»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.
TimToady m: class Foo { method foo(--> self) {}}; Foo.new.foo.say 00:00
camelia rakudo-moar de8024: OUTPUT«5===SORRY!5=== Error while compiling /tmp/CerzdjtmP3␤Type 'self' is not declared␤at /tmp/CerzdjtmP3:1␤------> 3class Foo { method foo(--> self7⏏5) {}}; Foo.new.foo.say␤»
TimToady awww
maybe we should special-case that for the people who inexplicably like fluent programming 00:01
me, I can't stand methods that don't know which way the data is flowing through them 00:02
TimToady is not particularly interested in making fluent programming easy though, but --> self is pretty good documentation that you've got a mutator 00:03
I wonder if it's possible to move the declaration of self up a little... 00:05
00:05 mattprelude left
skids
.oO(maybe an alternate method dispatcher to '.')
 00:05
TimToady like, say, .=
which could probably optimize more if it sees a --> self 00:06
well, self isn't exactly a compile-time value, so just declaring it earlier won't help, likely 00:07
00:09 salv0 left
ShimmerFairy I think chaining mutating method calls is really only useful where you find your self writing a lot of methods that don't return anything, because there you can say "why not return C<self>?". Otherwise, useful return values tend to trump fancy method chaining, I feel :) 00:09
00:10 mr_ron left 00:12 salv0 joined
pink_mist well of course if you're returning something else that trumps method chaining ... method chaining is for when you aren't returning something else 00:14
ShimmerFairy TimToady: another issue could be if someone tried method foo(Class \SELF: $param --> SELF) { ... } , so the initial impulse to just special-case 'self' there is obviously not adequate :) 00:16
m: my @a = 1,2,3; my $s = @a; for $s { say "A" }; $s.map: { say "B" } 00:18
camelia rakudo-moar de8024: OUTPUT«A␤B␤B␤B␤»
ShimmerFairy huh, I thought single-arg was supposed to fix that :(
00:18 yqt left
TimToady the GLRR pays attention to $ there 00:19
so you need @$s or so 00:20
m: class Foo { method foo(\SELF: --> SELF) {}}; Foo.new.foo.say
camelia rakudo-moar de8024: OUTPUT«===SORRY!===␤No compile-time value for SELF␤»
TimToady well, when we get --> $reval that should fall out pretty easily 00:21
ShimmerFairy that's unfortunate. I understand that $ protection was the sanest way to handle the problems people were having, but on the other hand I still think people should've been using given $s { } instead, in this one case :) 00:22
00:24 kanishka left 00:28 kanishka joined 00:30 leont left
TimToady it wasn't just that one case 00:30
@array.push: $element 00:31
we're not gonna define two different method names for every method that has the issue
ShimmerFairy Like I said, making $ protect against single-arg certainly seems like the best solution to the problem. If it means losing perfect for $s <-> $s.map equivalence, then oh well :) 00:33
00:39 adu left, sufrostico joined
zostay m: sub f(Pair:D $v) { ++$v.value }; my $x = 10; say f('x' => $x); say f('y' => 42); 00:44
camelia rakudo-moar 0024b0: OUTPUT«11␤Parameter '$a' expected a writable container, but got Int value␤ in sub f at /tmp/kYON57D75i:1␤ in block <unit> at /tmp/kYON57D75i:1␤␤»
zostay how do i tell if $v.value has been bound to an immutable value or to a variable?
ShimmerFairy m: sub f(Pair:D $v) { $v.value.VAR.WHAT }; my $x = 10; say f('x' => $x); say f('y' => 42); # you probably shouldn't check like this, though 00:47
camelia rakudo-moar 0024b0: OUTPUT«(Scalar)␤(Int)␤»
zostay yeah, that probably doesn't work well... i think i'd catch the failure and react instead 00:49
i'll go with that unless someone has a better answer 00:50
ShimmerFairy zostay: I think the best solution would be to do { ++$v.value; CATCH { #`(handle error here) } } 00:51
00:53 TEttinger left 01:05 adu joined
awwaiid m: use MONKEY-TYPING; augment class List { method first { self[0] } } ; (1, 2, 3).first # muahaha 01:06
camelia ( no output )
awwaiid m: use MONKEY-TYPING; augment class List { method first { self[0] } } ; say (1, 2, 3).first # muahaha
camelia rakudo-moar 0024b0: OUTPUT«1␤»
01:12 dnmfarrell left 01:30 vendethiel joined 01:33 tokuhiro_ joined 01:37 tokuhiro_ left
zostay whoo it even has a nice neat X::Assignment::RO exception to work from ++ 01:46
01:54 vendethiel left 02:09 espadrine left 02:10 sufrostico left 02:19 nightfrog left 02:24 nightfrog joined, adu left 02:33 aborazmeh joined 02:34 aborazmeh left, aborazmeh joined, tokuhiro_ joined
flussence
.oO( I see in the backlog I'm not the only one interested in gentoo integration :D )
 02:38
02:39 tokuhiro_ left
ShimmerFairy flussence: I couldn't even begin to actually _do_ it, mind you, but I do think it'd be neat if from the gentoo angle we could say "_our_ module system doesn't make gentoo use an external tool" :P 02:39
02:40 noganex_ joined
flussence it'd be pretty neat to have a CUR::Local::VDB or something along those lines to make p6 aware of system-installed stuff - but I've no idea where to begin with something like that either. 02:43
02:43 noganex left
JimmyZ .tell lizmat Do we need s/trait_mod/trait-mod/g too? I see a lot of trait_mod:<x> used. 02:43
yoleaux JimmyZ: I'll pass your message to lizmat.
02:46 adu joined
Timbus_ is `x` in this case any verb? or are we still limited to is, does, etc 03:23
03:29 kanishka left
dalek osystem: ce32c76 | (Sterling Hanenkamp)++ | META.list:
Remove KnottyList from the Perl6 ecosystem

Since Perl 6's own Pair know handles binding, KnottyPair is mostly redundant.		 03:41
c: 8e3c7d4 | (David Farrell)++ | lib/Type/Iterable.pod:
fix typos
 03:50
c: 4801399 | (David Farrell)++ | lib/Type/Iterable.pod:
Merge pull request #135 from dnmfarrell/master

fix typos
03:51 zostay left, zostay joined 04:02 rurban joined
dalek kudo/nom: c8ec299 | TimToady++ | src/core/Rat (2 files):
speed up Rat addtion by reducing lazily

About 20% faster when adding a bunch of things with the same denominator. We do this by being lazy about reducing the fractions till we need them reduced, that is, when we're about to display them in a way that reveals the
  .nude bits, or compare them in a way which needs to know integer status.		 04:11
04:23 Sgeo left 04:26 rurban left 04:30 pat_js` left
dalek kudo/nom: 0132b65 | TimToady++ | src/core/StrDistance.pm:
bit of GLR residue
 04:31
04:34 telex left 04:36 tokuhiro_ joined, telex joined 04:40 tokuhiro_ left
dalek ast: 90c84b8 | skids++ | S02-literals/heredocs.t:
Add tests for RT#117705
 04:45
TimToady couple of mysterious failures with tr/// that might or might not be my fault, though they look susiciously GLRish 04:52
but backing out various of my patches doesn't seem to fix it, so I dunno...
04:58 jack_rabbit_ joined 05:09 Sgeo joined, yuvipanda left 05:11 BenGoldberg left 05:14 Sgeo left 05:22 Sgeo joined 05:29 tokuhiro_ joined 05:33 adu left, tokuhiro_ left
ShimmerFairy blog post! shimmeryfairy.wordpress.com/2015/0...ally-done/ 05:34
05:34 TEttinger joined
JimmyZ oh, long post/gist, ShimmerFairy++ 05:35
but, why STRING? not String? 05:40
ShimmerFairy JimmyZ: that just takes advantage of the ::T syntax for using the same type as a constraint later on (as I understand it). 05:41
m: sub foo(Real ::R $a, R $b) { say 'OK!' }; foo(1, 2);
camelia rakudo-moar 0132b6: OUTPUT«OK!␤»
ShimmerFairy m: sub foo(Real ::R $a, R $b) { say 'OK!' }; foo(1, 2.0);
camelia rakudo-moar 0132b6: OUTPUT«Type check failed in binding $b; expected Int but got Rat␤ in sub foo at /tmp/CEZJ2A8OTX:1␤ in block <unit> at /tmp/CEZJ2A8OTX:1␤␤»
JimmyZ oh ,it is a type.
*type name 05:42
ShimmerFairy I used all-caps because the single-letter usage does, and because it makes it look like a funny type (so you're not so quick to think it's a "real" type) :)
JimmyZ: basically, I tried being as explicit as possible in writing those function signatures, so a potential implementor reading could see what the function expects to get and what it should return :) 05:46
JimmyZ ShimmerFairy: fair enough :) 05:50
05:56 skids left 06:18 szabgab joined
szabgab Good time! 06:18
rakudo: my %h; say %h{'a'}.elems;
camelia rakudo-moar 0132b6: OUTPUT«1␤»
szabgab what did I do there ? Why is that return 1 ? 06:19
rakudo: my %h; if not %h{'a'}:exists { %h{'a'} = [] }; say %h{'a'}.elems; 06:22
camelia rakudo-moar 0132b6: OUTPUT«0␤»
szabgab and probably this is what I wanted...
