»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg p6eval perl6: ... | irclog: irc.perl6.org/ | UTF-8 is our friend!
Set by sorear on 4 February 2011.
benabik Ooh, language design arguments. :-D 00:01
00:02 Chillance left 00:04 Trashlord left 00:06 soh_cah_toa left 00:07 [Coke] left 00:09 [Coke] joined 00:13 soh_cah_toa joined
colomon sorear: just e-mailed you the complete trig patch. 00:25
00:25 kfo joined
colomon still needs some work on real inverse hyperbolic functions, current implementation is hacky. 00:25
now to patch roast... 00:26
00:29 kfo_ left
colomon b: say True ?? "hello" !! "Goodbye" 00:42
p6eval b 1b7dd1: OUTPUT«hello␤»
sorear rakudo: say tan pi/2 00:45
p6eval rakudo aed9fe: OUTPUT«557135115.020977␤»
00:46 PacoLinux joined
dalek ecza: 08a7d84 | (Solomon Foster)++ | / (6 files):
Significant expansion of the C# Complex class in Utils.cs. Complete implementation of an auto-generated set of trig functions.
 00:47
colomon sorear: does niecza not allow todo fudging? 00:48
sorear colomon: correct 00:49
I don't actually quite know how it should work 00:50
I'll research that later maybe?
colomon sure
and I'll change them to skips for now
dalek ast: 46de8ad | (Solomon Foster)++ | S32-trig/generate-tests.pl:
Update trig generator with fudging for Niecza.
 00:54
ast: 4661e6b | (Solomon Foster)++ | S32-trig/ (13 files):
Update generated trig files with fudging for Niecza.
japhb Is there a way (other than string eval) to produce a role that closes over surrounding variables? 00:59
colomon oooo, there's a make spectest target now. 01:00
japhb e.g. my $foo = "blah"; my $role = role { method bar { $foo } };
01:06 REPLeffect left 01:19 REPLeffect joined
colomon sorear: whoops. forgot to implement atan2! 01:23
01:23 packetknife left 01:26 wolfman2000 joined 01:28 uasi joined
colomon sorear: you can turn on all the S32-trig tests except atan2.t 01:32
whee, sierpinski!! 01:44
01:58 envi joined 01:59 frhodes left
colomon sorear: you out there atm? 02:44
I just defined static Func<Variable,Variable,Variable> atan2_d = atan2; and that seems to compile okay, but when I add method atan2($x) { Q:CgOp { (atan2 {self, $x}) } } to Cool, niecza's compile blows sky high. What am I doing wrong? 02:47
02:48 flatwhatson left 02:50 envi left
colomon oh, should be atan2 {self} {$x} 02:51
02:56 envi joined
sorear japhb: role parameterization 03:02
colomon: I have just returned from dinner
colomon sorear: just sent you another patch 03:03
sorear colomon: (atan2 {self} {$x}). It's modelled on lisp.
{self, $x} should result in some kind of MethodNotFoundException
colomon: Why did you give atan2 a default argument? 03:04
colomon sprec
spec
sorear looks decent to me then. Which thing did you cargo-cult? 03:05
colomon all the stuff with the a2.Fetch() ; GetNumber etc
sorear sorry, I mean where did you get it from.
colomon I dropped the HandleSpecial bit because I wasn't sure how to do it with two arguments
oh, I started with the code for sin, I think 03:06
sorear add() contains an example of HandleSpecial used for two arguments
but you don't really need it here 03:07
colomon ah, yeah, I did find mul later, looking for something else. didn't occur to me to grab the HandleSpecial from there.
sorear the purpose of HandleSpecial is to implement junctional autothreading
colomon ah, right
sorear since you're only calling (atan2) from inside a function with Any args, atan2 can never see a junction
for add() and friends it's useful to avoid the cost of a function call and signature binding. not sure atan2 is the same 03:08
dalek ecza: 26ad592 | (Solomon Foster)++ | lib/ (2 files):
First stab at implementing atan2.
