 HoppingMadMan left
smls my @a = 0 xx 100; say @a.perl 00:05
m: my @a = 0 xx 100; say @a.perl
camelia rakudo-moar d8fa56: OUTPUT«[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …»
00:06 HoppingMadMan joined
shinobi-cl_ r: my @a = Nil xx 10; say @a.perl; 00:06
camelia rakudo-{moar,jvm} d8fa56: OUTPUT«[Any, Any, Any, Any, Any, Any, Any, Any, Any, Any]<>␤»
smls Note, however, that the expression on the left-hand-side of the xx operator is reevaluated each time 00:07
shinobi-cl_ Thanks smls!
smls m: say rand xx 10;
BenGoldberg m: say my @a = 0 x 100;
camelia rakudo-moar d8fa56: OUTPUT«0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000␤»
rakudo-moar d8fa56: OUTPUT«0.468089031244085 0.309870672209813 0.890902802842534 0.772970687409833 0.507221045662271 0.783112132530793 0.889978764459187 0.452996015354152 0.15961362780056 0.23165113780065␤»
BenGoldberg m: say my @a = (0,)0 x 100;
camelia rakudo-moar d8fa56: OUTPUT«5===SORRY!5=== Error while compiling /tmp/Pd0seky9LJ␤Two terms in a row␤at /tmp/Pd0seky9LJ:1␤------> 3say my @a = (0,)7⏏050 x 100;␤ expecting any of:␤ infix␤ infix stopper␤ postfix␤ statement end␤ …»
smls so if its a non-constant expression, you may want to store it in a variable first and then do $x xx 100
BenGoldberg m: say my @a = (0,) x 100; 00:08
camelia rakudo-moar d8fa56: OUTPUT«0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000␤»
BenGoldberg Does [x] now only do string repetition?
smls yes
x always does string repetition; xx always does list repetition
00:09 HoppingMadMan left
BenGoldberg Is there a convenient and efficient way to do list repetition without repeatedly evaluating the lhs? 00:10
timotimo if the lhs is compile-time constant or simple, it won't thunk
or something like that
what i mean is we have something in place in the hopes of making it smarter
BenGoldberg m: say my @a = do { my $x = rand; $x xx 10 }; 00:12
camelia rakudo-moar d8fa56: OUTPUT«0.945652991137229 0.945652991137229 0.945652991137229 0.945652991137229 0.945652991137229 0.945652991137229 0.945652991137229 0.945652991137229 0.945652991137229 0.945652991137229␤»
BenGoldberg m: say my @a = do { my \x = rand; x xx 5 }; 00:13
camelia rakudo-moar d8fa56: OUTPUT«0.0193072417998126 0.0193072417998126 0.0193072417998126 0.0193072417998126 0.0193072417998126␤»
smls you can always define a sub rep ($value, $n) { $value xx $n } 00:14
BenGoldberg m: sub infix::<xxx>(\value, \n) { value xx n }; say rand xxx 4; 00:15
camelia rakudo-moar d8fa56: OUTPUT«===SORRY!===␤Name infix:: ends with '::' and cannot be used as a sub declaration␤»
BenGoldberg m: sub infix<xxx>(\value, \n) { value xx n }; say rand xxx 4;
camelia rakudo-moar d8fa56: OUTPUT«5===SORRY!5=== Error while compiling /tmp/N2VWU5of82␤Missing block␤at /tmp/N2VWU5of82:1␤------> 3sub infix7⏏5<xxx>(\value, \n) { value xx n }; say ra␤ expecting any of:␤ new name to be defined␤»
smls timotimo (since you're here): Does this look like a sane thing to do? :
subset StrOrArrayOfStr where Str | (Array & {.all ~~ Str});
BenGoldberg: infix:<xxx> needs a colon :) 00:16
BenGoldberg m: sub infix:<xxx>(\value, \n) { value xx n }; say rand xxx 4; 00:17
camelia rakudo-moar d8fa56: OUTPUT«0.661307514012933 0.661307514012933 0.661307514012933 0.661307514012933␤»
BenGoldberg m: sub infix:<xxx>(\value, \n) { value xx n }; say (rand, rand) xxx 4;
camelia rakudo-moar d8fa56: OUTPUT«0.373813721034791 0.287736870594361 0.373813721034791 0.287736870594361 0.373813721034791 0.287736870594361 0.373813721034791 0.287736870594361␤»
BenGoldberg \o/
00:18 census left
timotimo sorry, very distracted 00:21
smls np
timotimo seems like it'd be very tough on the performance, but it should work 00:23
00:23 cognominal joined
smls Iterable is a class rather than a role? Interesting. 00:29
TimToady all bets are off after GLR 00:30
timotimo every time you check, it'll return a different result what kind of thing it is 00:31
"oh no, today iterable is a package again ... what should i do?!"
smls
.oO( class Foo does-or-is-or-whatever Iterable )
 00:32
00:32 vendethiel left
timotimo does-or-is-or-trusts-or-augments ... 00:33
00:33 ShimmerFairy left
TimToady we might not even have an Iterable type 00:36
timotimo we'll have a perl6 and a perl6-iterable instead
TimToady has flashbacks to incr-tcl 00:37
timotimo increase the tclness of your system by 1 every time this gets executed
smls TimToady: I expected Iterable to be a tag role to advertise "I have multiple elements and I implement .iterator, .list, keys & co to iterate them" 00:38
like Positional is a tag role to advertise "I have multiple elementd and I implement .AT-KET, .ASSING-KEY, etc. to index them" 00:39
TimToady we might well keep the type name, but we'll see what jnthn++ comes up with
00:40 fling joined
timotimo Iterobubble 00:41
00:47 ShimmerFairy joined 00:53 skids joined 01:03 cognominal left 01:10 vendethiel joined 01:22 rmgk_ joined, rmgk is now known as Guest74991, Guest74991 left, rmgk_ is now known as rmgk 01:24 shinobi-cl_ left 01:35 mr-foobar left 01:36 mr-foobar joined, Hor|zon joined 01:37 raiph left
timotimo who wants to write szabgab to improve his line use lib callframe(0).file.IO.dirname ~ '/lib'; 01:39
01:39 gfldex left 01:41 Hor|zon left 01:44 smls left 01:45 ilbot3 left 01:47 ilbot3 joined 01:51 llfourn joined, HoppingMadMan joined 01:56 llfourn left
timotimo hm, $?FILE may be bad if the module's installed, no? 01:57
02:00 ShimmerFairy left
skids ugexe suggested this after a long discussion: irclog.perlgeek.de/perl6/2015-07-14#i_10891237 02:01
timotimo but doesn't andthen not properly set $_? 02:06
02:11 cognominal joined 02:13 ShimmerFairy joined
skids timotimo: Dunno, I haven't tried to use it yet. 02:21
02:23 mr-foobar left 02:24 mr-foobar joined 02:25 vendethiel left
skids gist.github.com/skids/c4a548c8e829cd4dd480 # Hopefully a useful resource for both sides of the "implicit *%_" debate 02:26
02:45 mr-foobar left 02:47 noganex_ joined 02:49 noganex left 03:00 Sgeo joined 03:01 yqt left 03:08 dj_goku_ left
ShimmerFairy skids++ spotted just one typo: "wright a wrapper class" 03:23
skids fixed. thanks. I have a buidin named "wright" at work so that polluted my muscle memory :-) 03:25
ShimmerFairy skids: the only argument I think doesn't have any validity is the pro-*%_'s "But Perl 5 did it like this!". That's an argument that wouldn't work against a lot of P6's changes, and anyone who doesn't know P5 (such as myself) won't put much faith in the tradition argument wrt P6 :)
skids *building
ShimmerFairy skids: in reality, implicit *%_ doesn't affect my life so much that it'll upset me that it's around. I'm more bothered that it's a wildly inconsistent feature in the face of subs and the other kind of parameters, positionals. 03:26
(The fact that it's under the heading "Interface Consistency" is hilarious.) 03:27
03:29 laouji joined
skids I'd be less averse to changing it if the change came with a promise that most docs showing how to do OO stuff included *%_ in examples so it felt "endorsed" to use it. 03:29
ShimmerFairy Sure, I'm fine with encouraging the use of something that'll make your classes (allegedly) more robust. I just think it's a weird and unintuitive default :) 03:31
skids: Freedom #2 sounds reasonable, however I could argue against it. Rapid prototyping is one thing, being allowed to be sloppy is another. After all, there's a reason 'strict' and 'warnings' are on by default in P6 :P 03:33
03:36 dmitri joined 03:37 Hor|zon joined
skids I haven't felt chafed by the existing strictures and thankfully "my" is huffmanized. Though lately the idea that Failures should complain when GCd has been irking me. 03:38
ShimmerFairy m: class C { method mypush { say %_.perl } }; C.mypush(1=>2) # huh, I wasn't expecting this
camelia rakudo-moar d8fa56: OUTPUT«Too many positionals passed; expected 1 argument but got 2␤ in method mypush at /tmp/yK1XFU8qt4:1␤ in block <unit> at /tmp/yK1XFU8qt4:1␤␤»
ShimmerFairy m: class C { method mypush() { say %_.perl } }; C.mypush(a=>2)
camelia rakudo-moar d8fa56: OUTPUT«{:a(2)}<>␤»
skids and: 03:39
m: class C { method mypush() { say %_.perl } }; C.mypush("a"=>2)
camelia rakudo-moar d8fa56: OUTPUT«Too many positionals passed; expected 1 argument but got 2␤ in method mypush at /tmp/u11N0uZ0wT:1␤ in block <unit> at /tmp/u11N0uZ0wT:1␤␤»
ShimmerFairy _that's_ what I was expecting (my second line). (And a good demo of why Pairs and nameds shouldn't be conflated in syntax; hell, I got confused by it just now, and I've been using P6 for years!) 03:40
skids Though the first line is debatably a bug since "1" is a valid string. 03:41
03:41 Hor|zon left
ShimmerFairy m: my %hash; %hash.push(a=>42); say %hash.perl; # this is my bigger gripe concerning nameds; removal of *%_ would only make it a non-silent fail 03:41
camelia rakudo-moar d8fa56: OUTPUT«{}<>␤»
dalek osystem: 96747e1 | (Sterling Hanenkamp)++ | META.list:
Adding IO::Glob to the Perl6 ecosystem
 03:42
ShimmerFairy skids: but you can't do sub foo(:$1) as a named, so ~clearly~ it's always a positional Pair param.