06:25 rindolf joined
ShimmerFairy m: my %h; say %h<a>.perl 06:26
camelia rakudo-moar 0132b6: OUTPUT«Any␤»
06:31 helloworldlang joined
helloworldlang p6: my $lastval = 4; sub lastval is rw { $lastval }; dd $lastval; lastval() = 5; dd $lastval 06:33
camelia rakudo-moar 0132b6: OUTPUT«Int $lastval = 4␤Int $lastval = 5␤»
helloworldlang p6: my $lastval = 4; sub lastval is rw { return $lastval }; dd $lastval; lastval() = 5; dd $lastval
camelia rakudo-moar 0132b6: OUTPUT«Int $lastval = 4␤Cannot assign to a readonly variable or a value␤ in block <unit> at /tmp/tmpfile:1␤␤»
06:38 _mg_ joined
ShimmerFairy helloworldlang: you currently need return-rw to make the explicit return work right if you want is rw (and yes, it's not optimal) 06:38
helloworldlang p6: my $lastval = 4; sub lastval is rw { return-rw $lastval }; dd $lastval; lastval() = 5; dd $lastval 06:40
camelia rakudo-moar 0132b6: OUTPUT«Int $lastval = 4␤Int $lastval = 5␤»
06:45 _mg_ left 06:49 szabgab left 07:06 aborazmeh left
JimmyZ .tell Timbus_ is/does/of/returns/ etc, see traits.pm 07:20
yoleaux JimmyZ: I'll pass your message to Timbus_.
07:29 darutoko joined 07:31 tokuhiro_ joined 07:35 tokuhiro_ left
Timbus_ JimmyZ, design.perl6.org/S14.html#line_389 07:37
yoleaux 07:20Z <JimmyZ> Timbus_: is/does/of/returns/ etc, see traits.pm
07:38 vendethiel joined
Timbus_ 'feels like' it implies anything can be declared, but i guess it doesn't say it. 07:40
07:46 araujo_ left 07:49 araujo joined 08:06 bjz left, bjz joined 08:18 espadrine joined 08:22 domidumont joined 08:28 domidumont left 08:29 domidumont joined 09:00 vendethiel left, spider-mario joined 09:04 jack_rabbit_ left
masak antenoon, #perl6 09:12
DrForr Mere *minutes* after noon here... 09:18
09:21 samarkand joined
samarkand Hi, how can I add a dependency (third-party library) to my perl 6 project? 09:22
09:24 nowan joined
moritz samarkand: I don't think non-Perl 6 dependencies are tracked 09:28
samarkand: nor have I heard of a good way to do it yet
samarkand so if there's a perl 6 library, there's no way to use it in my perl 6 project, is that what you're saying? 09:29
@moritz 09:30
moritz samarkand: oh no, I misunderstood 09:31
samarkand: sorry, I thought you meant non-Perl6 dependencies by "third-party" :-)
DrForr Oh, third-party library as in other perl6 library. Isn't that just a dependency line in META.info?
samarkand only perl 6
moritz samarkand: you can specify dependencies in your META.info
see doc.perl6.org/language/modules 09:32
samarkand in META.info, is that it? and how to make them actually installed? like in ruby you'd say "bundle install"
DrForr panda track the dependencies.
*tracks 09:33
samarkand thank you, understood.
moritz github.com/tadzik/Bailador/blob/ma.../META.info for an example
samarkand thanks.
09:33 samarkand left 09:35 samarkand joined
samarkand I've forgotten to ask. Is that only modules.perl6.org? How about installing from github? 09:36
or from CPAN? Possible?
09:38 nowan left
JimmyZ samarkand: github.com/perl6/ecosystem 09:38
samarkand thx 09:39
09:39 samarkand left 09:41 nowan joined 09:43 leont joined
ShimmerFairy moritz: it's META6.json, not META.info, though 09:47
helloworldlang panda can install module from git://* uri 09:55
can perl6 run faster then perl5 in futrue ? 10:00
DrForr That's the hope... 10:01
10:04 AlexDaniel joined 10:05 xfix joined, xfix left, xfix joined 10:08 helloworldlang left
Woodi hi today :) 10:09
just this hit me - could we define here some small, simple, easy to implement language and name it PerlScript ? ;) 10:11
DrForr Just take JavaScript and add sigils? 10:13
Woodi but about "fluent programming" or APIs... I'm a bit surprised discussions about it do not mention other discusions on "Do not use method chaining!" topic
DrForr: I don't know, just idea was flying by ;)
DrForr The key would be to get it installed in browsers. 10:15
Woodi and looks "fluent APIs" are just method chaining but spirit of it it/quality of API is decided in domain space, not syntax or things like that...
maybe I should just learn JavaScript :) 10:16
DrForr That was ovid's talk? I haven't seen the talk, I was ... somewhat occupied. 10:17
Woodi what is purpose of "is raw" or "is Storable", from user perspective ? passing some data-blob via 2+ calls ? 10:19
DrForr Simpler way to serialize attributes? 10:23
Woodi if it helps somehow I think ppls buy 2x more sliced&packed bread ;)
DrForr: oops, probably is raw and Storable/is rw are 2 separate things... 10:24
DrForr Yep, 'raw' has no relation to 'rw' as far as I know. 10:29
10:31 zakharyas joined 10:36 leont left 10:53 spider-mario left
timotimo "is raw" causes no re-containerization at all, "is rw" causes no "readonly-recontainerization" 10:54
Woodi timotimo: but why user wants to use .raw ? at least examples needed... 10:56
psch Woodi: "is raw" is the trait-equivalent of using a sigilless parameter 10:57
"sub f(\a) { ... }" <=> "sub f($a is raw) { ... } "
Woodi actually, \a is very sigil-full ;) 10:58
timotimo only in declaration, not in use 10:59
you use "is raw" if you want to pass something through without imposing anything container-ish on it
Woodi timotimo: why anything need to be artificially packed in additional container ? 11:00
I know that it is VM and everything is kind of PMC... but one or the other is strange :) 11:01
jnthn Because $[1,2,3] shouldn't lose its itemization.
yoleaux 25 Sep 2015 20:29Z <nine> jnthn: the Slang.new failure started appearing in the range 94bc1a9..65129cf
25 Sep 2015 20:45Z <nine> jnthn: or...not. When I checkout 94bc1a9 and rebuild nqp/Moar I still get "Method 'new' not found for invocant of class '$?CLASS'". But with rakudo from this afternoon it worked.
25 Sep 2015 22:40Z <lizmat> jnthn: please read gist.github.com/lizmat/f3807956c354c14902a3 for first stab at CURLI revival
Woodi jnthn: ah, to pass something by. 11:02
11:02 TEttinger left
timotimo scalar containers is kind of how we do things 11:04
jnthn .tell nine Note that 663e71a4e0 and later in Moar require the change done in 44ad04a88 in Rakudo
yoleaux jnthn: I'll pass your message to nine.
jdv79 RabidGravy: oddly www.punoftheday.com/cgi-bin/todayspun.pl hangs up H::UA 11:08
but it finished. i think after an wait or sommething...
*finishes
11:09 RabidGravy left
jnthn .tell lizmat Read it, will ponder, thanks. :) 11:12
yoleaux jnthn: I'll pass your message to lizmat.
lizmat messages! 11:19
yoleaux 25 Sep 2015 23:09Z <tony-o> lizmat: does this mean that i'm open to putting whatever i want in a CURLI manifest file? how does CURLF store information about multiple modules with :auth and :ver? how do i handle *not* being able to write to a precomp directory, is that a total failure for installing the module? this doesn't seem to imply that a CURLF and CURLI should have the same interface, does the loader behave differently depending on
which on
25 Sep 2015 23:11Z <ShimmerFairy> lizmat: left a comment on your gist, hope it helps! :)
02:43Z <JimmyZ> lizmat: Do we need s/trait_mod/trait-mod/g too? I see a lot of trait_mod:<x> used.
11:12Z <jnthn> lizmat: Read it, will ponder, thanks. :) 11:20
nine Surprise: trying to turn Bool into a proper Enumeration is full of bootstrap issues 11:32
yoleaux 11:04Z <jnthn> nine: Note that 663e71a4e0 and later in Moar require the change done in 44ad04a88 in Rakudo
psch S99:Circularity_saw 11:33
psch stops snarking
nine: i've poked at that as well, quite some time ago 11:34
nine psch: any pointers at what cannot possibly work?