 03:09
colomon yeah, I have no real feel on junctions other than as a simple shortcut in comparisons. :)
sorear karma colomon
aloha colomon has karma of 152.
sorear karma Colomon Foster
aloha Colomon Foster has karma of 0.
sorear karma Solomon Foster
aloha Solomon Foster has karma of 112.
colomon I have a (not so) secret identity
sorear (also have to avoid typing 'colophon:') 03:10
colomon: My command on junctions isn't the best either. The niecza implementation of them is several orders of magnitude faster than I expected it to be 03:11
colomon that's handy!
sorear $obj == 1 | 3 | 5 is not significantly slower than $obj == 1 || $obj == 3 || $obj == 5
colomon btw, the GTK Sierpinski example Just Worked for me. That's terrific! 03:12
afk # bed 03:21
03:23 newsbad_com left 03:29 newsbad_com joined 03:36 donri left
ingy hi TimToady 03:56
TimToady: I read over % more in detail
TimToady: it's nice, but I still don't see how to allow for possible trailing comma 03:57
04:00 soh_cah_toa left
ingy sb end 04:02
moritz: funny how you tried to talk me out of a+ ** to be a +** and now its a+ ** in a05, albeit a+ % 04:05
japhb Once I have the message string, how do I create the $?USAGE variable in Rakudo, and set it to my newly-minted message? 04:15
(from within the setting, of course) 04:16
araujo hello 04:36
japhb o/ 04:40
05:05 REPLeffect left 05:13 djanatyn left, djanatyn joined 05:18 REPLeffect joined 05:20 Tene left, djanatyn left 05:21 bluescreen10 left 05:27 kaare_ joined 05:31 Tene joined, Tene left, Tene joined 05:34 flatwhatson joined 05:36 kaleem joined 05:44 flatwhatson left 05:50 kaleem left 05:52 mj41 joined 06:05 orafu left 06:06 orafu joined 06:07 risou_awy is now known as risou, mj41 left 06:10 REPLeffect left 06:12 risou is now known as risou_awy 06:13 f00li5h left
sorear hello araujo 06:17
06:19 araujo left 06:23 REPLeffect joined 06:24 Woodi joined
moritz good morning 06:41
ingy: funny indeed
Woodi morning #perl6 06:45
i got something like this:
rakudo: class A { has %!c = { 'a' => 1, 'b' => 2 } ; for %!c.keys -> $name { say $name } } 06:46
p6eval rakudo aed9fe: OUTPUT«Can only use repr_get_attr_obj on a SixModelObject␤ in <anon> at /tmp/6KlyczI8WZ:1␤ in <anon> at /tmp/6KlyczI8WZ:1␤ in <anon> at /tmp/6KlyczI8WZ:1␤␤»
Woodi and
rakudo: class A { our %c = { 'a' => 1, 'b' => 2 } ; for %c.keys -> $name { say $name } }
p6eval rakudo aed9fe: OUTPUT«Method 'STORE' not found for invocant of class 'Any'␤ in <anon> at /tmp/9HjmJPA6Ww:1␤ in <anon> at /tmp/9HjmJPA6Ww:1␤ in <anon> at /tmp/9HjmJPA6Ww:1␤␤»
Woodi hwhat should i use in nom ? works with star-07 06:47
06:48 woosley joined
moritz Woodi: the problem is that attributes are tied to specific instances, but the class body (in which you try to access the attribute) is no 06:49
Woodi our version works
hmm
moritz Woodi: so put the code into a method, create an instance of the class, and call the method with the code
Woodi you mean 'for' ? 06:50
moritz yes
you'll notice that all examples of attribute usage use them in methods, not in the class body 06:51
Woodi it is intentional outside method becouse it generate methods :) should be BEGIN probably
moritz then it can't be in an attribute
Woodi ok
so our should work ?