skids Ok I'll buy that. 03:43
The fact that a => 1 works is also a P5 legacy thing. 03:44
ShimmerFairy skids: I'm personally with lizmat in thinking that the conflation of named parameters and Pairs is a terrible mistake on P6's part. 03:45
skids I've rarely wanted to use pairs as items personally. 03:46
ShimmerFairy sure, there are at least a few legacy P5 things. Doesn't mean that's a solid argument though :P (I think it'd be nice if it quoted the right side too, but maybe that's just me).
skids: At the very least though, there should be a way to uniquely refer to nameds xor Pairs syntactically. 03:48
skids Pair.new? 03:49
I don't really view an itemized Pair as a thing useful enough to deserve special syntax.
ShimmerFairy Pair.new doesn't do anything, surprisingly 03:50
skids m: Pair.new(:key<a>,:value<b>).say
camelia rakudo-moar d8fa56: OUTPUT«a => b␤»
ShimmerFairy o_o 03:51
skids I wouldn't be opposed to Pair having a positional .new multi, personally.
zostay i'm in favor of eliminating implicit *%_ because it would greatly simplify ArrayHash and Hash::MultiValue... working out all the little fiddly bits and dealing with that always being implied has been a major pain 03:52
ShimmerFairy That's a horrendous way to have a unique Pair, esp. when you have syntax that _should_ do it already (as it is, you have to hope the key in a fat-arrow Pair isn't an ident literal)
03:52 llfourn joined
zostay i don't really understand why *%_ keeps on existing with an explicit signature when *@_ goes away 03:52
skids Well, from my perspective the syntax is there for parameters and Pairs are just hitchiking/an implementation detail leaking through. 03:53
ShimmerFairy skids: and even then, there may be an unambiguous way to specify a Pair, but there still isn't any such thing for nameds.
zostay seems like if i'm being explicit, i should need to add my own *%_ if i want it still
my 2¢
skids I thought colonpair was unambiguously a named in an arglist? 03:55
03:56 llfourn left
dmitri Survey: what is the extension you use for Perl 6 scripts: .p6, .pl6, .pl, none at all, or something else? 03:57
skids pm6 personally ut few others do.
dmitri I am adding Perl 6 support to ctags and it's important to it (that's the normal way to determine language) 03:58
well, .pm6 is a module
.pm is also used by Perl 6 modules
skids Yeah I stopped making that distinction years ago even for Perl 5 stuff. 03:59
03:59 lue joined, ShimmerFairy left, lue is now known as ShimmerFairy
ShimmerFairy I personally use .p6 for scripts, .pm6 for modules 04:00
skids Because if it is really a module it should be ensconsed in a /lib/ somewhere not in a likely-to-be cwd.
ShimmerFairy (and think it's a terrible missed opportunity that rakudo's core uses .pm :P)
skids Yeah that bothers me.
dmitri ShimmerFairy: Yes -- I had to add "tasting" code for the .pm files -- is it Perl 5 or Pelr 6?
ShimmerFairy skids: I think colonpairs work out as always nameds in arglists just by virtue of only allowing a legal variable name as the key in them 04:01
skids I think that's not an accident though. 04:02
ShimmerFairy m: my %h = :a(42); say %h.perl
camelia rakudo-moar d8fa56: OUTPUT«{:a(42)}<>␤»
ShimmerFairy ^ in reality it's not an unambiguous syntax for nameds
dmitri: One of the specs (S01?) and maybe somewhere on docs.perl6.org describes various features distinguishing 5 and 6 --- 'use v6' (or v5), the presence of a 'unit module/class/etc.' statement, and so on. 04:03
skids updates gist to add a sentence about interaction of Pair vs named with *%_. 04:04
dmitri ShimmerFairy: yes, but 04:05
"use v6" is not followed by most of Perl 6 code I've looked at
(even though, admittedly, it's not much)
skids ShimmerFairy: no not in an assignment it isn't. 04:06
ShimmerFairy dmitri: I personally always 'use v6' to make sure nobody dares mistake my code for the other Perl :P . If there's a lack of people using it, my guess would be because we've only had one (evolving) version of Perl 6 so far.
skids: ? I don't quite follow 04:07
dmitiri: I imagine in lots of cases there won't be an easy way to detect between P6 and P5 files, just like C++ .h files :(
dmitri ShimmerFairy: I think it's not that bad, actually, at least if we assume that .pm is either Perl 5 or Perl 6 *module* 04:09
ShimmerFairy yeah, and .pl either P5 or P6 script :)
skids ShimmerFairy: The right side of = is not exactly an arglist, it is its own thing.
04:10 beastd left
dmitri The .pl case is obviously harder 04:10
ShimmerFairy skids: sure, but I'm talking about ambiguity of syntax across the language. I can't immediately think of another case like => vs. :() in the language.