11:34 tokuhiro_ joined
masak "hopefully in a way that the resulting splice will be very difficult for the end-user to notice in day-to-day use." -- I think the general landscape of hope is slightly better than this description implies. 11:36
the place where the circularity saw cuts matters a lot to the implementor, but for the user it can usually be made either 100% invisible or very nearly so. 11:37
psch nine: nothing helpful comes to mind, unfortunately
masak (I liked the rest of design.perl6.org/S99.html#circularity_saw -- nicely done, whoever wrote it)
11:39 tokuhiro_ left
lizmat commented on gist.github.com/lizmat/f3807956c354c14902a3 ShimmerFairy tony-o nine 11:43
JimmyZ: I think at least at the Perl 6 level, it would need to be called trait-mod 11:44
but this is one of the things all the way down into nqp itself 11:45
jnthn will not be happy with such a change
m: say +StrDistance.new(:before("foo"),:after("zoo")) # shouldn't that just work ? 11:48
camelia rakudo-moar 0132b6: OUTPUT«Cannot make a Block object using .new␤ in block <unit> at /tmp/RPwa8ZBCR8:1␤␤»
11:49 FROGGS joined
FROGGS lizmat: will read now 11:50
yoleaux 25 Sep 2015 21:11Z <hoelzro> FROGGS: that S26 example you posted that looked buggy is fine; you need a whitespace between '#|' and your comment
25 Sep 2015 22:40Z <lizmat> FROGGS: please read gist.github.com/lizmat/f3807956c354c14902a3 for first stab at CURLI revival
FROGGS lizmat: I very much disagree about the statements about CURLF 11:51
Woodi just found R* 2015.09.tgz ! o/ # impatience-tortures: off :)
lizmat FROGGS: could you leave it as a comment?
FROGGS lizmat: and I think we already agreed on a need of an install method of CURLF, which also implies that it should support precomp 11:52
lizmat: will do
lizmat okidoki! :-)
[Tux] has RT#124082 been resolved? Really? WOW 11:53
This one is new: 11:54
Parameter '$self' expected a writable container, but got Str value
$name ~~ s{"-"} = "_"; 11:55
psch m: my $n = "foo"; $n ~~ s{"f"} = "F"; say $n 11:56
camelia rakudo-moar 0132b6: OUTPUT«Foo␤»
psch m: my $n := "foo"; $n ~~ s{"f"} = "F"; say $n
camelia rakudo-moar 0132b6: OUTPUT«Parameter '$self' expected a writable container, but got Str value␤ in block <unit> at /tmp/ZiVu90D9Gc:1␤␤»
lizmat afk for a few hours& 12:09
12:13 yqt joined
FROGGS .tell lizmat I commented: gist.github.com/lizmat/f3807956c35...nt-1582671 12:25
yoleaux FROGGS: I'll pass your message to lizmat.
12:26 spider-mario joined 12:31 aristotle joined
aristotle TimToady: a thought I had recently is that one day it may become a good idea to remove all the then-formerly helpful error message hints à la “where you trying to write Perl 5 here?” 12:35
c.f. nntp.perl.org/group/perl.perl5.porters/231036
(the thread is about a bug in the Perl 5 regexp compiler’s attempt to parse patterns for a POSIX regexp construct solely to warn that it isn’t supported) 12:37
FROGGS moritz: HTTP::Easy needed a fix, which is already in
aristotle: yes, I also think that the P5ism catchers can go away at some point 12:39
ShimmerFairy I don't think they're quite useless just yet, but they definitely aren't meant to be forever :) 12:40
aristotle certainly not useless now
FROGGS nods
moritz fwiw some of the (like for =~, and . for string concatenation) are often useful to me 12:41
because I switch a lot between 5 and 6, and sometimes my brain lags behind
FROGGS aye 12:42
let's just rethink their usefulness in two decades :o)
ShimmerFairy None of them are useful for me :P 12:43
aristotle as I said, this is about one day someday in the future. Perl 5 is far that point in its cycle; for Perl 6 it’s a long ways off. that’s what I wrote in the post I linked to
ShimmerFairy The only ones I have an issue with, though, are where they break things out of superstitious worry of P5isms. The thing I'm thinking of:
m: say <>
camelia rakudo-moar 0132b6: OUTPUT«5===SORRY!5=== Error while compiling /tmp/OSyuRs0jnu␤Unsupported use of <>; in Perl 6 please use lines() to read input, ('') to represent a null string or () to represent an empty list␤at /tmp/OSyuRs0jnu:1␤------> 3say <7⏏5>␤»
ShimmerFairy m: my %h; say %h<> 12:44
camelia rakudo-moar 0132b6: OUTPUT«␤»
ShimmerFairy Even better considering the zen slice. I don't see any good reason <> should be more severe than a warning.
masak nine: re turning Bool into an enumeration: the reason it isn't one this late in the game is that it's been full of bootstrap issues this whole time.
FROGGS .tell hoelzro I agree that there is whitespace missing afetr the #|, but why does the absence of the #| change anything?
yoleaux FROGGS: I'll pass your message to hoelzro.
masak nine: I also don't have any answers, but I will consider the day we succeed in making it one a fairly big watershed moment :) 12:45
aristotle FROGGS: I’m bringing it up because it seems that it hasn’t crossed anybody’s mind when it comes to Perl 5 – or maybe the people who thought of it at the time are longer around –, so those porting aids and their usefulness never did get revisited (/cc TimToady) 12:47
FROGGS yeah, long forgotten probably 12:48
12:48 rurban joined
[Tux] test 50000 34.845 34.738 12:49
test-t 50000 35.543 35.436
RT#124082 looked fixed because the error message changed 12:52
psch m: Num.range.say 12:57
camelia rakudo-moar 0132b6: OUTPUT«Method 'range' not found for invocant of class 'Num'␤ in block <unit> at /tmp/fqWlgOlyDj:1␤␤»
psch m: Num.Range.say
camelia rakudo-moar 0132b6: OUTPUT«-Inf..Inf␤»
12:58 mr_ron joined
psch [Tux]: unless you're about Inf !~~ Int..? 12:58
+talking :l 12:59
[Tux] tux.nl/Talks/CSV6/test-t.png 13:00
13:01 rurban left
[Tux] m: my Int $i; $i = Int.Range.max; 13:01
camelia rakudo-moar 0132b6: OUTPUT«Type check failed in assignment to $i; expected Int but got Num␤ in block <unit> at /tmp/I1RROPjblb:1␤␤»
[Tux] the error changed from "expected 'Int' but got 'Num'" to "expected Int but got Num" 13:02
the quotes were dropped
13:02 Begi joined
psch [Tux]: so the actual bug is RT #61602, not RT #124082? 13:03
Begi Hey ! What is the easiest way to code a chat application?
ShimmerFairy Yeah, that's the infamous Inf problem right there (and the Int.Range.max example isn't even that new :P) 13:04
moritz Begi: the easiest way is to use an existing one 13:05
Begi moritz: educational purposes, I'd build my own 13:06
btyler_ Begi: here's a non-blocking TCP chat* server in ~20 lines: gist.github.com/kanatohodets/8ffb1...04c80225cc
[Tux] psch, maybe. That got rejected though
btyler_ * preferred chat client: telnet
13:07 zakharyas1 joined, zakharyas left
btyler_ (I hacked that up for a lightning talk at YAPC::EU; you might want to do something more elaborate, like using a real chat protocol instead of bytes on TCP :)) 13:07
ShimmerFairy [Tux]: it's a mistake that it's rejected; it's an issue, stalled as it may be
psch [Tux]: it got moved to the specs github issue, i agree with ShimmerFairy that it shouldn't be outright rejected, but i don't know if RT has "stalled" or something similar 13:08
oh
ShimmerFairy psch: it does, actually. I think it was the only stalled ticket left (or at least one of them) :)
psch RT has stalled, yeah
i just saw [Coke] changing it from stalled to open to rejected
[Tux] I had several discussions about it with lizmat and masak and I *do* understand the problem 13:09
I'll change the RT number in my bug tester to prevent future confusion
ShimmerFairy .tell [Coke] I believe it's a mistake for you to reject RT #61602 -- it may not have a solution in the foreseeable future, but rejection should apply to tickets we _know_ are invalid, not tickets about the still-decided. 'stalled' is more appropriate for it, like it was 13:10
yoleaux ShimmerFairy: I'll pass your message to [Coke].
13:11 zakharyas1 left 13:14 Begi left 13:17 rurban joined 13:24 diana_olhovik joined 13:25 brrt joined
masak "The main concern among us Angular developers recently has been the future of our existing Angular 1 apps." -- p5:p6 :: ng1:ng2 -- www.codelord.net/2015/09/10/angular...t-we-know/ 13:31
moritz m: gist.github.com/moritz/5cda6371ab3cfec3254e
camelia rakudo-moar 0132b6: OUTPUT«Proxy␤Method␤»
moritz why does the last line print Method here? 13:32
and not Int?
masak good question. 13:33
does just `$mv.varies` print 6?
moritz masak: no, it says <anon>
13:34 kaare__ left, RabidGravy joined
ShimmerFairy moritz: Well, it's because you mapped those keys to Method objects, no? 13:34
masak sounds wrong to me. 13:35
ShimmerFairy: isn't that the way you implement a Proxy, though?
ShimmerFairy masak: I've never seen the method keyword used with it, I don't think. Just plain closures
masak you give FETCH and STORE a callable each. doesn't mean they get exposed.