moritz yes 06:52
Woodi b: class A { our %c = { 'a' => 1, 'b' => 2 } ; for %c.keys -> $name { say $name } }
p6eval b 1b7dd1: OUTPUT«a␤b␤»
Woodi oo
hmm
06:52 lue left
Woodi ok, so b is ok 06:53
in nom i got STORE problems...
rakudo: class A { our %c = { 'a' => 1, 'b' => 2 } ;BEGIN { for %c.keys -> $name { say $name } } } 06:57
p6eval rakudo aed9fe: OUTPUT«===SORRY!===␤Method 'keys' not found for invocant of class 'Mu'␤»
07:00 mj41 joined 07:04 SHODAN joined
Woodi today release of next Star ? 07:04
07:04 lue joined 07:08 senobmeht left 07:09 lue left 07:10 senobmeht joined 07:13 lue joined
sorear -> sleep 07:15
tomorrow: study nom's implementation of parametric roles 07:16
snarkyboojum Woodi: haven't heard about a date for next Star yet - devs are waiting to bed down nom first I believe 07:21
TiMBuS is there a perl5 version of the cross operator? list::moreutils doesnt have one :/ 07:24
Woodi perl6: my $a = "A\n"; lc $a; $a.chomp ; say $a;
p6eval pugs, rakudo aed9fe, niecza v9-50-g26ad592: OUTPUT«A␤␤»
Woodi perl6: my $a = "A\n"; lc $a; chomp $a; say $a;
p6eval pugs, rakudo aed9fe, niecza v9-50-g26ad592: OUTPUT«A␤␤»
TiMBuS Algorithm::Loops might have one... 07:25
snarkyboojum perl6: my $a = "A\n"; say $a.chomp; 07:31
p6eval pugs, rakudo aed9fe, niecza v9-50-g26ad592: OUTPUT«A␤»
Woodi perl6: my $a = "A\n"; $a.lc; $a.chomp; say $a; 07:35
p6eval pugs, rakudo aed9fe, niecza v9-50-g26ad592: OUTPUT«A␤␤»
Woodi perl6: my $a = "A\n"; $a.chomp; say $a;
p6eval pugs, rakudo aed9fe, niecza v9-50-g26ad592: OUTPUT«A␤␤»
Woodi perl6: my $a = "A\n"; say $a.lc; 07:36
p6eval pugs, rakudo aed9fe, niecza v9-50-g26ad592: OUTPUT«a␤␤»
Woodi perl6: my $a = "A\n"; say $a.lc; say $a;
p6eval pugs, rakudo aed9fe, niecza v9-50-g26ad592: OUTPUT«a␤␤A␤␤»
Woodi perl6: my $a = "A\n"; say $a.chomp; say $a; 07:37
p6eval pugs, rakudo aed9fe, niecza v9-50-g26ad592: OUTPUT«A␤A␤␤»
Woodi it little strange but consistent :)
moritz not strange if you remember that Str and Int methods don't modify the invocant, but return a modified copy 07:38
you can do $a.=chomp if you want in-place modification 07:39
Woodi moritz: chomp at least was modyfing in-place and it was perfect 07:41
lc not
07:51 mj41 left 07:55 lue left 08:03 lue joined 08:06 mj41 joined 08:09 lue left 08:18 im2ee joined
im2ee Hello! :) 08:18
moritz good morning im2ee 08:24
perfectly inconsistent, yes
08:28 risou_awy is now known as risou 08:32 cognominal joined 08:33 lue joined 08:36 masak joined
masak good morning, #perl6! 08:36
moritz good am, ma<tab> 08:38
masak ;)
08:38 lue left
moritz you will love this one: 08:39
multi sub infix:«<=>»(Buf:D $a, Buf:D $b) { [||] $a.list Z<=> $b.list or $a.elems <=> $b.elems
}
08:41 araujo joined, lue joined
masak ooh 08:41
moritz thought so :-) 08:42
08:42 molaf joined 08:43 REPLeffect left
moritz I'm sure it's terribly inefficient in contemporary rakudo 08:43
but it's so beautiful that I'll keep it that way until somebody complains about speed :-) 08:44
szabgab_ guys, can I ask you for a quick but unrelated help. I am trying to run a poll and I need a quick feedback before I launch it 08:45
perlpolls.com/perl-news is the url, I'd appreciate your comments
08:45 MayDaniel joined, szabgab_ is now known as szabgab 08:46 lue left
masak looks ok. 08:50
from a sixer's perspective, it's more "Perl 5 news" than "Perl news", but I'm not sure the distinction matters much to the rest of the world. ;)
szabgab shall I add the planetsix? 08:51
masak wouldn't hurt.