dmitri I think I'll just pretend that .pl is always Perl 5 04:11
ShimmerFairy dmitri: I think that's a good choice, all things considered (just like, to use the same comparison, assuming .h is C by default ☺) 04:12
04:15 isBEKaml joined 04:16 Calibellus joined
skids Well, I guess it hasn't bothered me because I learned => as "there in case you can't let go of P5 habits" and I've rarely ever used it. 04:17
Which is odd because I'm horrible at losing habits. 04:18
ShimmerFairy among other issues, consider that you're allowed to substitute colonpairs for fat arrow in arglists, but not in adverbs. (Not that I advocate allowing fatarrow for adverbs, it's just another minor inconsistency)
skids: Well, for me => is a Pair-y syntax, and :() is a named-y syntax. It frustrates me that the two concepts have been conflated just because nameds can kinda be treated as Pair-like maybe. 04:19
The only other conflation I can think of right now is that of { } for either hashes or code blocks, and that's really an issue for the parser (whereas what I'm complaining about is an issue for the humans writing code) 04:24
04:25 bin_005_x joined, khw left 04:26 bin_005 left 04:34 laouji left, laouji joined 04:39 laouji left 04:40 laouji joined 04:43 isBEKaml left 04:44 isBEKaml joined 04:47 isBEKaml left
dmitri ShimmerFairy: github.com/dtikhonov/ctags/blob/ba...arse.c#L44 04:52
04:56 skids left, cognominal left 05:06 llfourn joined 05:08 xinming_ left 05:14 davido__ left 05:15 Calibellus left 05:16 davido__ joined 05:17 cognominal joined
cognominal \o #perl6 05:19
I am attempting to provide a minimal example for QAST::ParamTypeCheck. I am doing it wrong. Can someone help? gist.github.com/cognominal/efa78cd052e6b9ef7268 05:20
05:20 llfourn left 05:21 dmitri left 05:34 laouji left 05:38 Hor|zon joined
Sgeo Wait, what's *%_? 05:41
Almost thinking it's 'allow arbitrary named arguments and ignore them' because in other languages _ would mean unused variable 05:42
But $_ has special meaning, and @_ did in Perl5
05:42 laouji joined, Hor|zon left 05:45 BenGoldberg left 05:46 ChoHag joined
ChoHag How do you provide values to the attributes of an inherited class? 05:47
05:57 pierrot left 05:59 llfourn joined 06:08 pierrot joined 06:20 laouji left 06:26 aindilis joined 06:27 aindilis` joined 06:35 brrt joined, darutoko joined 06:38 llfourn left 06:39 Hor|zon joined
FROGGS Sgeo: *%_ is 'slurp any additional named argument' 06:40
Sgeo And do what with it? Ignore it?
FROGGS [...]'into an anonymous hash'
m: sub foo(*%_) { say %_ }; foo :a(42)
camelia rakudo-moar d8fa56: OUTPUT«a => 42␤»
FROGGS Sgeo: you can access %_, but usually that idiom is used to ignore them 06:41
Sgeo Ah
jdv79 its just bikeshedding but who came up with this IETF::RFC_Grammar namespace? 06:43
just seems odd to use _ instead of just :: and RFC is unambiguous so why IETF...
06:44 Hor|zon left
jdv79 is the top level Grammar name not a good idea? 06:45
seems not - things exist already
07:08 RabidGravy joined 07:33 azawawi joined
azawawi hi 07:39
07:40 Hor|zon joined
azawawi japhb: ping 07:40
07:41 aindilis left, aindilis` left 07:42 aindilis joined 07:44 Hor|zon left
azawawi .tell japhb re your discussion with timotimo++ about Electron. It is basically chrome + nodejs + API which we are accessing via JSON::RPC. We're setting up a nodejs jsonrpc server and using it via JSON::RPC and wrapping all into lovely p6 classes :) 07:47
yoleaux azawawi: I'll pass your message to japhb.
azawawi .tell japhb examples are found here github.com/azawawi/perl6-atom-elec...r/examples 07:48
yoleaux azawawi: I'll pass your message to japhb.
07:52 pmurias joined 07:53 rurban joined 07:57 TEttinger left
pmurias hi 07:59
08:01 virtualsue joined 08:02 HoppingMadMan left
brrt hi pmurias 08:12
cognominal hum, there was many typos in my code. No wonder it did not work. 08:17
08:18 diana_olhovik joined 08:27 xfix joined 08:28 rurban left
jdv79 m: class Foo:::Bar {} # a bug in the error? 08:29
camelia rakudo-moar d8fa56: OUTPUT«===SORRY!===␤Name Foo:::Bar ends with '::' and cannot be used as a package name␤»
08:33 rurban joined, rurban left 08:34 llfourn joined 08:38 llfourn left
jdv79 is there a way to combine a rule and a lit for purposes of match/backtrack? 08:54
nm. ?? seems good enogh. 09:00
09:05 rurban joined 09:08 rurban left 09:23 dsm joined
jdv79 nice when a seemingly simple rfc is itself confused 09:27
:(
09:27 rindolf joined 09:34 llfourn joined 09:35 telex left 09:36 telex joined 09:39 llfourn left 09:41 Hor|zon joined
jdv79 m: class A { has $.a; }; my $s = 'a'; A.new.$s; # how does one do that? 09:43
camelia rakudo-moar d8fa56: OUTPUT«Cannot find method 'postcircumfix:<( )>'␤ in block <unit> at /tmp/iyCAhJXeDF:1␤␤»
09:45 Hor|zon left
psch m: class A { has $.a = "foo"; }; my $s = 'a'; say A.new."$s"(); 09:46
camelia rakudo-moar d8fa56: OUTPUT«foo␤»
jdv79 doh!
thanks
was just a little taken aback that $o.perl kicks out a constructor call that may not be valid 09:47
i was attempting to dome testing but in aggregate. wonder how to do that now. 09:48
09:48 virtualsue_ joined 09:49 virtualsue left, virtualsue_ is now known as virtualsue
nine S06 says the want function is gone and I should "return an object instead that responds accordingly to the various contextual methods". But only the "sink" method seems to ever be called. "my $item =" and "my @list =" just assign without calling any methods. So how is this really supposed to work? 09:52
pmurias as far as I know you can't emulate the old context stuff 09:53
the object you return should just behave like both a scalar/list instead of having a conditional in the function 09:54
do I guess correctly you want that for the p5 interop? 09:55
nine pmurias: yes
pmurias I think specific the context manually when calling the method/sub is the only way 09:56
09:56 xfix left
psch i'd guess you have to kind wrap what wantarray does 09:56
pmurias * specifying
psch i.e. tell perl6 if perl5 would return true or false for wantarray in that spot
psch .oO( that's what pmurias++ says i guess ) 09:57
pmurias that's the opposite
psch ...it is? better hurry with the coffee, me :) 09:58
pmurias when calling a p5 method we would need something like foo(:context<scalar>, 1, 2,3)
s/method/sub/
nine But in essence, what S06 says is just plain wrong?
psch nine: S06 talks about contextual methods, i'm not sure assignment is one of those 10:00
m: class A { method list { 1,2,3 }; method item { 6 } }; .say for A.new; say A.new
camelia rakudo-moar d8fa56: OUTPUT«A.new␤A.new␤»
pmurias it tells you what to do if you want context specific behaviour
nine pmurias: I want context specific behaviour and I did what it told me, but it just doesn't work that way as we can see in psch++'s example 10:01
psch yeah, that got me too just now
nine m: class A { method sink { say "sink!"; } }; A.new
camelia rakudo-moar d8fa56: OUTPUT«sink!␤»
nine Only sink ever gets called
psch m: class A { method list { 1,2,3 }; method item { 6 }; method sink { "sunk" } }; .say for A.new; say A.new 10:02
camelia rakudo-moar d8fa56: OUTPUT«A.new␤A.new␤»
10:02 araujo left
psch m: class A { method list { 1,2,3 }; method item { 6 }; method sink { "sunk" } }; my @a = A.new; my $a = A.new; say @a ~ "|" ~ $a 10:02
camelia rakudo-moar d8fa56: OUTPUT«A<140404705145096>|A<140404705147544>␤»
psch that's .gist
?
ah no .Str
m: class A { method list { 1,2,3 }; method item { 6 }; method sink { "sunk" }; method Str { }; my @a = A.new; my $a = A.new; say @a; say $a 10:03
camelia rakudo-moar d8fa56: OUTPUT«5===SORRY!5=== Error while compiling /tmp/wdKvX1Yv2q␤Missing block␤at /tmp/wdKvX1Yv2q:1␤------> 3a = A.new; my $a = A.new; say @a; say $a7⏏5<EOL>␤ expecting any of:␤ postfix␤ statement end␤ statement modifier…»
psch m: class A { method list { 1,2,3 }; method item { 6 }; method sink { "sunk" }; }; my @a = A.new; my $a = A.new; say @a; say $a
camelia rakudo-moar d8fa56: OUTPUT«A.new␤A.new␤»
ShimmerFairy m: class C { method list { 1,2,3 } }; say @(C)
camelia rakudo-moar d8fa56: OUTPUT«1 2 3␤»
psch i'm really missing something there
pmurias nine: re plain wrong, the old behaviour is gone and not emulatable and what we now have is something different
ShimmerFairy m: class C { method list { 1,2,3 }; method item { 6 } }; my $a = C.new; say $a.perl; say @$a.perl; say $$a.perl; 10:05
camelia rakudo-moar d8fa56: OUTPUT«C.new␤(1, 2, 3)␤6␤»
nine pmurias: but it tells me that an object can respond accordingly to various contextual methods, when there seems to only be "sink" which is a single method.
