»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or rakudo:, or /msg camelia p6: ... | irclog: irc.perl6.org or colabti.org/irclogger/irclogger_logs/perl6 | UTF-8 is our friend!
Set by moritz on 22 December 2015.
BenGoldberg Does anyone know if there is, somewhere in moar, an identity function which takes a void* and returns the same? 00:24
BenGoldberg I want to turn a Routine into a Pointer, which can be stored in a CStruct, which would then get passed to a library routine... I'm 99% sure that I can make an identity function dwiw. 00:31
lookatme morning 00:33
rouking_ what's the difference between having a whenever block inside a react block or not? 02:56
moritz lichtkind: pong 05:48
yoleaux 22 Jun 2017 21:07Z <eveo> moritz: do you remember what "not raising with Junctions" means in this commit? We now have a situation that Junction == Numeric results in a Junction, but Junction != Numeric collapses the Junction and returns a Bool. That difference wasn't intended, was it? github.com/rakudo/rakudo/commit/4a...d8d88c7774
moritz .tell eveo the idea was that any(2, 3) != $x is interpreted as !(any(2,3) == $x), because the former is basically never what you want. That's why the != variant returns a Bool and the == a Junction 05:50
yoleaux moritz: I'll pass your message to eveo.
eveo moritz: the problem with it is that it's a special case that tries to guess "what I really want". .Bool collapses junctions, but now it's .Bool, or !=, 'cause there it's not really what you want 09:49
moritz: I'd still evaluate 5|6 != 42 in a Bool context, so I'm not really following what I'd not really want :/
m: say one(5, 6) != 6 09:50
camelia False
eveo m: say one(5, 6) == 6
camelia one(False, True)
eveo m: say so one(5, 6) == 6
camelia True
eveo Yeah, and it's wrong too
Xliff BenGoldbert: Let me know if you ever find an answer to that question. 10:04
^^ BenGoldberg
Xliff BenGoldberg: Wouldn't nativecast work, though? 10:06
tbrowder hi. #perl6! is the person responsible for perl6.party present? 10:26
moritz tbrowder: that would be eveo 10:27
eveo: the problem with != and junctions was that people were quite often surprised by the results 10:28
eveo moritz: currently result is wrong
I'd say that's quite surprising
moritz eveo: no, it's as designed and tested
it's not consistent with the rest, which you can count as surprising
tbrowder thanks, moritz. i was going to ask about the site details, but i just found the github link WAY at the bottom. (anyone else not fond of the single-page site format?) 10:30
eveo Make your own site then
tbrowder but the blog entries look nice.
moritz eveo: search for "Use of negative operators with junctions" in S03 10:31
eveo m: say one(5, 6, 6) !== 6 10:32
camelia True
eveo m: say one(5, 6, 6) !=== 6
camelia True
eveo S03:00
synopsebot6 Link: design.perl6.org/S03.html#line_00
tbrowder eveo: maybe i will, that's why i was looking for site construction details.
moritz S03:3327 10:32
synopsebot6 Link: design.perl6.org/S03.html#line_3327
eveo tbrowder: out of your own articles. The ones on perl6.party are copyrighted. 10:33
tbrowder sure, i understand
eveo moritz: that section entirely ignores that there's more than one type of junction 10:34
eveo "as designed and tested" for a poorly thought special case 10:34
eveo m: say all(5, 6) != 6 10:37
camelia True
eveo m: say none(5, 6) != 6
camelia True
eveo It only works for an `any` junction which coincidentally is the only junction shown in the spec 10:38
tbrowder i've been looking for a good blogging platform and attended Henry Van Styn's Rapi::Blog session at TPC NA 2017. looks like a good possibility.
eveo huggable: 6.d 10:43
huggable eveo, Proposals for 6.d language: github.com/perl6/specs/blob/master/v6d.pod
eveo m: say 2 ne 1 & 2 & 3 10:50
camelia True
eveo m: say 2 !~~ 1 | 2 | 3 10:51
camelia False
eveo m: say 2 !== 1 | 2 | 3
camelia False
eveo At least this shit is documented. 10:55
Perl 6 just lost one more piece of my heart due to this.
Consistently inconsistent.
perlpilot With $obj.can('method'), is there a way to find out where candidates came from without inspecting their signatures? 13:01
lollercopter perlpilot: yes, .package 13:03
m: say "".can("contains")».package 13:04
camelia ((Str) (Cool))
perlpilot ah, thanks. I knew there must've been something.