szabgab unfortunatelly I am not sure in that :( but I added as I think it is important 08:54
masak hey, it's your poll. ;) 08:56
szabgab masak: in any case, thanks for you quick feedback!
masak I'm not sure use.perl.org belongs there. 08:57
it's literally impossible to blog there nowadays.
szabgab this is the beginnng of those polls I mentioned almost a year ago
I put it there as sort of a reality check 08:58
masak ok.
szabgab I am quite sure many people will still check it
as happened a while ago
but we'll see
masak I think a fair number of people would check "planet perl 6" if it were in that list.
szabgab it is already
masak \o/ 08:59
so it is. szabgab++
moritz: how come split gets a Str $delimiter variant, but comb doesn't get a Str $matcher variant? 09:01
09:05 mj41 left
masak re the discussion about "the final release" yesterday. I'd *love* for there to be such a moment, where we put away our tools, wipe our hands and say "there. we did it." it's just -- from what I've seen of the world, it doesn't work that way. :/ 09:06
any 6.0.0 point will by necessity be arbitrary and thus a disappointment to some. 09:07
Patterner are those people paying you? 09:08
masak so ingy claimed.
see backlog. 09:09
09:09 MayDaniel left
Patterner unless you have proof: ignore some. and continue the good work. 09:09
masak aye. 09:10
though possibly I'll get the energy to blog about this.
I was happy to see Rakudo gain speed. I was even happier about Rakudo Star. if we manage to make the jigsaw pieces fit, nom will be our awesomest product yet. 09:11
09:12 MayDaniel joined
masak it just feels so skewed to be talking about failed 6.0.0 expectations in the middle of that. 09:12
masak continues the good work 09:13
std: my $a = "HAI"; say "OH $a # comment" 09:15
p6eval std c22b2ee: OUTPUT«ok 00:01 121m␤»
masak nom: my $a = "HAI"; say "OH $a"
p6eval nom aed9fe: OUTPUT«OH HAI␤»
masak my $a = "HAI"; say "OH $a # comment"
nom: my $a = "HAI"; say "OH $a # comment"
p6eval nom aed9fe: OUTPUT«===SORRY!===␤Confused at line 1, near "say \"OH $a"␤»
masak submits rakudobug
I think we've had this one before.
b: my $a = "HAI"; say "OH $a # comment"
p6eval b 1b7dd1: OUTPUT«===SORRY!===␤Confused at line 22, near "say \"OH $a"␤»
masak oh, ok. 09:16
masak searches RT for this bug
09:17 MayDaniel left
masak aye, here: rt.perl.org/rt3/Ticket/Display.html?id=78202 09:18
some goof named masak already found that one a year ago. 09:19
Woodi masak: i do not think that Perl6 is spec only. it totally do not matter to user, especially outside Perl sphere. only implementations matter and spec is just one of design stage for them. they think that way. and now: if Perl6 spec is like ANSI specification will be + on top of the first
so ingy problem was just name... 09:20
masak Woodi: no disagreement there.
well, Perl 6 *is* a spec. maybe if he said that's what people had been paying for all along, and not the implementations too, that's why they revolted. 09:21
but note that things like Hague grants are never on just the spec, but on Rakudo primarily and the spec secondarily.