ShimmerFairy nine, psch: list and item do get called when requested, just perhaps not in all the places you expect :) 10:06
psch ShimmerFairy: yeah, the "contextual" bit is missleading in S06 in my opinion
pmurias nine: it's a conjencture
psch ShimmerFairy: but maybe that's my understanding, i'd want .list when iterating...
pmurias nine: and even if it worked it would be really bad to use that for p5 interop
psch s/ und/ misund/
ShimmerFairy m: my $a = Array.new(1,2,3); say $a.perl; say $_ for $a; say $_ for @$a 10:07
camelia rakudo-moar d8fa56: OUTPUT«5===SORRY!5=== Error while compiling /tmp/Cm3aVKYMXg␤Bogus postfix␤at /tmp/Cm3aVKYMXg:1␤------> 3y $a.perl; say $_ for $a; say $_ for @$a7⏏5␤ expecting any of:␤ infix␤ infix stopper␤»
ShimmerFairy m: my $a = Array.new(1,2,3); say $a.perl; say $_ for $a; say $_ for @$a
camelia rakudo-moar d8fa56: OUTPUT«[1, 2, 3]␤1 2 3␤1␤2␤3␤»
ShimmerFairy psch: ^^^ doesn't iterate $a in this case either, without explicitly list-ifying it
pmurias nine: because in p5 we know the context at sub call time and in p6 we only know that at return value use time
psch ShimmerFairy: yeah, and that's actually fine. looking at it from wantarray, we had context information in perl5 on the LHS, but now we need it on the RHS 10:08
nine pmurias: hence the idea to lazily execute the function once the context is clear
psch because the assignment itself doesn't look at the LHS anymore
pmurias nine: seems horrible to me
nine pmurias: if there are better ideas, I'd be happy to hear them :) 10:09
pmurias assume a context (scalar?) and allow the caller to override with a named argument? 10:10
"foo(1,2,3)" and "foo(:list, 1, 2, 3)"
nine: slightly less terse but much less likely to explode 10:11
nine pmurias: the use case I have is a context sensitive P5 method (from DBIx::Class) that's being delegated to by a fallback in a P6 subclass of the P5 ResultSet and the original caller being in P5 code. So we have P5 -> P6 -> P5 and I need to propagate the context through this P6 code somehow. 10:12
pmurias: I love your idea for the case when P6 code just invokes a P5 method and knows what it needs. Unfortunately in my case the layer between P5 and P6 (the "->" in P5 -> P6) doesn't know that the P6 code will just delegate to a P5 method. 10:13
pmurias $context = v5::wantarray(); :context($context)
$context = v5::wantarray(); $foo.bar(:context($context), ...) 10:14
nine That may end up passing a :context argument to some P6 method that wouldn't know what to do with it. Or worse that has exactly such a named argument but for a different purpose.
pmurias the second -> doesn't know that we are propagating context? 10:16
nine Worse: I translate all named arguments to positional arguments when calling into P5 code, so we can still do $p5obj->foo(foo => 'bar');
pmurias: yes, the second -> would be where I need the context information. But I'd have to be sure we're just delegating and that we need to propagate. Otherwise we'd propagate the context to completely unrelated calls. 10:18
10:19 rurban joined 10:20 rurban left
pmurias having to manually setting the context propagation doesn't seem like a heavy price to me 10:20
s/setting/set/
nine pmurias: manually where?
10:20 rurban joined
pmurias P5 -> (fetch the context here, using something like $context = v5::wantarray) P6 (set the context using something like :context($context) here) -> P5 10:21
nine The P6 code would have to know that it is called by P5 code directly. 10:23
pmurias yes
I don't think there is a way to that otherwise
lazily executing function sounds like something that would cause needless debugging on the users 10:24
nine So I need to check my callframe in the fallback method 10:25
10:26 rurban left
pmurias fallback method? 10:27
10:28 spider-mario joined
[TuxCM] so, "," has higher precedence than ?? !! ? 10:28
m: my $a = 1; my @b = $a == 1 ?? 1, 2, 3 !! 4, 5, 6
camelia rakudo-moar d8fa56: OUTPUT«5===SORRY!5=== Error while compiling /tmp/z984s0_Ssr␤Precedence of , is too loose to use inside ?? !!; please parenthesize␤at /tmp/z984s0_Ssr:1␤------> 3my $a = 1; my @b = $a == 1 ?? 17⏏5, 2, 3 !! 4, 5, 6␤ expecting any of:␤ i…»
[TuxCM] the message is clear, but I am still surprised
10:29 llfourn joined, vendethiel joined
nine pmurias: the fallback method in Inline::Perl5::Perl5Parent that's delegating to the P5 method 10:31
ShimmerFairy [TuxCM]: it's actually saying that , has a lower precedence
[TuxCM] if it had, it would not be ambiguous 10:32
FWIW, I have no problem using parens here 10:33
ShimmerFairy [TuxCM]: lower precedence means the commas get their arguments _after_ ?? !! ; i.e. the parser would read it as (1 ?? 1), 2, (3 !! 4), 5, 6 . Which makes no sense, thus the error. 10:34
10:36 leont joined 10:39 cognominal left 10:54 brrt left 10:56 llfourn left 11:05 rindolf left
[TuxCM] what discriminates a Block from a Callable? 11:13
how do I dynamically pass a Callable?
nine [TuxCM]: a block is Callable 11:16
psch Block does Callable
nine or rather does Callable AFAIK
psch the latter is a role, yes
11:16 Psyche^_ joined
psch well, Block is Code, Code does Callable 11:17
11:18 azawawi left
[TuxCM] can I combine "things" in a when? 11:19
given ($x.WHAT) { when Routine|Sub|Block|Callable|code { ... }
moritz yes
[TuxCM] Code
moritz though note that Callable immplies all of the others 11:20
m: for Routine, Sub, Block, Code -> $t { say $t ~~ Callable }
camelia rakudo-moar d8fa56: OUTPUT«True␤True␤True␤True␤»
psch Sub <: Routine <: Block <: Code i think?
11:20 Psyche^ left
psch might be wrong order on the first two 11:20
moritz doc.perl6.org/type/Callable has a nice type graph 11:22
[TuxCM] given ($in.WHAT) { when Callable { @foo = gather while $in() -> $x { take $x }; } 11:24
ok for Sub, but fail for Block
what am I missing there?
11:27 Hor|zon joined
[TuxCM] Found it: my block had a return 11:28
moritz [TuxCM]: fwiw given $in { when Callable { ... } } 11:30
11:30 llfourn joined
moritz [TuxCM]: should work the same way 11:30
(of course, only if you other when clauses also work with the object itself, not just the type) 11:31
[TuxCM] thanks, /me checks
11:31 Hor|zon left
moritz m: sub f { state $x = 10; $x-- }; my @a = gather while f() -> $x { take $x } 11:35
camelia ( no output )
moritz m: sub f { state $x = 10; $x-- }; my @a = gather while f() -> $x { take $x }; say @a
camelia rakudo-moar d8fa56: OUTPUT«10 9 8 7 6 5 4 3 2 1␤»
nine Seems like callframe won't give me the caller, while &?CALLER::ROUTINE doesn't work in FALLBACK methods :/ 11:37
11:47 gfldex joined, Peter_R left
timotimo oh, because the caller of the fallback method is in the meta object protocol implementation, ib et 11:52
11:54 leont left
nine timotimo: that's not a real problem. Problem is jnthn broke &?CALLER::CALLER::ROUTINE :/ It always returns the FALLBACK regardless of how many CALLER:: I stuff in there. It did work on Tuesday's rakudo. 11:55
11:57 Peter_R joined 11:58 smls joined
nine m: role Marker { }; sub foo { say &?CALLER::ROUTINE ~~ Marker; }; my $test = sub { foo(); }; $test(); $test does Marker; $test(); 11:59
camelia rakudo-moar d8fa56: OUTPUT«False␤False␤»
timotimo blogs.perl.org/users/aaron_baugher/...arted.html ← cool 12:00
oh, that's bad :|
smls .tell dmitri For .pm files, you can assume it's Perl 5 by default, and only Perl 6 if the first non-comment line starts with one of: use v6 unit module class 12:01
yoleaux smls: I'll pass your message to dmitri.