ShimmerFairy: the presence of the `method` keyword there is practically irrelevant. 13:36
13:36 tokuhiro_ joined
moritz ShimmerFairy: the 'method' keyword just means they get an invocant 13:36
masak m: my &c1 = sub ($obj, $x) { say "$obj: $x" }; my &c2 = method ($obj: $x) { say "$obj: $x" }; say c1("OH HAI", 42); say "OH HAI".&c2(42) 13:37
camelia rakudo-moar 0132b6: OUTPUT«OH HAI: 42␤True␤OH HAI: 42␤True␤»
masak m: my &c1 = sub ($obj, $x) { say "$obj: $x" }; my &c2 = method ($obj: $x) { say "$obj: $x" }; c1("OH HAI", 42); "OH HAI".&c2(42)
camelia rakudo-moar 0132b6: OUTPUT«OH HAI: 42␤OH HAI: 42␤»
masak ShimmerFairy: ^
it's only a matter of intent, not a matter of differing semantics.
moritz huh 13:38
but changing it to a sub instead helps indeed
wtf?
masak ok, bug
masak submits rakudobug
it's probably special-cased where it oughtn't be
moritz lib/NativeCall/Types.pm also uses Proxy with method 13:39
jdv79 i think i ran into that a few months ago and just ignored it when someone showed me the working one
masak let's fix it instead :P
moritz does that mean the ones in Nativecall::Types are also broken? And why has nobody noticed that, if true?
masak ++moritz # for finding out 13:40
13:40 tokuhiro_ left 13:50 brrt left
hoelzro FROGGS: are you talking about your class Foo { method bar { } #= baz␤}; say Foo.^find_method('bar')[0].WHY example? 13:58
yoleaux 12:44Z <FROGGS> hoelzro: I agree that there is whitespace missing afetr the #|, but why does the absence of the #| change anything?
hoelzro m: class Foo { method bar { } #= baz␤}; say Foo.^find_method('bar')[0].WHY
camelia rakudo-moar 0132b6: OUTPUT«(Any)␤»
moritz hoelzro: fwiw a nice syntax would be Foo.WHY('bar')
hoelzro I don't know why that doesn't work with camelia, but it works on a local Perl 6
moritz: that's not a bad idea 13:59
jdv79 ok, so a .pm file with no use v6 and no modeline
is it acceptable to not consider that p6?
13:59 brrt joined
jdv79 Algorithm::Elo for instance 13:59
moritz jdv79: well, I think that's a pretty simple heuristic, but often good enough 14:00
brrt .tell jnthn using nqp i can make a reliable inlined-notinlined pair using a getlex to another frame
yoleaux brrt: I'll pass your message to jnthn.
jdv79 i'm asking when its not
brrt .tell jnthn i can't repeat the issue outside of the compiler yet, and can't see anything obviously wrong with it 14:01
yoleaux brrt: I'll pass your message to jnthn.
jdv79 context is cpanish things but i think in general its a valid question
hoelzro jdv79: what's this heuristic for? 14:02
jdv79 determining if a file is p6 pod extractable
14:02 diana_olhovik_ joined
jdv79 in this very specific case 14:02
but is it a bad thing, i guess i'm asking roudabout, to enforce the simple heuristics 14:03
14:03 diana_olhovik left
moritz well, if a repo has a META.info, you could look into is "provides" section 14:03
jdv79 huh? 14:04
hoelzro ah 14:05
well, S02 says that use v6 is an indicator, or if the first non-comment non-empty line has /'unit'? ['module' || 'class' || 'grammar' || 'role']/
I think that listing is flexible
if the first non-comment, non-empty line of a file contains a module, class, grammar, or role declaration
you can consider it Perl 6
jdv79 oh right, forgot about that. duh.
thanks 14:06
hoelzro np
moritz jdv79: github.com/tadzik/Bailador/blob/ma.../META.info if you look into "provides", it lists all Perl 6 modules there 14:07
that way you don't need a heuristic at all
colomon moritz++
14:07 samarkand joined
moritz it won't work for separate .pod files, but then the "use v6;" heuristic doesn't work for those either 14:08
also we could add references to documentation files in META.info too
jdv79 yeah, its just a matter of how certain things poke at files in isolation
samarkand Hi, how can I do this?
$my_template_str = "My string $var1 something";
# some stuff.....
# and then applying some variable to $my_template_str
$var11 = "aaabbb";
$my_template_str.apply($var1 => $var11);
# => "My string aaabbb something";
sorry for the bad formatting
14:08 brrt left
jdv79 trying to avoid large changes 14:08
moritz samarkand: there's a very simple way: use a subroutine 14:09
hoelzro FROGGS: interestingly enough, a newline between { and method fixes the problem
samarkand @mortiz, I mean, string interpolation or something
moritz samarkand: my $template = sub (:$var1) { " string $var1 something" }
ShimmerFairy moritz: as I understand, it's not called META.info. Or did I miss something?
moritz samarkand: and then my $result = $template( var1 => 'abc')
jdv79 hoelzro: where is that in S02 again? i'm blind. 14:10
moritz ShimmerFairy: could also be META.json or so
samarkand thanks, is that the only way?
hoelzro jdv79: I have misquoted, looking
ShimmerFairy moritz: I wish we would settle on one name (I thought we had with META6.json). It's kinda annoying, honestly :/
14:10 diana_olhovik joined
jnthn m: my $template = { "My string $:var1 something" }; say $template(var1 => 'eats') # another way 14:10
yoleaux 14:00Z <brrt> jnthn: using nqp i can make a reliable inlined-notinlined pair using a getlex to another frame
camelia rakudo-moar 0132b6: OUTPUT«My string eats something␤»
yoleaux 14:01Z <brrt> jnthn: i can't repeat the issue outside of the compiler yet, and can't see anything obviously wrong with it
samarkand jnthn, mortiz - that's it, thx
jnthn samarkand: That ways (wht $:foo placeholder params) is perhaps the least repetitive way I can think of :) 14:11
moritz ShimmerFairy: aye; though the list in ecosystem points to the file directly, so it's only half bad
14:11 FROGGS_ joined, diana_olhovik_ left
ShimmerFairy moritz: for as long as we use the ecosystem over CPAN, at least :P 14:11
hoelzro jdv79: design.perl6.org/S01.html#line_104
jnthn .tell brrt Ugh...I think we may really have to catch the mis-lookup under the debugger to actually figure this one out, or just dump a crapload of debugging output 14:12
yoleaux jnthn: I'll pass your message to brrt.
moritz ShimmerFairy: well, there's an easy fix: submit pull requests to all modules to change the name
jdv79 cpan uses META6.json as specced
pink_mist jnthn++ that template example was great! awesomely powerful feature of perl6 =)
jdv79 hoelzro: thanks 14:13
hoelzro sure thing!
moritz ah yes, the good old $:foo trick 14:14
Math::Model uses that for good measure
14:15 FROGGS left 14:18 diana_olhovik left 14:19 samarkand left 14:21 zakharyas joined 14:25 zakharyas left
jdv79 still stuff in the ecosystem that doesn't compile 14:26
14:32 diana_olhovik joined
jdv79 because of unit i mean 14:33
a little surprising
[Coke] ShimmerFairy: it's a bug in the spec, not a bug in rakudo. 14:34
yoleaux 13:10Z <ShimmerFairy> [Coke]: I believe it's a mistake for you to reject RT #61602 -- it may not have a solution in the foreseeable future, but rejection should apply to tickets we _know_ are invalid, not tickets about the still-decided. 'stalled' is more appropriate for it, like it was
[Coke] once the spec issues are resolved, if the answer is "yes, we need that", then sure, we reopen the ticket.
14:37 diana_olhovik left
moritz anybody want to discuss the security implications of NFG? 14:37
friends of mine organize a capture-the-flag contest, and I foolishly promised to write an exploitable service for them 14:38
in Perl 6, no less :-)
so, I thought about having a secret flag in a file that's in a decomposed form, and blacklist that name, but make is accessible via the composed form, or something 14:39
or maybe that the character length of the file name is limited, and in NFG it's shorter than the limit because it's more compactly composed, or something 14:40
anybody have any good ideas on an NFG-exploitable service that looks plausibly secure to somebody not familiar with NFG?