nom: my $a = "HAI"; say "OH $a \# comment" 09:22
p6eval nom aed9fe: OUTPUT«OH HAI # comment␤»
masak workaround ^^
09:31 lue joined
Woodi that is nice: Perl isn't a programming language. It's a life support system for CPAN. :) 09:34
szabgab masak: btw I'd really like to have at least one interesting news item about Rakudo every week
I'd put it in the Perl Weekly and hope to get more and more people paying attention 09:35
masak szabgab: that's a nice idea. I support it wholeheartedly. 09:37
Woodi szabgab: there is new module Simple::Redis for perl6 on github. i was angry of no working db bindings available :)
still in heavy development
masak Woodi++ 09:38
I hope to use that module soon.
Woodi pls give me know if you notice problems. and if module will help in something then pls too :) 09:39
09:39 woosley left
moritz Woodi: does my IO::Socket.recv patch from yesterday night make it more useful for you? 09:40
szabgab Woodi: it would be much easier for me if you blogged about it :) and if that bog was aggregared on planet six 09:41
but if you give me a url I can link to Github as well
Woodi szabgab: link is: github.com/slunski/perl6-simple-redis 09:43
but still not decided where to put eventual blogs :) 09:44
masak blogs.perl.org is nice.
and easy to start using.
Woodi will look
szabgab Woodi: why not write a blogging platform in Perl 6 :) 09:45
you could start by pre-generated html files and RSS feed so they are static for the reader 09:46
Woodi szabgab: agree, definitly someone should do that :)
but as for me i am terrible programmer. on the other side i can start project or two :) 09:47
just porting Net::DBus looks important :)
szabgab Oh I am a terrible programmer too
I create bugs annoying enough for others to come in and fix them :) 09:48
Woodi if it works for you then maybe it is possible to follow :) 09:49
09:59 zby_home joined 10:04 Chat3240 joined 10:05 Chat3240 left, risou is now known as risou_awy 10:06 risou_awy is now known as risou 10:12 lue left 10:16 molaf left
dalek ast: e9ba740 | moritz++ | S03-operators/buf.t:
start to test comparison operators for Buf
 10:18
kudo/nom: 6fd0773 | moritz++ | src/core/Buf.pm:
comparison operators for Buf
masak moritz: I'm thinking of spec'ing a Str $matcher form for comb. any objections? 10:19
10:20 djanatyn joined
mberends masak: I can't think os many cases where that would be used. 10:20
*of
moritz nom: say 'abc'.comb('b')
p6eval nom aed9fe: OUTPUT«No applicable candidates found to dispatch to for 'comb'. Available candidates are:␤:(Str, Mu %_)␤:(Str, Regex $pat, Any $limit, Any $match, Mu %_)␤:(Cool, Mu %_)␤:(Cool, Regex $matcher, Any $limit, Mu %_)␤␤ in method comb at src/gen/CORE.setting:1559␤ in <anon> at …
masak mberends: I had one yesterday.
moritz nom: say 'abc'.match('b')
p6eval nom aed9fe: OUTPUT«b␤»
masak mberends: $string.comb("1").elems # count the number of 1s in string 10:21
moritz nom: say "Lasst uns die Anzahl der a's zählen'.match('a', :g).elems
p6eval nom aed9fe: OUTPUT«===SORRY!===␤Confused at line 1, near "say \"Lasst"␤»
moritz nom: say "Lasst uns die Anzahl der a's zählen".match('a', :g).elems
p6eval nom aed9fe: OUTPUT«3␤»
masak that's all fine and good.
moritz masak: no objections
masak \o/
masak makes it so 10:22
Woodi nom: say "Lasst uns die Anzahl der a's zählen".match('a', :g).perl
p6eval nom aed9fe: OUTPUT«Array.new(a, a, a)␤»
moritz in nom I've just written a low-level Str.match(Str) that creates a Match object
masak Woodi: that looks like a faulty .perl output.