12:06 rindolf joined
ChoHag How do you initialise the values of a superclass other than by stuffing them in the new() call? 12:08
nine .tell jnthn your commit 56ae33ea0a5ac3d54a5fef2aa15eab34f9e9594b broke &?CALLER::ROUTINE: role Marker { }; sub foo { say &?CALLER::ROUTINE ~~ Marker; }; my $test = sub { foo(); }; $test(); $test does Marker; $test(); 12:09
yoleaux nine: I'll pass your message to jnthn.
smls ChoHag: With `nextwith` I think 12:11
ChoHag And BUILD is no good because the superclass' attribute is required but the subclass' BUILD is called last. 12:12
nine ChoHag: I guess, you'd need to write your own "new" 12:13
12:18 mohij joined 12:19 araujo_ joined, araujo_ left 12:21 census joined 12:27 spidermario joined 12:28 spider-mario left, spidermario is now known as spider-mario 12:45 beastd joined 12:48 mohij left 12:56 mohij joined 13:12 spider-mario left 13:24 virtualsue left 13:28 Hor|zon joined 13:33 Hor|zon left 13:35 xinming joined, virtualsue joined, virtualsue left 13:36 virtualsue joined 13:40 khw joined 13:41 mohij left 13:47 bin_005_x left 13:54 FROGGS_ joined 13:57 FROGGS left 13:59 mr_ron joined 14:05 virtualsue left
smls m: my @a = 5, 10, 15; @a[1]:delete; dd @a, +@a; 14:13
camelia rakudo-moar d8fa56: OUTPUT«@a = [5, Any, 15]<>␤3␤»
smls ^^ so delete no longer actually creates "arrays with holes"?
(Not that I'm complaining, I always considered that feature a mistake.) 14:14
m: my @array = 5, 10, 15; @array[2]:delete; dd @array, +@array 14:15
camelia rakudo-moar d8fa56: OUTPUT«@array = [5, 10]<>␤2␤»
smls looks like it can cut off elements from the end though
smls would still prefer :delete being replaced by a :remove with splice semantics. 14:16
14:31 skids joined
nine m: use nqp; role Marker { }; sub foo { say nqp::getcodeobj(nqp::ctxcode(nqp::ctxcaller(nqp::ctx))) ~~ Marker; }; my $test = sub { foo(); }; $test does Marker; $test(); 14:31
camelia rakudo-moar d8fa56: OUTPUT«True␤»
nine What a workaround :)
14:32 haroldwu joined
nine .tell jnthn I can workaround missing &?CALLER::ROUTINE with nqp::getcodeobj(nqp::ctxcode(nqp::ctxcaller(nqp::ctx))) 14:32
yoleaux nine: I'll pass your message to jnthn.
14:39 BenGoldberg joined, virtualsue joined 14:43 laouji joined
timotimo m: my @a = 5, 10, 15; @a[1]:delete; dd @a, +@a; say @a[1]:exists; 14:44
camelia rakudo-moar d8fa56: OUTPUT«@a = [5, Any, 15]<>␤3␤False␤»
timotimo ^- it makes an array with a hole, smls
smls So it's just that .elems doesn't know about holes? 14:45
(nor does .perl, because it won't round-trip there, will it) 14:46
(yet more evidence that holes are a mis-feature...)
(and deprecated/shunned in Perl 5 for good reason.) 14:47
timotimo i suppose we don't want .elems to go through and count the holes 14:48
14:51 laouji left
timotimo what's problematic about having a hole? 14:51
psch @a[*-1]:delete not leaving a hole at the edge is weird to me
i mean, i kinda see why a hole at the edge doesn't really make sense, but that feels like stretching the metaphor a bit too far
smls "what's problematic" -- the fact that, as .elems and .perl show, it does cause an overall consistent design to fall into place. 14:54
and the fact that it blocks the more useful splice-based :delete/:remove adverb from being added
timotimo that sentence makes no sense to me
smls Well, .perl is supposed to round-trip, isn't it? with holes it doesn't, and there's no declarative way to specify holes which .perl could use. 14:55
14:55 raiph joined
smls And .elems matching the number of elements that :exists would not be an unreasonable expectation w.r.t. consistency. But again, holes make that unfeasible. 14:56
timotimo m: my @a = 5, 10, 15; @a[1]:delete; say @a.perl
camelia rakudo-moar d8fa56: OUTPUT«[5, Any, 15]<>␤»
timotimo m: say ([5, Nil, 15]<>).perl
camelia rakudo-moar d8fa56: OUTPUT«[5, Any, 15]<>␤»
timotimo m: say ([5, Nil, 15]<>)[1]:exists
camelia rakudo-moar d8fa56: OUTPUT«True␤»
timotimo interesting.
psch m: my @a = 5, 10, 15; @a[1]:delete; my @b = EVAL @a.perl; say @b.perl; say @a[1] === @b[1]
camelia rakudo-moar d8fa56: OUTPUT«[5, Any, 15]<>␤True␤»
psch m: my @a = 5, 10, 15; @a[1]:delete; my @b = EVAL @a.perl; say @b.perl; say @a[1]:exists; say @b[1]:exists 14:57
camelia rakudo-moar d8fa56: OUTPUT«[5, Any, 15]<>␤False␤True␤»
psch m: my @a = 0, Any, Nil; say +@a
camelia rakudo-moar d8fa56: OUTPUT«3␤»
psch m: my @a = 0, Any, Nil; say +@a; @a[2]:delete; say @a[2] 14:58
camelia rakudo-moar d8fa56: OUTPUT«3␤(Any)␤»
timotimo m: my @a = 5, 10, 15; @a[1] = Nil; say @a[1]:exists
camelia rakudo-moar d8fa56: OUTPUT«True␤»
timotimo m: my @a = 5, 10, 15; @a[1] = Nil; say @a.perl
camelia rakudo-moar d8fa56: OUTPUT«[5, Any, 15]<>␤»
timotimo i thought assigning Nil to a slot in the array would set its :exists no False
smls from perldoc (perldoc.perl.org/functions/delete.html): "WARNING: Calling delete on array values is strongly discouraged. The notion of deleting or checking the existence of Perl array elements is not conceptually coherent, and can lead to surprising behavior." 14:59
psch m: my @a = 0, 1, 2; @a[1] = nqp::null(); say @a[1]:exists
camelia rakudo-moar d8fa56: OUTPUT«===============================================================================␤The use of nqp::operations has been deprecated for non-CORE code. Please␤change your code to not use these non-portable functions. If you really want␤to keep using nqp:…»
psch m: use NQP; my @a = 0, 1, 2; @a[1] = nqp::null(); say @a[1]:exists
camelia rakudo-moar d8fa56: OUTPUT«===SORRY!===␤Could not find NQP in any of:␤ file#/home/camelia/.perl6/2015.07.1-51-gd8fa568/lib␤ inst#/home/camelia/.perl6/2015.07.1-51-gd8fa568␤ file#/home/camelia/rakudo-inst-2/share/perl6/lib␤ file#/home/camelia/rakudo-inst-2/share/perl6/v…»
smls This seems like one of the features which Perl 6 carried over from old Perl 5 versions, and never got the memo that the Perl 5 community found out that they were a mistake.
psch eh, how was that pragma..?