14:43 skids joined 14:46 diana_olhovik joined, yqt left 14:49 samarkand joined, felher left
samarkand I wonder, what's the default or most popular naming convention? 14:49
Is it
a) my-metho
b) my_method
It seems it's "a", doesn't it? But I do keep seeing "b"
14:51 khw joined
jdv79 there's seems some sort of bent for kebabing 14:52
i don't get it - maybe because it can be done 14:53
14:53 diana_olhovik_ joined 14:54 diana_olhovik left
samarkand I wonder, what's the default or most popular naming convention? 14:54
Is it a) 'my-method' or b) 'my_method'
It seems it's "a", doesn't it? But I do keep seeing "b" as well.
jdv79 a
aka kebab casing
samarkand ok
14:55 samarkand left
jdv79 design.perl6.org/S99.html#kebab_case 14:55
awwaiid yay new rakudo star! 14:57
14:57 diana_olhovik joined, diana_olhovik_ left
awwaiid I offered to show rakudo off to some people here at StrangeLoop (all of you should be here, btw), but haven't had takers. I might start tackling people. I REALLY wish I had hit the deadline and had a chance of presenting here... this is definitely a prime target audience 14:58
14:59 mn080202 joined
mn080202 hey guys can I call methods on action object from lambda inside of grammar? 15:00
*or girls
so I want inline action in a grammar to call method on action object when parsing 15:01
is that possible
or do I have to have a method on the grammar to do it like <.methodIwanttocall>
moritz mn080202: iirc the actions object is in $*ACTIONS, so you can just call a method on it 15:02
mn080202 thanks moritz you are the man 15:03
moritz mn080202: you're welcome 15:04
15:04 mattprelude joined, diana_olhovik left
confuseAcat wouldn't it be cool if rakudo.org/how-to-get-rakudo/ reflected the latest release? 15:05
[Coke] RT: 1,024; GLR: 4, WEIRD: 12, LTA: 78; testneeded: 9; xmas: 98 15:06
[ptc] sometimes wonders if [Coke] is a bot...
[Coke] That's interesting. Please continue. 15:07
15:08 skids left
[ptc] gets eliza flashbacks from the 90's 15:08
15:09 diana_olhovik joined
[Coke] Why do you think that? 15:10
raydiak good morning, #perl6 15:11
15:12 mn080202 left
[Coke] regarding stress test chat from the other day - we might want to re-evaluate the stress tests and see if we can pull any into spectest. 15:12
given speed improvements, it might be possible for us to get better coverage in the common case.
15:13 felher joined 15:15 yqt joined 15:17 diana_olhovik_ joined, mn080202 joined, diana_olhovik left, ggoebel2 left 15:18 pmurias joined
pmurias hi 15:18
mn080202 @moritz the thing you showed me before where can one find doc's for it
hoelzro m: class Foo { method bar { } #= baz␤}; say Foo.WHY 15:20
camelia rakudo-moar 0132b6: OUTPUT«baz␤»
hoelzro ah ha
it's all clear now
"Declarator blocks that start with #= attach to the declarator declared at the start of the line immediately before them." 15:21
it *is* up to spec it seems, but definitely LTA
[Coke] stresstest failures in t/spec/S02-types/WHICH.t and t/spec/S05-transliteration/trans.t 15:22
nine [Coke]: there are certainly stresstess that are not that stressfull at all 15:23
15:27 _mg_ joined
pmurias hoelzro: what would be the awesome choice here? warn in such cases? 15:35
[Coke] timtoady, lizmat: t/spec/S16-io/quoting-syntax.t - this appears to be testing something that isn't in the specs any more. might be removable.
hoelzro pmurias: that's a good question =/
lately, I often find myself divided between warnings in the compiler and having a really nice linter 15:36
15:37 tokuhiro_ joined
pmurias tools support for Perl 6 is lacking at the moment :( 15:38
dalek ast: 7666314 | moritz++ | S12-attributes/mutators.t:
Clean up S12-attributes/mutators.t
 15:39
15:39 kaare__ joined 15:40 ggoebel joined 15:42 tokuhiro_ left
[Coke] t/spec/S03-operators/shortcuts.t looks dodgy. 15:43
15:48 brrt joined 15:51 zed_ joined
masak hoelzro: just to be clear, what is it that's LTA about that formulation? is it the "start of the line" bit? 15:58
(and yes, I think I agree. that goes against expectations, and isn't very considerate to one-liners)
[Coke] how often are you using docs on a one liner? 16:02
16:03 diana_olhovik joined 16:04 diana_olhovik_ left
TimToady the main thing you're declaring on a line is usually the first thing; otherwise you'll be attaching stuff to the last item in the signature, if it's on the same line 16:05
16:06 pmurias left 16:08 pmurias joined 16:14 _mg_ left 16:17 ShimmerFairy left
masak I agree with TimToady, but if I saw such a doc-comment right after a method, I would probably assume it was for the method, not the class. if it was for the class, I'd expect it to occur after the class. 16:17
*mumble mumble* Least Surprise
16:18 rurban left
TimToady maybe #= can't see inside parens? 16:18
zostay m: my %x = foo => 1, bar => 2; baz => 3, qux => 4; dd %x; # DANGER DANGER WILL ROBINSON 16:22
camelia rakudo-moar 0132b6: OUTPUT«Hash %x = {:bar(2), :foo(1)}␤»
16:22 spider-mario left
TimToady that oughta warn about Useless use of...something... 16:23
16:24 spider-mario joined 16:26 Begi joined
Begi I'm using the simple code of doc.perl6.org/type/IO::Socket::INET. I've this error : Failed to connect : connection refused. I don't think this is my firewall. An ideas ? Thanks 16:29
16:30 ShimmerFairy joined
dalek p: 0198475 | (Pawel Murias)++ | src/vm/js/ (9 files):
[js] Store the contents of hashes in a separate inner object. Support elems on hashes.
 16:30
p: 7914bf6 | (Pawel Murias)++ | t/nqp/18-associative.t:
Test calling nqp::elems on hashes.
nine Ok, I successfully created an EnumHOW based Boolean class in BOOTSTRAP that has a gist method. Now if I try to give it a proper multi method gist, things get difficult. If I just do that, the build complains about the missing proto. But I've not found a way to create that proto 16:34
[Coke] has trouble running the profile html in chrome. keeps asking me if I want to kill it. 16:37
brrt know i know why there are so many germans in #perl and #perl6. in german, there is also more than one way to do it 16:39
yoleaux 14:12Z <jnthn> brrt: Ugh...I think we may really have to catch the mis-lookup under the debugger to actually figure this one out, or just dump a crapload of debugging output
nine brrt: so you say that Perl is a good fit for a German native brain? 16:40
or Germanic probably
brrt seems like a decent hypothesis 16:41
in translating, i was thinking to myself 'but why didn't they use this other construct for consistency'
dalek osystem: 53f978b | timotimo++ | META.list:
i've renamed a bunch of META.info to META6.json
 16:44
osystem: d54bd92 | timotimo++ | META.list:
another rename of META.info to META6.json
flussence facepalms (no wonder they mysteriously stopped working the first time I tried that in the code repo...) 16:45
dalek k-simple: 1aaee2b | timotimo++ | META (2 files):
rename META.info -> META6.json
 16:47
p: b9d9198 | (Pawel Murias)++ | src/vm/js/nqp-runtime/serialization.js:
[js] serialize hashes
 16:48
[Coke] fired up the profile in firefox, getting a lot of inclusive times of Inf & NaN. :( 16:49
16:50 Begi left
timotimo [Coke]: yes, i've seen things like that, too ... and i don't really have an idea why that happens ;( 16:50
dalek p: dc5da29 | (Pawel Murias)++ | src/vm/js/QAST/Compiler.nqp:
[js] Implement nqp::bindpos_s.
 16:51
16:51 travis-ci joined
travis-ci NQP build failed. Pawel Murias '[js] serialize hashes' 16:51
travis-ci.org/perl6/nqp/builds/82316551 github.com/perl6/nqp/compare/7914b...d9198dd36e
16:51 travis-ci left
timotimo 362 META.info, 12 META6.json 16:51
in the ecosystem META.list
flussence [Coke]: might be something to do with 32-bit number cutoffs in JS and microseconds? That would cause problems for values around 35 or 70 minutes depending on if it's signed or not 16:54
16:54 travis-ci joined
travis-ci NQP build failed. Pawel Murias '[js] Implement nqp::bindpos_s.' 16:54
travis-ci.org/perl6/nqp/builds/82316931 github.com/perl6/nqp/compare/b9d91...5da29a855b
16:54 travis-ci left 16:56 colomon left
flussence (that doesn't seem right though, I'm pretty sure JS silently promotes ints to floats on overflow...) 16:56
16:57 brrt left
AlexDaniel how can I "break" multiple loops? 16:58
flussence m: outer: loop { loop { say 1; last outer; } }; say 2 16:59
camelia rakudo-moar 0132b6: OUTPUT«1␤2␤»
AlexDaniel ok, what about "break 2;" ?
TimToady why do you want to break things? 17:01
AlexDaniel I mean, it is a frequent problem when you want to break out of multiple loops, and creating a label (or whatever it is called) is too difficult to type!