why aren't the letters quoted? 10:23
moritz indeed
masak submits rakudobug
also, why Array.new rather than just [] ?
moritz because [] never flattens
Woodi nom: my $a = "Lasst uns die Anzahl der a's zählen"; $a.match('a', :g) ; $a.WHAT;
p6eval nom aed9fe: ( no output )
Woodi nom: my $a = "Lasst uns die Anzahl der a's zählen"; $a.match('a', :g) ; say $a.WHAT;
p6eval nom aed9fe: OUTPUT«Str()␤» 10:24
Woodi nom: my $a = "Lasst uns die Anzahl der a's zählen"; $b = $a.match('a', :g) ; say $b.WHAT;
p6eval nom aed9fe: OUTPUT«===SORRY!===␤Symbol '$b' not predeclared in <anonymous> (/tmp/MDZILoYUJ6:1)␤»
Woodi nom: my $a = "Lasst uns die Anzahl der a's zählen"; my $b = $a.match('a', :g) ; say $b.WHAT;
p6eval nom aed9fe: OUTPUT«Array()␤»
Woodi nom: my $a = "Lasst uns die Anzahl der a's zählen"; my $b = $a.count('a', :g) ; say $b; 10:25
p6eval nom aed9fe: OUTPUT«Method 'count' not found for invocant of class 'Str'␤ in <anon> at /tmp/eMq5STVE0D:1␤ in <anon> at /tmp/eMq5STVE0D:1␤␤»
Woodi what od .match ?
s/od/do ?
grrr
ah, know
moritz nom: say 'abc'.match('a').WHAT 10:26
p6eval nom aed9fe: OUTPUT«Match()␤»
moritz nom: say 'abc'.match('b').from
Woodi nom: my $a = "Lasst uns die Anzahl der a's zählen"; $a.match(/[ab]/, :g) ; say $a.WHAT;
p6eval nom aed9fe: OUTPUT«1␤»
nom aed9fe: OUTPUT«Str()␤»
Woodi nom: my $a = "Lasst uns die Anzahl der a's zählen"; $a.match(/[ab]/, :g) ; say $a.perl;
p6eval nom aed9fe: OUTPUT«"Lasst uns die Anzahl der a's zählen"␤»
10:26 Trashlord joined
Woodi nom: my $a = "Lasst uns die Anzahl der a's zählen"; my $b = $a.match(/[ab]/, :g) ; say $b.perl; 10:26
p6eval nom aed9fe: OUTPUT«[]␤»
moritz Woodi: $a.match doesn't modify the string, so if you ignore the return value, there's no reason for matching at all
ah, you found out too :-)
char classes are written as <[ab]> in p6 10:27
moritz -> afk
Woodi nom: my $a = "Lasst uns die Anzahl der a's zählen"; $a.match(/<[ab]>/, :g) ; say $a.perl;
p6eval nom aed9fe: OUTPUT«"Lasst uns die Anzahl der a's zählen"␤»
10:27 risou is now known as risou_awy
Woodi nom: my $a = "Lasst uns die Anzahl der a's zählen"; my $b = $a.match(/<[ab]>/, :g) ; say $b.perl; 10:27
p6eval nom aed9fe: OUTPUT«[a, a, a]␤»
10:27 lue joined
Woodi nom: my $a = "Lasst uns die Anzahl der a's zählen"; my $b = $a.match(/<[az]>/, :g) ; say $b.perl; 10:28
p6eval nom aed9fe: OUTPUT«[a, z, a, a, z]␤»
dalek ecs: d7b5637 | masak++ | S32-setting-library/Str.pod:
[S32/Str] added Str form to .comb

In parallel with the Str form in .split -- it's nice to have the few times one needs it.