m: use nqp; my @a = 0, 1, 2; @a[1] = nqp::null(); say @a[1]:exists 15:00
camelia rakudo-moar d8fa56: OUTPUT«True␤»
15:02 kaare_ joined
psch anyway, DELETE-POS binds with nqp::null 15:02
m: my @a = ^3; @a[1] := Nil; say @a[1]:exists
camelia rakudo-moar d8fa56: OUTPUT«True␤»
psch so i don't get it :) 15:03
m: use nqp; say nqp::isnull(Nil) 15:04
camelia rakudo-moar d8fa56: OUTPUT«0␤»
psch m: use nqp; say nqp::isnull(nqp::null())
camelia rakudo-moar d8fa56: OUTPUT«1␤»
psch List.pm defines EXISTS-POS with an nqp::isnull, like that
and DELETE-POS binds nqp::null 15:05
assigning Nil doesn't assign nqp::null
m: use nqp; my @a = 1,2; @a[0]:delete; say nqp::isnull(@a[0]) 15:06
camelia rakudo-moar d8fa56: OUTPUT«0␤»
psch m: use nqp; my @a = 1,2; @a[0]:delete; say nqp::isnull(@a[0]); say @a[0]:exists
camelia rakudo-moar d8fa56: OUTPUT«0␤False␤»
psch ...but AT-POS hllizes?
does that fall into the GLR? :) 15:07
15:09 mohij joined
timotimo if there's a hole, you'll get something that can autovivify, no? 15:13
jdv79 hmm, so how to d the p6 equiv of a p5 if ( my $lex = 42 {lex here } no lex here? 15:15
timotimo those parens are unbalanced 15:16
psch m: -> my $a = 0 { say $a }; try { say $a }
camelia rakudo-moar d8fa56: OUTPUT«5===SORRY!5=== Error while compiling /tmp/9URYq93zcM␤Invalid typename 'my' in parameter declaration.␤at /tmp/9URYq93zcM:1␤------> 3-> my7⏏5 $a = 0 { say $a }; try { say $a }␤»
psch m: -> $a = 0 { say $a }; try { say $a }
camelia rakudo-moar d8fa56: OUTPUT«5===SORRY!5=== Error while compiling /tmp/9kqswJDpWb␤Variable '$a' is not declared␤at /tmp/9kqswJDpWb:1␤------> 3-> $a = 0 { say $a }; try { say 7⏏5$a }␤»
timotimo a missing lexical is a compile-time error 15:17
jdv79 hmm, so how to d the p6 equiv of a p5 if ( my $lex = 42 ){lex here } no lex here?
super laggy here
timotimo you can indirectly access it via ::('$a')
psch m: -> $a = 0 { say $a }; try { say ::('$a') }
camelia ( no output )
moritz jdv79: if 42 -> { $lex 7 ... }
timotimo m: if 42 -> $lex { say $lex }; try { say ::('$lex') }
camelia rakudo-moar d8fa56: OUTPUT«42␤»
psch oh that way around
i did remember the pointy block...
timotimo hehe
jdv79 thanks 15:20
15:21 llfourn left 15:26 lizmat joined 15:27 cognominal joined
jdv79 are types and subsets lexical? 15:28
timotimo m: { subset Test } my Test $a
camelia rakudo-moar d8fa56: OUTPUT«5===SORRY!5=== Error while compiling /tmp/76syBQMgH9␤Strange text after block (missing semicolon or comma?)␤at /tmp/76syBQMgH9:1␤------> 3{ subset Test }7⏏5 my Test $a␤ expecting any of:␤ infix␤ infix stopper␤ …»
timotimo m: { subset Test }; my Test $a
camelia ( no output )
timotimo m: subset Test; my Test $a
camelia ( no output )
timotimo seems to be our-scoped
15:29 Hor|zon joined, llfourn joined
jdv79 o 15:31
timotimo that's only by default, of course 15:32
m: { my subset Test }; my Test $a
camelia rakudo-moar d8fa56: OUTPUT«5===SORRY!5===␤Type 'Test' is not declared␤at /tmp/DOdn1UjlR_:1␤------> 3{ my subset Test }; my Test7⏏5 $a␤Malformed my␤at /tmp/DOdn1UjlR_:1␤------> 3{ my subset Test }; my7⏏5 Test $a␤␤»
15:33 Hor|zon left
moritz types and constants are our-scoped by default, yes 15:37
15:43 beastd|2 joined 15:45 beastd left
jdv79 i would like to have an attr that i can set and get in both list and item context 15:46
not sure how to do taht in p6
moritz jdv79: example? 15:47
smls When subscripting an array or hash with :k, :v or :kv, any non-existent keys are silently skipped (whereas without a subscripting adverb, they cause an undefined value at the corresponding place in the output). Is this by design? 15:48
jdv79 a path attr, $.path.[2] = 'foo', $.path = "foo/bar/", $.path.list, $path.item
moritz create a Path class that's indexable, and return a Proxy attribute which does something different in STORE 15:49
jdv79 any examples? 15:50
the proxy docs and tests i saw were scant
smls m: my %x = a=>1, c=>3; dd %x<a b c>; dd %x<a b c>:v # example for undefined elements being silently skipped
camelia rakudo-moar d8fa56: OUTPUT«(1, Any, 3)␤(1, 3)␤»
dalek ast: b60e0a8 | lizmat++ | S32-list/ (3 files):
Test for Index rather than Int
 15:58
lizmat smls: yes that is by design 16:00
smls ok
lizmat m: my %x = a=>1, c=>3; dd %x<a b c>; dd %x<a b c>:!v # don't skip undefined elements
camelia rakudo-moar d8fa56: OUTPUT«(1, Any, 3)␤(1, Any, 3)␤»
lizmat smls: not the !v 16:01
smls interesting.
lizmat *note
smls This is not in the design docs, is it?
lizmat yes, it is...
smls oh, it's in S02 rather than S09 16:02
thanks
itz I seem to be able to create instances of roles directly with .new which I wouldn't have expected 16:03
lizmat itz: that's called punning 16:06
punning a role into a class 16:07
m: role A { }; say A.new.WHAT
camelia rakudo-moar d8fa56: OUTPUT«(A)␤»
lizmat an anonymous class is created in which the role is imported
itz ah thanks .. it was quite convenient but just surprised me 16:09
moritz m: role A { }; say A == A, A === A.new.WHAT
camelia rakudo-moar d8fa56: OUTPUT«Use of uninitialized value of type A in numeric context in any at src/gen/m-Metamodel.nqp:1677␤Use of uninitialized value of type A in numeric context in any at src/gen/m-Metamodel.nqp:1677␤TrueTrue␤»
moritz m: role A { }; say A === A, A === A.new.WHAT
camelia rakudo-moar d8fa56: OUTPUT«TrueTrue␤»
moritz oh, I guess doing === always gets the punned class, never the role 16:10
m: role A { }; say A.HOW.^name; say A.new.WHAT.HOW.^name
camelia rakudo-moar d8fa56: OUTPUT«Perl6::Metamodel::ParametricRoleGroupHOW␤Perl6::Metamodel::ClassHOW␤»
16:14 vendethiel left 16:21 spider-mario joined 16:28 perturbation joined 16:30 Hor|zon joined 16:34 vendethiel joined 16:35 Hor|zon left 16:39 brrt joined
jdv79 gist redispatches to perl for defined - weird; why not just gist recursively? 16:47
16:47 araujo joined, araujo left, araujo joined
lizmat jdv79: context? 16:48
16:51 brrt left 16:52 raiph left
jdv79 just reading docs for say 16:52
i mean gist 16:53
raydiak jdv79: that's just the default provided by Mu...another way to say "if there isn't a (more specific) .gist defined, fall back to .perl" 16:54
16:54 mr-foobar joined
lizmat multi method gist(Mu:D:) { self.perl } # what raydiak said 16:54
jdv79 i defined a gist and it didn't take 16:56
idk
lizmat m: class A { method gist { "foo" } }; say A 16:57
camelia rakudo-moar d8fa56: OUTPUT«foo␤»
16:58 pmurias left 17:02 lizmat left 17:06 raiph joined 17:13 llfourn left 17:40 HoppingMadMan joined
jdv79 why doesn't a proxy behave well with say? 17:51
skids wonders at the rationale behind checking if a runtime exception's backtrace is_runtime 17:55
Well, only one way to find out. Recompile without the check and see what explodes. 17:56
jdv79 what's wrong with this?: gist.github.com/anonymous/634aa3faecd763e77966 17:58
skids jdv79: wrong self, I think. 18:01
I did make that work once... 18:02
smls jdv79: You can do FETCH => -> ($) { $!a } I think, instead of using anon methods 18:06
itz googles "perl6 punning" and discovers a PPT about how the .sk have opium for Xmas dinner
smls jdv79: Or, do my $a := $!a; and then refer to $a in the Proxy
skids thinks the whole Proxy thing looks like it was designed in a hurry and never proofread. 18:08
jdv79 nothing i try works 18:09
now i get: No such private method 'a' for invocant of type 'A'
i feel like this should be easier
skids m: gist.github.com/skids/322c06d96ea46a2ce67c 18:11
camelia rakudo-moar d8fa56: OUTPUT«gist not found»
skids really?