I wish I could just say "break 2" to get out from two loops, that would be great
TimToady no it wouldn't, what if you miscount? 17:02
AlexDaniel then I'd expect some awesome error message
TimToady how would it detect that you miscounted?
flussence
.oO( I usually use «return» for that, keeps my subs from growing to insane complexity )
 17:03
AlexDaniel I don't know? "torture the implementor on behalf of the user"…
TimToady and should it count hidden loops in closures and such?
does it count the implicit loop in grep and map?
labels are really much saner 17:04
AlexDaniel maybe! That being said, I think that I can count up to two… Sometimes I can count up to 3, when I really need that. But that happens tops 1 time in a year… 17:06
And if it was implemented, I'd expect it not to count anything besides explicit loops that I've typed 17:07
I don't think that I've ever wanted to jump out of anything like that 17:08
but jumping out of two of my own explicit loops is pretty frequent
m: outer: loop { loop { say 1; last owter; } } say 2; 17:13
camelia rakudo-moar 0132b6: OUTPUT«5===SORRY!5=== Error while compiling /tmp/rWqzRE7lHW␤Strange text after block (missing semicolon or comma?)␤at /tmp/rWqzRE7lHW:1␤------> 3er: loop { loop { say 1; last owter; } }7⏏5 say 2;␤»
AlexDaniel m: outer: loop { loop { say 1; last owter; } }; say 2;
camelia rakudo-moar 0132b6: OUTPUT«5===SORRY!5=== Error while compiling /tmp/UQoYbSSvXa␤Undeclared routine:␤ owter used at line 1␤␤»
AlexDaniel shouldn't it say “undeclared label”?
masak arguably, yes. 17:14
AlexDaniel m: my $test = 'outer'; outer: loop { loop { say 1; last $test; } } say 2; 17:15
camelia rakudo-moar 0132b6: OUTPUT«5===SORRY!5=== Error while compiling /tmp/_fbuLShMhq␤Strange text after block (missing semicolon or comma?)␤at /tmp/_fbuLShMhq:1␤------> 3er: loop { loop { say 1; last $test; } }7⏏5 say 2;␤»
masak also, Levenshtein.
AlexDaniel m: my $test = 'outer'; outer: loop { loop { say 1; last $test; } }; say 2;
camelia rakudo-moar 0132b6: OUTPUT«1␤Cannot call last(Str); none of these signatures match:␤ ()␤ (Label:D \x)␤ in block <unit> at /tmp/lKSxzLUfcw:1␤␤»
masak I'm glad that one doesn't work :) 17:17
17:17 yqt left 17:19 rindolf left
AlexDaniel m: say Label.Int 17:20
camelia rakudo-moar 0132b6: OUTPUT«46809168␤»
AlexDaniel m: say Label.Int
camelia rakudo-moar 0132b6: OUTPUT«68591696␤»
AlexDaniel m: Label.last 17:22
camelia rakudo-moar 0132b6: OUTPUT«control exception without handler␤ in block <unit> at /tmp/7X8ICvNxsN:1␤␤»
17:24 _mg_ joined 17:25 brrt joined
timotimo arguably, those shouldn't work on undefined labels, eh? 17:26
m: say Label.defined
camelia rakudo-moar 0132b6: OUTPUT«False␤»
17:27 rindolf joined
AlexDaniel timotimo: I've no idea. This is just so weird… 17:27
timotimo: I will not submit a report just because I don't get it… :) 17:28
timotimo what exactly is weird to you about this? 17:29
AlexDaniel I mean, Label.Int – what is this and why would anybody want to do that? 17:30
timotimo you'd not really likely want to do that
.Int returns the nqp::where of the label
AlexDaniel should that even be accessible for regular users? 17:32
timotimo well, that's kind of how you differentiate them, i guess?
hm 17:33
well, there's .gist which shows more info
AlexDaniel m: outer: loop { loop { say 1; last last last last last; } }; say 2; 17:34
haha
camelia rakudo-moar 0132b6: OUTPUT«(timeout)1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1␤1…»
timotimo i don't think "last last" works
AlexDaniel of course not 17:35
timotimo what does that even do? %)
the same as "last" would do?
AlexDaniel yeah
timotimo because last returns .. Any?
Nil?
AlexDaniel doesn't really matter?
timotimo right
AlexDaniel but that's an interesting question, why would anybody want to use last in non-void context? 17:36
maybe in ?? !!
17:39 rindolf left, tokuhiro_ joined 17:43 tokuhiro_ left 17:46 rindolf joined
flussence m: my $r = do loop { loop { last; } }; say $r 17:48
whoops
m: my $r = outer: do loop { loop { last outer; } }; say $r
camelia rakudo-moar 0132b6: OUTPUT«(timeout)»
rakudo-moar 0132b6: OUTPUT«5===SORRY!5=== Error while compiling /tmp/_THUJPl5NB␤Confused␤at /tmp/_THUJPl5NB:1␤------> 3my $r = outer:7⏏5 do loop { loop { last outer; } }; say $␤ expecting any of:␤ colon pair␤»
flussence well that answers that...
moritz m: my $r = do { outer: loop { loop { last outer } }; say $r.perl 17:55
camelia rakudo-moar 0132b6: OUTPUT«5===SORRY!5=== Error while compiling /tmp/FF5DUgkj69␤Missing block␤at /tmp/FF5DUgkj69:1␤------> 3oop { loop { last outer } }; say $r.perl7⏏5<EOL>␤ expecting any of:␤ postfix␤ statement end␤ statement modifier…»
moritz m: my $r = do { outer: loop { loop { last outer } } }; say $r.perl
camelia rakudo-moar 0132b6: OUTPUT«Label.new␤»
[Coke] "my $x = last;" kills the REPL. 17:58
moritz it probably last()s the L in repL :-) 17:59
mr_ron moritz: You revised S12-attributes/mutators.t. Should I close github.com/perl6/roast/issues/71? 18:00
moritz mr_ron: yes please 18:01
18:02 FROGGS_ is now known as FROGGS
mr_ron done 18:02
FROGGS .tell jnthn I try to implement 'my $a; "a" ~~ /$a=[.]/' and fail to do that in P6::A... is that something I should be able to do in the actions? 18:03
yoleaux FROGGS: I'll pass your message to jnthn.
mr_ron Just a last grumble. FETCH for the proxy is sort of an accessor method for an attribute. Normal accessor methods for attributes only get called once to get the value why should we allow the FETCH method to be called multiple times to get the value once? 18:05
brrt .tell jnthn (poor jnthn for all the spam) i've created a gist with dynvar log results: gist.github.com/bdw/992dfd5b69b7d42f2239 18:11
yoleaux brrt: I'll pass your message to jnthn.
brrt i'm pretty sure he's going to see the same interesting thing :-)
moritz mr_ron: the compiler needs the value for type-checking, for example 18:13
mr_ron: it's not about return values; it's being allowed to read values; and proxies are special in that reading a value can have a side effect 18:14
mr_ron ... checking whether the same could be said for ordinary custom object attr accessors ... 18:16
m: class C{has $.a = 0; method a{ return 42 + ++$!a }}; my $o = C.new; say $o.a; say $o.a; 18:19
camelia rakudo-moar 0132b6: OUTPUT«43␤44␤»
mr_ron m: class C{has Int $.a = 0; method a returns Int { return 42 + ++$!a }}; my $o = C.new; say $o.a; say $o.a; 18:21
camelia rakudo-moar 0132b6: OUTPUT«43␤44␤»
18:26 brrt left 18:30 domidumont left, larion left 18:32 yqt joined
mr_ron It still doesn't make sense to me so maybe I'll do more research on Proxies and take up the question in that context ... 18:32
18:46 aristotle left
timotimo For more information, see the perl6(1) man page. 18:46
do we actually have that? 18:47
raydiak mr_ron: I've looked at it like this: a method *returns* a value (in normal use), but a Proxy *is* a value (albeit a special one). If you take an ordinary value and pass it around, ordinary stuff happens. If you take a Proxy and pass it around, it does special stuff for every read or write (even internal to rakudo), because that's what it's designed to do. 18:49
18:50 leont joined
raydiak mr_ron: iow the difference is that an accessor runs the code once and returns a value that you use, while a Proxy is re-called for every access and (possibly) returns a new value every time 18:54
mr_ron just got back ... 18:55
Sort of the heart of the problem I see is that if you want an object attribute mutator you need a Proxy which gives you a FETCH accessor which may now be called more than once to get the value of the attribute once. A mutator by proxy seems to mean a potentially difficult or broken attr accessor. 19:00
timotimo huh why?
mr_ron why what? 19:01
psch Seq in a Proxy is something that immediately comes to mind as "weird"
or "from a Proxy" 19:02
timotimo why would that mean "a potentially difficult or broken attr accessor"?
psch although i'd question the design decision behind putting a Seq into an Attribute
AlexDaniel timotimo: I don't think that there is a man page, but there is going to be one eventually, I guess 19:03
timotimo: where can you see that message?
timotimo perl6 --help
AlexDaniel timotimo: I mean, is it going to be displayed if you're using for example windows?
timotimo i was looking for the name of the module debug env var
AlexDaniel I'm not sure that there is “man” for Windows…
psch rakudo-specific env vars are terribly underdocumented :/ 19:04
RAKUDO_ERROR_COLOR is the worst offender, in my opinion
'cause it checks for "%*ENV<RAKUDO_ERROR_COLOR> // !$is-win;"
which is assigned to $color 19:05
which in turn is treated as "do we want color"
so it needs RAKUDO_ERROR_COLOR to exist but be False...
AlexDaniel “--optimize=[level] use the given level of optimization (0..3)” – which value is default?