I even looked at the list of methods and didn't see that one. 13:05
perlpilot Hmm. That doesn't work too well with multi methods. (or it's making some sort of indictment of my intelligence ... 13:12
m: class C { multi method m() { }; multi method m($a,$b) { } }; C.can('m').[0].package
camelia ( no output )
perlpilot m: class C { multi method m() { }; multi method m($a,$b) { } }; C.can('m').[0].package.say 13:13
camelia (Dummy)
lollercopter You get a proto. You need to get to the actual .candidate you want
perlpilot sure, but that just means I need to adjust my idea of what .can does. 13:15
(I've been thinking of it as returning candidates)
lollercopter perlpilot: what are you trying to accomplish?
There's .cando method you can use to pick a candidate
(OTOH, it'd need the invocant) 13:16
perlpilot "accomplish" is a strong word. I'm just playing. I was hoping for something like $obj.can('method').classify(*.package.perl) to give me candidates per class 13:17
actually, I was imagining that there should be a modifier to .can that does that, but I don't know how useful it would be to the masses. 13:20
lollercopter perlpilot: you just need to classify the candidates themselves
m: class C { multi method m() { }; multi method m($a,$b) { } }; C.can('m')».candidates.flat.classify(*.package.perl).say
camelia {C => [m m]}
lollercopter m: class C { multi method m() { }; multi method m($a,$b) { } }; C.can('m')».candidates.flat.classify(*.package.perl)<C>.head.signature.say 13:21
camelia (C $: *%_)
lollercopter m: class C { multi method m() { }; multi method m($a,$b) { } }; C.can('m')».candidates.flat.classify(*.package.perl)<C>.tail.signature.say
camelia (C $: $a, $b, *%_)
lollercopter m: "".can('contains')».candidates.flat.classify(*.package.perl).say
camelia {Cool => [contains], Str => [contains contains contains contains]}
lollercopter No need for shoving "modifiers" all over the codebase. 13:22
perlpilot yeah, I suppose so. I just didn't want to type so much ;) 13:23
araraloren evening 14:03
[Coke] morning. 14:13
Geth ecosystem: 827738a5ef | (Curt Tilmes)++ | META.list
Add TCC, Tiny C Compiler, to ecosystem

See github.com/CurtTilmes/perl6-tcc		 14:16
[Coke] coconut-lang.org/
prodotiscus_ how to read a variable from STDIN? 14:19
rightfold prodotiscus_: That's a rather vague question.
Stdin provides a stream of bytes. 14:20
ctilmes prodotiscus_: Check out docs.perl6.org/routine/prompt
rightfold You can decode the bytes into some data structure.
prodotiscus_ `Enter your action (,) > Cannot flush handle: Failed to flush filehandle: Invalid argument` 14:33
I get this when i try `$action = prompt "Enter your action..";` 14:34
jnthn prodotiscus_: What Rakudo version do you have? 14:35
prodotiscus_ 2017.05-389-g326faed built on MoarVM version 2017.05-71-gad6ab26 14:36
jnthn Ah, not a release version
prodotiscus_ I use that on my laptop I'm now working now
what version can I use to avoid this?
jnthn There was a bug that caused that; it never appeared in a release, but there were several days where a HEAD built could get you it 14:37
The 2017.06 release version was fine in this regard
prodotiscus_ ok, thank you
jnthn Or latest if you like to live on the edge :) 14:38
raschipi Living on the edge means one will get things like that from time to time. 14:40
prodotiscus_ m: class A { has Str $!foo is required; method TOP { say "string, $!foo and something else" }; }; 14:57
camelia ( no output )
prodotiscus_ m: class A { has Str $!foo is required; method TOP { say "string, $!foo and something else" }; }; A(foo => "bar");
camelia Cannot coerce to A with named arguments
in block <unit> at <tmp> line 1
prodotiscus_ m: class A { has Str $!foo is required; method TOP { say "string, $!foo and something else" }; }; A.new(foo => "bar");
camelia The attribute '$!foo' is required, but you did not provide a value for it.
in block <unit> at <tmp> line 1
ctilmes prodotiscus_: Set private attributes with a submethod BUILD() 14:59
prodotiscus_: docs.perl6.org/language/classtut#Constructors
prodotiscus_ thank you ctilmes 15:00
m: class A { has Str $.foo is required; method TOP { say "string, $.foo and something else" }; }; A.new(foo => "bar");
camelia ( no output )
prodotiscus_ that's if I use public variable 15:01
ctilmes prodotiscus_: Right, it works win the public variable 15:05
*with 15:06
timotimo we should probably put TWEAK in there instead of BUILD 18:16
because if you have a BUILD, no default values or anything like that will be done for you 18:18
but TWEAK doesn't have that property
mst hence why I whined until somebody added TWEAK :) 18:19
moritz mst++ 18:23
samcv my git branch output is so nice now comfy.moe/llcmld.png 20:28
:-)
will provide code if anybody wants their git branch output to look like that :)
rindolf TimToady: hi 20:53
lizmat: hi
lizmat rindolf o/
rindolf lizmat: sup? 20:54
lizmat not much, with you?