smls jdv79: With my $a := $!a; it works for me
skids: That used to work 18:12
18:12 davido__ left
skids jdv79: The above gist WFM. You get to put the indetation back in :-) 18:12
18:13 davido__ joined 18:14 virtualsue left
jdv79 yeah, yours have $self 18:14
i don't get what the errors i was getting were 18:15
but at least something works - thanks
skids np
dalek line-Perl5: 03bba94 | (Stefan Seifert)++ | / (4 files):
Pass through scalar context when delegating to P5 base object

We now support calling Perl 5 methods in scalar context even when the Perl 5 module is partially written in Perl 6 using v6-inline:		 18:16
jdv79 let me guess, obj attr sets on construction don't go through accessors? 18:18
18:19 census left
smls yes 18:21
jdv79 does Moose do that ^H? 18:22
i know Rose::Object does
18:23 HoppingMadMan left 18:24 beastd|2 left 18:25 raiph left 18:26 raiph joined 18:27 mr_ron left, mr_ron joined 18:30 mr_ron left 18:31 Hor|zon joined 18:35 Hor|zon left
Sgeo When would someone use a sigilless variable? 18:40
And what's the point of sigils exactly?
skids They are noun markers.
jdv79 i think they are "contextualless" or something 18:41
moritz Sgeo: sigils enforce context; sigilless variables are good when you don't want that (mostly to pass something through unchanged)
skids And some of the docs about sigilless variables are a bit treacherous because they are not as simple as that.
for example: 18:42
m: my \a = 1; a = 3
camelia rakudo-moar d8fa56: OUTPUT«Cannot modify an immutable Int␤ in block <unit> at /tmp/vE4HPi1Fgi:1␤␤»
skids m: my \a = 1; a := 3
camelia rakudo-moar d8fa56: OUTPUT«5===SORRY!5=== Error while compiling /tmp/Lt7Ht6h65q␤Cannot use bind operator with this left-hand side␤at /tmp/Lt7Ht6h65q:1␤------> 3my \a = 1; a := 37⏏5<EOL>␤»
skids Where you actually use them is in special parameters like captures. 18:43
moritz fwiw doc.perl6.org/language/variables#Si..._variables makes it quite clear that you can only assign them once 18:44
Sgeo m: my \a = 1; \a := 3 18:45
camelia rakudo-moar d8fa56: OUTPUT«5===SORRY!5=== Error while compiling /tmp/NdcY7Aa7t3␤Cannot use bind operator with this left-hand side␤at /tmp/NdcY7Aa7t3:1␤------> 3my \a = 1; \a := 37⏏5<EOL>␤»
skids mortiz: Yeah I think I've saw some adventish stuff that threw them around casually, is all.
Sgeo m: my \a = 1; my \a := 3
camelia rakudo-moar d8fa56: OUTPUT«5===SORRY!5=== Error while compiling /tmp/4OiknKNW07␤Redeclaration of symbol a␤at /tmp/4OiknKNW07:1␤------> 3my \a = 1; my \a := 37⏏5<EOL>␤»
Sgeo Can't rebind them by any means :/ 18:46
18:46 espadrine joined
timotimo m: my \a = 1; my $b; a := $b; 18:46
camelia rakudo-moar d8fa56: OUTPUT«5===SORRY!5=== Error while compiling /tmp/B2gTo_u7lQ␤Cannot use bind operator with this left-hand side␤at /tmp/B2gTo_u7lQ:1␤------> 3my \a = 1; my $b; a := $b7⏏5;␤»
timotimo mhm
skids They are also commonly used in e.g. "my constant e"
And I guess the core uses them a lot to avoid making lots of multis. 18:47
dalek c: 41c9e4b | moritz++ | lib/Language/variables.pod:
Mention that sigilless vars do not enforce context
 18:49
moritz skids: avoiding both multis and containers
skids OK, well, if the is-runtime backtrace check is still protecting anything explody, no spectest has been written for it. 18:50
18:50 spider-mario left
skids I should probably check jvm though. 18:50
18:52 spider-mario joined 18:56 BooK joined, dagurval joined, rudi_s joined, [ptc] joined, StavroMueller joined, broquaint joined, timbunce joined 18:57 go|dfish joined 19:10 llfourn joined, virtualsue joined, lizmat joined, charsbar__ left, f3ew joined, jordman joined, SHODAN joined, Ulti joined 19:11 bobkare joined, ingy joined, raydiak joined, charsbar__ joined, oha joined, xxx joined, tadzik joined 19:14 llfourn left 19:16 ponbiki joined, ponbiki is now known as Guest44551 19:24 HoppingMadMan joined
ChoHag m: class BloodyThing { has Int $.whatsit = 42; submethod BUILD (:$!whatsit) { say "Hi!" } }.new 19:29
camelia rakudo-moar d8fa56: OUTPUT«Type check failed in assignment to '$!whatsit'; expected 'Int' but got 'Any'␤ in submethod BUILD at /tmp/M1EPynx_Oh:1␤ in block <unit> at /tmp/M1EPynx_Oh:1␤␤»
lizmat ChoHag: if you're using your own BUILD, the default setting doesn't work 19:31
m: class BloodyThing { has Int $.whatsit; submethod BUILD (:$!whatsit = 42) { say "Hi!" } }.new
camelia rakudo-moar d8fa56: OUTPUT«Hi!␤»
lizmat m: class BloodyThing { has Int $.whatsit; submethod BUILD (:$!whatsit = 42) { say "Hi!" } }.new.perl.say
camelia rakudo-moar d8fa56: OUTPUT«Hi!␤BloodyThing.new(whatsit => 42)␤»
ChoHag The spec suggests otherwise. 19:32
"Whether you write your own BUILD or not, at the end of the BUILD, any default attribute values are implicitly copied into any attributes that haven't otherwise been initialized."
lizmat well, yes, perhaps the spec needs changing :-) 19:34
ChoHag Or perhaps BUILD needs changing.
lizmat I'm just telling how it is now :-)
perhaps.... TimToady might want to chime in on this 19:35
ChoHag BUILD/new contains lots of magic. If you want to change any of it, no matter in how minor a way, you need to reproduce *all* of the rest of it.
skids m: class BloodyThing { has Int whatsit = 42; submethod BUILD () { say "Hi!" } }.new
camelia rakudo-moar d8fa56: OUTPUT«5===SORRY!5=== Error while compiling /tmp/Wi9lEpLX26␤Malformed has (did you mean to declare a sigilless \whatsit or $whatsit?)␤at /tmp/Wi9lEpLX26:1␤------> 3class BloodyThing { has Int whatsit7⏏5 = 42; submethod BUILD () { say "Hi!" } ␤»
skids erm.. darn howd that paste wrong. 19:36
m: class BloodyThing { has Int $.whatsit = 42; submethod BUILD () { say "Hi!" } }.new # The mere presense of :$whatsit sets it to any.
camelia rakudo-moar d8fa56: OUTPUT«Hi!␤»
skids *:$!whatsit. 19:37
Also, read RT#125437 before you beat your head against a wall. 19:38
ChoHag Seems to me there should be three interfaces - Inflexible magic; Partially-customisable magic; DIY magic - instead it seems the magic is entirely on or off.