Also, added the :match parameter mentioned in the text.		 10:29
10:30 wamba joined
masak phenny: de en "Lasst uns die Anzahl der a's zählen"? 10:33
phenny masak: "Let us count the number of a's" (de to en, translate.google.com)
masak thought so 10:34
moritz masak: now I'd like to see tests :-)
masak moritz: I'm working on tests, though not those :)
I'm 31% through pack.t
moritz wow
masak 230 tests running. 10:35
10:35 lue left
moritz are you implementing as you go along? 10:35
masak heck, no. :P
that would be *impressive*, though.
no, I have my hands full with "what did they actually want to test here?" and "is this still relevant for us?" 10:36
just deleted a bunch of 'p' template tests.
szabgab masak: and I did not had PErl 6 in the "what articles would you like to read" section either, now fixed that too 10:38
masak szabgab: nice.
szabgab I hope to reach 1000 responses but I only have 4 days :)
I'll start promoting it very heavily in a few hours from now 10:39
10:39 pmurias joined
masak best of luck. 10:40
what about s! et al in pack/unpack. perldoc says 'forces native (short, long, int) sizes instead of fixed (16-/32-bit) sizes'. 10:54
I'm thinking that's another thing that doesn't apply when you're on Parrot or CLR or JVM.
any thoughts?
mberends masak: it's one of those 'can argue both ways' cases. Initially go minimalist, so drop 'em. 10:59
masak aye. 11:02
mberends masak: comments by moritz++ and colomon++ present a good case for a separate but related serialization/Storable spec. It would be worth collecting these kind of types to help set up that other spec, since it doesn't really exist yet apart from scattered mentions of native types in the Synopses. 11:04
11:07 molaf joined
masak aye. this is the thing that's related to the Compact structs in S09, right? 11:08
mberends reads
masak "(This is to be construed as a substitute for at least some of the current uses of C<pack>/C<unpack>.)"
whether or not it is, it feels like there's a number of ideas here that need coordinating and nurturing. 11:09
11:11 Kivutarrr joined
mberends yes. The C language bias in S09 should be softened to accommodate various platforms and implementations. It's interesting that jnthn++ is currently working hard on native types, and much of this arises in Niecza CLR Interop as well. 11:14
pmurias mberends: what are the current problems with CLR Interop? 11:17
mberends pmurias: there are issues #56 and #57 on github, sorear++ has done half a workaround for #56. 11:18
11:21 lue joined
mberends pmurias: I'm still exploring the CLR Interop territory, it's vast. I recently found (for the matchboxes puzzle) that Niecza has no pick or roll. I tried to use the System.Random class from Niecza and failed, that one has not been ticketed yet because there may some usual suspects still to try (I don't want to flood github issues with noise). 11:22
11:23 replore joined 11:25 lue left
pmurias mberends: would eval(:from<c#>,...) help with anything? 11:33
mberends pmurias: thanks, I never thought of that! I'll give it a try :)
pmurias it's not implemented yet ;) 11:34
mberends ok, tomorrow then ;)
pmurias just trying to determine if i should implement that ;)
11:35 type joined
type o/ 11:35
hello masak
mberends \o type: strong or weak? 11:36
tadzik duck!