FROGGS AlexDaniel: 3 19:06
AlexDaniel nice
FROGGS and 'off' is also a valid level
timotimo hold on
no, 2 is normal, 3 is used for the setting
FROGGS ahh, interesting
mr_ron Accessors can have side effects, even ordinary accessor methods. Ordinary object attr accessor methods only run once to get their results. The FETCH version may run more than once.
AlexDaniel hm
psch mr_ron: auto-generated accessors can have side-effects? 19:07
mr_ron not auto-generated but custom ones yes
psch oh, yeah custom ones are just methods after all 19:08
mr_ron m: class C{has Int $.a = 0; method a returns Int { return 42 + ++$!a }}; my $o = C.new; say $o.a; say $o.a;
camelia rakudo-moar 0132b6: OUTPUT«43␤44␤»
psch i'm not sure i'd still call that an accessor, fwiw 19:09
but maybe that's Java-damage, 'cause accessor is pretty much getter for me, which should just return the value
...which is of course an implicit contract, but seems useful enough to me to assume
mr_ron An accessor could fetch a web page and have a side effect on the remote web server for example. 19:10
I guess we don't actually have working object attr mutators yet and the question isn't urgent. 19:11
raydiak if it's not meant to re-fetch the page every time, you might want to cache the result inside the FETCH, e.g. with a state var 19:12
raydiak thinks of side-effects in FETCH as falling under the "enough rope to hang yourself" category :)
psch raydiak: i pretty much agree
an accessor that needs to fetch a remote value isn't an accessor to me
if the object needs the remote value, i'd fetch it in &new or &BUILD 19:13
that's of course preference :)
mr_ron An example that just popped into my head - fetching from a sequence generator in a database ... 19:16
dalek rl6-roast-data: a0d610d | coke++ | / (9 files):
today (automated commit)
 19:17
psch but that'd be something like "method getNextId { $db.getCurrentId + 1 }", which is not a value inherent to the object
(plus neccessary async precautions etc) 19:18
psch is probably mostly arguing that accessors *should* be side-effect free by definition
which means the language is not at fault when someone builds something that *has* side-effects and calls it an accessor 19:19
19:19 nowan left
mr_ron What if the accessor needs to access a remote resource that has a limit on the number of accesses per minute/hour etc. 19:22
yes you can code around it
It still looks to me like you coded a class with an ordinary custom attribute accessor and then when you recoded for a mutator and put the accessor code in the FETCH method the multiple accesses could cause trouble. 19:25
psch mr_ron: i'm sorry, this looks to me like "if you stuff that it's safe it might be blow up"
mr_ron: that's not saying you can't have a point there
mr_ron: i just don't see it
err 19:26
"if you code stuff that isn't safe"
psch suspicously eyes the keyboard
mr_ron In regular Perl accessor methods can have side effects. I think that's true of most languages. Give me an example of a major programming language where you can't? 19:29
psch now we're back at semantics. i'm coming from a Java-POV, where an accessor is a getter, which is a method that accessed and returns a value that's internal to the class 19:31
19:31 FROGGS left, nowan joined
psch if an accessor can have side-effects, it's not a getter 19:31
if it's not a getter, none of the things i said apply, because i am working under the assumption that it is a getter
if it is a getter, it can't have side-effects, otherwise it's not a getter
the important phrase here is "internal to the class", i suppose 19:32
19:32 apotheon left
psch if you have a method that returns a value that's not internal to the class *i* *personally* wouldn't call that an accessor 19:32
19:33 apotheon joined, kaare__ left
psch wikipedia calls an accessor method a "[method] which returns the value of the private member variable" 19:33
to clarify: it's not about "you can't write a method that returns and changes a value" it's about "you can't call a method that changes an accessor" 19:35
at least that's where i see the communicative dissonance 19:36
19:37 cognominal joined
raydiak has $!id; method id { Proxy.new: FETCH => { $!id //= $db.getNextID }, STORE => -> $, $v { $!id = $v } } looks pretty usable to me 19:38
19:38 mn080202 left 19:39 pmurias left
psch raydiak: again, agreed. i'd struggle to call it an accessor though 19:39
mr_ron Now it looks like you can only get one sequence number from the database for the object ever, which doesn't look right ...
my $first = $obj.id; my $second = $obj.id; 19:40
looks like they get the same value which doesn't look right 19:41
raydiak ah I see, thought you meant an id for the object with the id method...then perhaps this is what you want: method id { my $id; Proxy.new: FETCH => { $id //= $db.getNextID }, STORE => -> $, $v { $id = $v } }
19:41 pmurias joined
raydiak every call to .id will return a new proxy with it's own $id 19:41
now, with that said, I do see what you're saying, and it would be nice to have a way to write it without the Proxy/caching complication...have often wished the rval could be part of the signature, myself 19:42
mr_ron I think you may be right about the sequence example but I have to leave for an appointment now. BBL. 19:43
raydiak np, good luck
lizmat messages ? 19:45
yoleaux 12:25Z <FROGGS> lizmat: I commented: gist.github.com/lizmat/f3807956c35...nt-1582671
raydiak is it too late to add rval(s) to the signature, separated from the normal params list by a '=', complete with multidispatch? :)
psch ...i'm not sure i can imagine what that's supposed to look like or do 19:47
20:04 telex left 20:06 telex joined
moritz raydiak: uhm, you do know that = introduces a default value? 20:08
20:09 pmurias left 20:13 xfix left
masak I have jotted down a few notes about macros and the direction I think we should take them in Perl 6. the gist is called "three types of macros": gist.github.com/masak/13210c51f034f931af0c 20:14
comments are greatly appreciated and warmly welcomed, etc. etc. 20:15
grondilu m: say (1/3).base-repeating 20:16
camelia rakudo-moar 0132b6: OUTPUT«(0. 3)␤»
masak I still don't have all the answers, but things are becoming much clearer since YAPC::EU.
lizmat .tell FROGGS I commented on your comment on gist.github.com/lizmat/f3807956c354c14902a3
yoleaux lizmat: I'll pass your message to FROGGS.
masak m: say (1/7).base-repeating 20:17
camelia rakudo-moar 0132b6: OUTPUT«(0. 142857)␤»
psch masak: the ordering for "ex1 compute()" seems confusing?
masak: that is, the commented/to-be-expected-output
masak psch: arguments are evaluated before they are sent to ex1
psch masak: yes, but &compute doesn't &say 20:18
masak hm, yes.
fixing.
fixed, hopefully: gist.github.com/masak/13210c51f034f931af0c 20:19
psch++
psch m: my $x = -> { say "foo" }; $x;
camelia rakudo-moar 0132b6: OUTPUT«WARNINGS:␤Useless use of variable $x in sink context (line 1)␤»
psch m: my $x = sub { say "foo" }; $x;
camelia rakudo-moar 0132b6: OUTPUT«WARNINGS:␤Useless use of variable $x in sink context (line 1)␤»
psch masak: ^^^ that still kind of throws a wrench in there, i think? 20:20
unless macros in SINK always .CALL-ME() (or something like that...)
masak well, let's work this out.
in the ex1 case, compute() is called in the mainline, so we're fine there. 20:21
psch ah, right
masak in the ex2 case, `compute()` is *sent* in that form as a Qtree, and then evaluated in the quasi.
psch and ex2 also calls explicitely
yeah
masak so we're fine there :)
psch i didn't pay enough attention to the calls to ex{1,2} there :)
masak *nod*
talking about macros is confusing by default :) 20:22
psch oh, but
sub ex1($value) { ... $arg; ... }
masak heh
fixing
psch :)
20:24 darutoko left
masak fix'd: gist.github.com/masak/13210c51f034f931af0c 20:24
psch the "Unhygiene" section is wonderful, masak++ 20:25
masak :D
20:26 rurban joined
timotimo likes it, too 20:34
masak maybe I should have a motivating example of unhygiene, too.
grondilu m: printf "%s(%s)", |(355/113).base-repeating
camelia rakudo-moar 0132b6: OUTPUT«5===SORRY!5=== Error while compiling /tmp/LGMytANGEQ␤Variable '%s' is not declared␤at /tmp/LGMytANGEQ:1␤------> 3printf "%s(7⏏5%s)", |(355/113).base-repeating␤»
masak but that's a little bit of uncharted territory, still.
maybe we can find a good example from some other macro-blessed language.
grondilu m: printf '%s(%s)', |(355/113).base-repeating
camelia rakudo-moar 0132b6: OUTPUT«3.(1415929203539823008849557522123893805309734513274336283185840707964601769911504424778761061946902654867256637168)»
timotimo can we do better than to create a Scalar container to hold the capture when we have a method foobar(|args) ? 20:35
masak Common LISP rather than Scheme, I guess...