At the least it could warn that my attribute's default values (and other ignored magic) is wasted. 19:39
If this was making me beat my head against a wall, there'd be neither wall nor head. 19:40
skids (Give the above RT I would not expect the behavior is necessarily representative of what "should happen.")
lizmat +1 on warning on not using the default value with custom BUILD
ChoHag Computers these days elicit little more than a sigh from me, no matter what they do.
lizmat +1 on doing the same on "is required" and having your own BUILD 19:41
ChoHag And while you're at it, very strong words saying that BUILD turns off the various construction magics and that they need replicating. 19:42
The spec just says, almost en passant, about needing to include settable arguments in BUILD's signature. 19:43
19:44 HoppingMadMan left
ChoHag rt.perl.org: "The workaround is to change your full name" 19:45
I think that about sums up how I feel about the modern internet.
That and the fact that rt.perl.org doesn't have a search or even a go to button. 19:46
19:46 telex left
ChoHag Because who'd want to do either of *those* weird things? 19:46
19:47 Hor|zon joined
skids rt.perl.org/Public/Bug/Display.html?id=125437 # since the bots seem to be on vacation 19:47
mst lizmat: hrm, we've found a lot of use from BUILD in M* stuff with attribute initialization happening normally first
lizmat mst: well, if you do your own BUILD, it's just as easy to add defaults, as I shown above 19:48
19:48 telex joined
lizmat now, with "is required", that's a different issue 19:48
skids m: class BloodyThing { has Int $.whatsit = 42; submethod BUILD () { say "Hi!" } }.new.perl.say 19:49
camelia rakudo-moar d8fa56: OUTPUT«Hi!␤BloodyThing.new(whatsit => 42)␤»
ChoHag s/add/duplicate/
lizmat m: class BloodyThing { has Int $.whatsit; submethod BUILD (:$!whatsit!) { say "Hi!" } }.new 19:51
camelia rakudo-moar d8fa56: OUTPUT«Required named parameter 'whatsit' not passed␤ in submethod BUILD at /tmp/Xw27tZhPLx:1␤ in block <unit> at /tmp/Xw27tZhPLx:1␤␤»
mst lizmat: so you just need to move the default somewhere else
lizmat: and the proposed warning should tell people to do that
19:52 Hor|zon left
skids m: class BloodyThing { has $.whatsit = 42.say; submethod BUILD () { say "Hi!" } }.new 19:53
camelia rakudo-moar d8fa56: OUTPUT«Hi!␤42␤»
19:54 HoppingMadMan joined
lizmat skids: interesting ... 19:55
skids m: class BloodyThing { has $.whatsit = 42.say; submethod BUILD (:$whatsit) { say "Hi!"; $!whatsit = 43 } }.new.perl.say
camelia rakudo-moar d8fa56: OUTPUT«Hi!␤BloodyThing.new(whatsit => 43)␤»
19:56 ShimmerFairy left
skids So if you want your default applied, just don't set $!foo 19:56
m: class BloodyThing { has $.whatsit = 42.say; submethod BUILD (:$whatsit) { say "Hi!"; } }.new.perl.say 19:57
camelia rakudo-moar d8fa56: OUTPUT«Hi!␤42␤BloodyThing.new(whatsit => Bool::True)␤»
lizmat skids++ 19:58
19:58 ggoebel left
skids As to whether an unset :$!foo should set the attribute to Any, that could be intentional, or could be the bug in the RT. 20:00
I'm guessing that behavior will not go away though even after the bug is fixed.
ChoHag m: class BloodyThing { has Int $.whatsit = 42; submethod BUILD (:$whatsit) { say "Hi!" } }.new(whatsit=>7).perl.say 20:02
camelia rakudo-moar d8fa56: OUTPUT«Hi!␤BloodyThing.new(whatsit => 42)␤»
20:02 ggoebel joined
skids You do have to do the assign. 20:03
lizmat ChoHag: BUILD is just receiving a named param that you're not doing anything with
ChoHag Yes.
lizmat m: class BloodyThing { has Int $.whatsit = 42; submethod BUILD (:$!whatsit) { say "Hi!" } }.new(whatsit=>7).perl.say
camelia rakudo-moar d8fa56: OUTPUT«Hi!␤BloodyThing.new(whatsit => 7)␤»
ChoHag So that's more magic down the shitter.
lizmat well, one could argue that a warning should be issued at compile time when a parameter with a name is not being used in the sub/method 20:04
m: sub a($) { } # don't warn
camelia ( no output )
lizmat m: sub a($a) { } # $a not used in sub a 20:05
camelia ( no output )
skids No, I would not like that.
ChoHag Is there any particular reason why bypassing the attribute's default in a custom BUILD is advantageous?
timotimo maybe if you have a multi submethod BUILD 20:06
skids lizmat: (Because often you want functions that throw away parameters.) 20:07
lizmat then just don't name the parameters
m: sub a($) {} 20:08
camelia ( no output )
skids Then you have to write a comment saying what this thing is you threw away.
20:09 ShimmerFairy joined 20:10 llfourn joined
ChoHag Regardless of absurd notions of purity, if an attribute has a default, and an attribute is being set by virtue of being a named parameter during customised object construction, unless a parameter in the signature overrides its attribute's default with its own that the attribute's default default be the default. 20:10
That there's no default supplied in the signature obviously makes it "right" that in that case the variable doesn't have a default value, but it makes even less sense for an attribute's default value to be ignored. 20:12
20:14 llfourn left
skids Maybe $!-twiggled params should default to their current value in general. 20:15
ChoHag Either has $.foo = "thing" should imply '= "thing"' in 'sub BUILD (:$!thing)', or a qualifier could/should be applied to drag in the default default, along the lines of 'sub BUILD (:$!thing with attribute-default)'
Personally I think 'with attribute-default' should be assumed (ie. the first option), but that probably makes overriding a default to nothing difficult. 20:16
And now I need to sleep, so I don't get to see the ensuing discussion, if any... 20:17
skids Or like I said, maybe :(:$!foo) would cause $!foo to remain untouched if it was not provided, in which case it would get the default unless you otherwise frobbed it.
ChoHag That's both options combined, which seems very perl-lik. 20:18
skids The one thing that is missing though is the ability to see what the "has" default is from inside BUILD. 20:19
20:19 ChoHag left
skids But there may be laziness reasons why you should not be able to do that. 20:19
20:26 diana_olhovik left
skids adds this conversation to RT#125437 since it is likely when this code will next be touched 20:32
20:32 darutoko left
skids ChoHag++, lizmat++ 20:32
20:41 araujo left, araujo joined 21:05 kaare_ left
lizmat wishes #perl6! a good night! 21:07
skids o/ 21:09
21:12 jjido joined, lizmat left 21:25 rindolf left 21:34 dustinm` joined 21:48 Hor|zon joined 21:52 Hor|zon left 21:54 TEttinger joined
dalek p/jvm-force-gc: 3fec6e0 | hoelzro++ | src/vm/jvm/ (2 files):
Create mapping and implementation for force_gc on JVM
 21:56
p/jvm-force-gc: a317d34 | hoelzro++ | src/vm/jvm/stage0/ (10 files):
Re-bootstrap JVM files for force_gc
p/jvm-force-gc: 426ca43 | hoelzro++ | src/vm/jvm/HLL/Backend.nqp:
Add force_gc implementation for JVM backend
22:11 llfourn joined 22:15 yqt joined, BenGoldberg left, BenGoldberg joined, llfourn left 22:16 BenGoldberg left, BenGoldberg joined 22:21 BenGoldberg left 22:29 virtualsue left 22:31 smls left, smls joined 22:33 diakopter joined
dalek c: b82c2cf | smls++ | lib/Language/subscripting.pod:
Add the /language/subscripting document

Some sections still need filling in. I've added POD comments to guide further work on it.		 22:34
22:35 leont joined
smls ^^ review welcome. 22:39
22:39 smls left 22:44 Zoffix joined 23:01 espadrine left 23:19 raiph left 23:24 cognominal left 23:28 jjido left 23:35 mohij left, raiph joined 23:43 jjido joined 23:47 RabidGravy left 23:49 Hor|zon joined, spider-mario left 23:53 Hor|zon left 23:55 jjido left 23:56 cognominal joined 23:58 vendethiel left 23:59 nys 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!