rindolf lizmat: i closed an old KDE bug today as it was fixeed 20:55
lizmat: and I fixed a PySol bug that I reported
lizmat rindolf++
rindolf lizmat: thanks :-) 20:56
lizmat: none of these are particularly p6-related
lizmat it's still early :-) 20:57
ctilmes Some thoughts on eqv() I shared with TimToady et. al. at YAPC: curt.tilmes.org/2017-TPC-eqv
lollercopter ctilmes: +1 on that. The only worry is performance. And if perhaps we should leave the .perl's as is and only do the more expensive per-attribute eqv if the class wants to (e.g. if .eqv method is present but returns Nil) 21:07
lollercopter ctilmes: and sidenote: is-deeply uses "eqv semantics" not eqv op (it does Seqs differently for example). The propper way to run that test would've been cmp-ok $o1, &[eqv], $o2 21:09
ctilmes Note that .perl iterates over the attributes anyway...
lollercopter ctilmes: only by default. 21:10
ctilmes: and simply making a string out of them is probabpy cheaper than eqv'ing each
ctilmes I think you could construct scenarios where each was cheaper/more expensive 21:11
lollercopter m: my @a = ^1000; my @b = ^1000; for ^1000 { $ = @a eqv @b }; say now - INIT now 21:14
camelia 1.32335701
lollercopter m: my @a = ^1000; my @b = ^1000; for ^1000 { $ = @a eq @b }; say now - INIT now
camelia 1.13419479
ctilmes This is how Test runs is-deeply: github.com/rakudo/rakudo/blob/nom/...t.pm6#L638 21:15
$got eqv $expected
lollercopter ctilmes: look before that 21:16
ctilmes Yes, I see, but I still maintain most people would expect what I tried to work
lollercopter ctilmes: also, your example with sets is wrong. The only reason the perls differed is due to a bug in Setty.perl. It's meant to be ordered 21:17
ctilmes There are other scenarios where you pass classes (or subclasses) around, and would expect it to work
lollercopter ctilmes: I think I'm retracting that +1 now that I think of it. .perl is meant to roundtrip to give the original object back. Doing eqv on attributes is kinda pointless because you don't test any extra information .perl doesn't give you and you still don't solve the problen entirely since the private attributes aren't tested, you can still end up with uneqv eqv objects 21:19
ctilmes What do you think about .eqv as a method, and have the fallback check for that? 21:20
(also iterating the attributes and eqv-ing them can short cut on failure, whereas .perl-ing always iterates every attribute)
lollercopter True 21:21
lollercopter ctilmes: perhaps .EQV method? something less-likely to exist aacidentally for other reasons 21:23
\o
ctilmes yes, good point 21:23
tbrowder ctilmes: what about the fact that you can overload "==" and it works? 22:12
tbrowder it works on your class and least it's a workaround until eqv is fixed (if ever) 22:19
nebuchadnezzar new official Debian packages will be published soon 22:32
nebuchadnezzar really need a personal ZofBot ;-) 22:33
timotimo it's probably open source? 22:57
Voldenet m: { my $x=$0; $x ~~ s:g/<[\ \-]>/_/; $x.say } if "Hostname=This is a-test" ~~ /^Hostname\=(.*)$/ 23:22
camelia This_is_a_test
Voldenet is there any faster way to write this? 23:23
lollercopter Voldenet: yup 23:57
m: for ^10000 { { my $x=$0; $x ~~ s:g/<[\ \-]>/_/; } if "Hostname=This is a-test" ~~ /^Hostname\=(.*)$/ }; say now - INIT now
camelia 2.1162005
lollercopter m: for ^10000 { my $v = "Hostname=This is a-test"; if $v.starts-with: 'Hostname=' { my $x = $v.substr(9).subst(' ', '_', :g).subst('-', '_', :g); } }; say now - INIT now
camelia 0.1612127
lollercopter m: my $v = "Hostname=This is a-test"; $v.substr(9).subst(' ', '_', :g).subst('-', '_', :g).say if $v.starts-with: 'Hostname=' 23:58
camelia This_is_a_test
lollercopter Looks right

Powered by Cro and the Raku® Programming Language.

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

Thank you!