20:35 tokuhiro_ joined
grondilu m: printf '%s(%s)', |(1/30).base-repeating 20:35
camelia rakudo-moar 0132b6: OUTPUT«0.0(3)»
20:36 vendethiel joined
psch masak: the last two paragraphs make me want to grasp for "use" and "no" for visitor macros 20:36
masak: i'm pretty sure that's a not-so-great idea, but that's what the writing seems to lead to
20:37 mr_ron left 20:39 tokuhiro_ left 20:40 _mg_ left
psch i like the hint of s-expr that comes through in the whole gist 20:40
especially the "Visitor macros" section
as in, with those we'd have predeclarable ways of complete code transformation 20:41
which starts at linting/coding-standards and doesn't really stop
that's to say, we'd gain a mechanism that allows code transformation on a level pretty close to what LISPs allow 20:46
although i suppose in the end that hangs on the exact Qtree implementation 20:47
(i realize that was pretty much the idea behind it, but i've not really paid close attention before, shelving it as something that's probably beyond me...)
masak yes, Qtrees are the "hidden player" which would make much of this work. 20:48
also absent from that gist (but which I know a fair bit more about nowadays) is the distinction between "analytic Qtrees" and "synthetic Qtrees". 20:49
that is, the parser can create a Qtree, or you can create a Qtree, and they won't end up being exactly the same. 20:50
mainly because the parser has the benefit of parsing thing in a context, and "annotation" its Qtree with that context. (surrounding declarations, etc)
but the Qtrees built node-by-node by the user are "free-floating" until they're actually attached/splied into the program. only then can they get such a context. 20:51
specifically, a Qtree that you built with something like `$x = 5;` is never illegal, but it might cause a compilation error when you add it to a part of the program which doesn't have `$x` declared. 20:52
vendethiel psch: I disagree it's a bad idea 20:54
masak vendethiel: did you mean "I disagree, it's a bad idea" or "I disagree that it's a bad idea"? :) 20:55
also, what idea?
vendethiel masak: second. the idea being "enable or disable visitor macros using 'use' and 'no'"
masak actually, that's how I envisioned they'd be pulled in normally. 20:56
20:56 diana_olhovik left
psch i dislike the conflation, mostly 20:56
masak but I'm thinking there might also be a lower-level API for that, something where you register visitors on a given parser.
psch "use/no" for pragmas, modules, vmacros
seems a bit much to me 20:57
masak they all share a certain goal -- pulling something in from the outside.
I'd much rather have "use/no" than "use-module"/"no-module", "use-pragma"/"no-pragma", "use-vmacro"/"no-vmacro"
psch i suppose arguing from the alternatives i can't disagree 20:58
as in, having distinct keywords for each would be horrible, yes 20:59
masak .oO( strawman accomplished! ) :P
psch :P
21:00 mr_ron joined
psch i mean, there's no reason why vmacros are the one thing that "do something similar but are imported differently" 21:00
compared to pragms and moduels (and slangs)
masak I see vmacros as pretty similar to routines, actually. they don't get called by the user code, but they *do* get called by the parser that reads the user code. 21:01
lizmat .tell ugexe I hopefull answered your questions in gist.github.com/lizmat/f3807956c354c14902a3
yoleaux lizmat: I'll pass your message to ugexe.
masak for this reason, they don't really need a name and a place in a lexpad or symbol table. but maybe having a name there helps sometimes.
psch (fwiw, vmacros and slangs are becoming pretty much TWTDI in my head by now)
lizmat afk again& 21:02
psch that's not *the* but *two* :)
masak TWTDI? "The Worst Thursday Double Integral"?
psch two ways to do it :P 21:03
masak to me it's more like the difference between BUILD and new
you can use BUILD as long as you're fine with just sending positionals to the default .new method -- that is, you're not changing the default interface of the object
similarly, you can keep using macros (in all their flavors) if you're staying reasonably within limits of what the language already does 21:04
ramping up, you might reach for `is parsed` or syntax macros or both
but at some point that grows unwieldy, kinda like a bunch of scattered regexes that want to be a grammar 21:05
and then you'd consolidate things into a slang (API pending)
psch the regex :: grammar analogy makes sense to me... :) 21:07
masak sorry for the wonky analogy :) but both `method new` and slangs are for when the things already provided by default are not enough and you need to change things *more*
psch yeah, that's how i'd order it too
masak `method new` helps override the default signature of the constructor. (and I meant "nameds" above, not "positionals") 21:08
psch macros can fix a few hickups here and there, while slangs can turn everything into COBOL
just like &new can let you BUILD your god object from a single hash
masak slangs help put you in a new (sub-)language, either a completely different one or one that's "mostly Perl 6 main" or (like regexes) one where Perl 6 main isn't all that far away.
psch: right, exactly. 21:09
it's different steps up the power ladder.
in 9 out of 10 classes, I'm content with defining BUILD.
awwaiid which will let me make it so I can put whitespace around method invocation and move the --> ReturnType outside of the parens? 21:10
(which I think are my top-two warts)
(I should make a personal wart list) 21:11
masak yes, please make a personal wart list.
21:11 rurban left
mst I'm assuming the perl5 rule of 'if you defined your own new() method you've already gone too far' tends to apply 21:11
masak awwaiid: both of those sound more slang-y to me. since they change syntax.
awwaiid masak++ # encouraging things
I am also ok if they are fixed in the core language :) 21:12
masak mst: that's a personal rule of mine, yes, which I hadn't really verbalized. but yes.
mst awwaiid: $foo.bar\n .baz is my big objection
awwaiid yes, exactly that
masak mst: as time goes by, I'm less and less tempted to override .new
mst awwaiid: I like lining up method calls vertically
masak good news, awwaiid and mst: there's already a slang for this! :D
it's called Slang::Tuxic
awwaiid I had a really nic elooking chain going on and it was so beautiful but didn't work
mst masak: no, it isn't
awwaiid orly
mst awwaiid: if you stick a \ on the end of each line ala Tcl it works for the moment
masak github.com/FROGGS/p6-Slang-Tuxic/ 21:13
mst Slang::Tuxic does a bunch of whitespace related stuff, most of which is weird
21:13 rurban joined, rurban left
masak heh, no arguments there :) 21:13
awwaiid ah yeah. really it is specifically the one mst mentioned that has hit me a few times
I ended up using several .= because it was prettier
mst masak: yeah, so please don't say "there's a slang for this"
masak mst: not really sure I understand your objection, but OK. 21:14
mst: Slang::Tuxic does more than you want?
mst it introduces a bunch of things I actively don't want
like, say, every single other change it makes
awwaiid yeah. I am totally fine with no-space-allowed between method and param paren
mst "there is a slang from which what you want could totally be extracted", OTOH, I will absolutely stipulate to 21:15
21:15 rindolf left
mst it's just a question of which of the bajillion people annoyed by this gets to it first 21:15
awwaiid reads source
This is definitely one of those which I would place money on will be Really Annoying for Every New Beginner Ever
masak I see three things Slang::Tuxic does (though I may have missed things)
(a) allow whitespace in "$obj .meth" 21:16
(b) allow whitespace in "$thing {'lookup'}"
(c) allow lack of whitespace in "sub(){}", specifically the lack after "sub"
mst pretty sure it allows 'thing (<args>);' 21:17
masak yeah -- all those things can go under (b), I guess. 21:18
awwaiid forks. starts chopping things
masak basically, postcircumfixes can be preceded by whitespace.
mst right, but that I actively don't won't
awwaiid++
masak awwaiid++
awwaiid: looking at an earlier version/commit might help, too.
awwaiid oh good idea
21:24 BenGoldberg joined 21:51 pmurias joined
vendethiel awwaiid: nice, I'd love to be able to go and nelwine 'em as well :D 21:57
22:10 cognominal left 22:14 RabidGravy left, apotheon left
timotimo oh 22:14
should we set arguments like |args "is raw" by default (or just generate the same code either way?) 22:15
because i don't really see a reason why |args should be stored inside a scalar container
vendethiel m: sub foo { $[1, 2, 3] }; sub bar { [1, 2, 3] }; say foo.perl, bar.perl # tell me they're different :o 22:16
camelia rakudo-moar 0132b6: OUTPUT«[1, 2, 3][1, 2, 3]␤»
timotimo m: sub foo { return $[1, 2, 3] }; say foo.perl 22:17
camelia rakudo-moar 0132b6: OUTPUT«[1, 2, 3]␤»
timotimo m: sub foo { return [1, 2, 3] }; say foo.perl
camelia rakudo-moar 0132b6: OUTPUT«[1, 2, 3]␤»
timotimo m: sub foo { return 1, 2, 3 }; say foo.perl
camelia rakudo-moar 0132b6: OUTPUT«(1, 2, 3)␤»
timotimo hum.
22:18 apotheon joined
masak 'night, #perl6 22:21
pink_mist good night masak
22:22 pmurias left 22:23 TEttinger joined 22:24 apotheon left 22:28 apotheon joined 22:33 yqt left 22:36 tokuhiro_ joined 22:41 tokuhiro_ left 22:45 cognominal joined 22:47 spider-mario left 22:51 araujo_ joined 22:53 mr_ron left 22:54 araujo left 22:56 mr_ron joined 22:57 leont left
lizmat ugexe: another comment on gist.github.com/lizmat/f3807956c354c14902a3 23:02
23:04 vendethiel left
lizmat enough CURLI ideas for now: gist.github.com/lizmat/f3807956c35.../revisions 23:20
good night, #perl6!
23:44 cognominal left 23:50 BenGoldberg 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!