tadzik ducks
11:36 Psyche^ joined
type not strong but strange 11:36
nom: > subset Even of Int where { $_ % 2 == 0 } Even() > my Even @a > @a=1..5 1 2 3 4 5 > my Int $x =1.3 Type check failed in assignment to '$x'; expected 'Int' but got 'Rat' > my Int @b > @b=(1,2.2,3) 1 2.2 3
p6eval nom 6fd077: OUTPUT«===SORRY!===␤Confused at line 1, near "> subset E"␤»
tadzik type: 1.3 is not an Int, hence the fail 11:37
nom: say so 1.3 ~~ Int
p6eval nom 6fd077: OUTPUT«Bool::False␤»
tadzik it works for @b, because typed arrays are not yet implemented
as in, the types are not checked
type thanks !! 11:38
tadzik you're welcome, glad I could help 11:39
11:39 Patterner left, Psyche^ is now known as Patterner, wooden joined, wooden left, wooden joined
masak type: hi there! 11:40
mberends: I must've missed any C language bias in S09. to me it feels like the decisions about the actual memory layout belong in a repr somewhere. 11:41
type hi masak: tadzik gave me the answer. thanks!! 11:42
11:45 woosley joined, bbkr left
pmurias mberends: would eval(:lang<C#>,...) be usefull to you? 11:46
11:46 bbkr joined 11:47 type left
mberends pmurias: It would be far from the first thing I would think of using. Without knowing the spec in enough details, I would also suspect that such code would be very unportable. It reminds of a Q:PIR() syntax that Rakudo used to have and happily got rid of. So the short answer is, probably not, except in desperate times. 11:49
(I've satisfied my need for random numbers with the seriously ugly now.to-posix[0].Str.comb.[*-1] %% 2) 11:50
masak would love to see some Apocalypses fanfic 11:51
a really good A09, for example :)
pmurias mberends: it's equaly unportable to other forms of C# interop 11:55
mberends masak: "and so they went back to the bikeshed. It was blue, which jnthn always hated, but he'd been outvoted that last time they chose its color." 11:56
im2ee Hmm, can i check if there is a client who is waiting to be accepted without accepting? (in IO::Socket) 11:58
tadzik ...why would you need that? :) 12:02
ingy o/
tadzik as in, what is the problem you want to solve? 12:03
ingy: o/
im2ee Hmm, tadzik im training perl6. I have some ideas and i want to try them out. :)
tadzik im2ee: I can see that, but why do you need to check if there's someone to accept()?
12:03 drbean left
tadzik im2ee: you want a non-blocking IO, maybe? 12:03
im2ee tadzik, yes, something like that. :) 12:04
mberends pmurias: you're right. Interop spec will only become standardized after a number of styles have been tried and compared. Styles that isolate Perl code in one file and and C# code in another file would be preferred, however.
im2ee tadzik, in p6 there is no multi-threading, right? :) So i want to get around :)
tadzik im2ee: IO::Select has some non-blocking IO capabilities, but I don't know how to apply them to this particular case
im2ee: maybe can_read() will return your socket if it has someone to accept()? 12:05
bbl
im2ee tadzik, i want to make tcp server which can work with more than 1 clients. :)
Woodi im2ee: on server side: accept returns new socket for each connsction probably 12:15
and to switch betwin clients usually select-like is used 12:16
12:20 drbean joined
Woodi do threads work in parrot ? 12:21
im2ee tadzik, where i can find something more informations about IO::Select? :) 12:24
12:25 lue joined
pmurias mberends: i don't think there is much difference between using C# code and .net specific libraries 12:26
jnthn mberends: wtf, blue is my fave color! :P 12:27
I mean, er, morning :)
12:28 Kivutarrr left
pmurias we might end up with CPAN modules that expose the interface to say GTK on multiple platforms, but programs that use interop generally won't be portable 12:28
mberends jnthn: morning blues, eh? good * anyways :)
tadzik im2ee: you can read the source :)
im2ee: that's precisely what IO::Select is for :) 12:29
pmurias mberends: unless both implementations support interop with the same platform
tadzik Woodi: threads don't
im2ee: I wrote IO::Select quite recently, but without any usecase for it. If you find something that sucks, tell me, and we'll think about making it better
im2ee: it's supposed to be modelled after Perl 5's IO::Select
12:30 lue left
im2ee tadzik, so, i can use can_read, yes? 12:33
tadzik im2ee: I think so, yes
mberends pmurias: I agree there is not much difference. I would just prefer to keep the two languages separate, so that changes in one file affect only the Niecza compiler and changes in the other file affect only the C# compiler. Merging them I don't prefer, which makes me feel desperate. Note that asp.net opts for language separation too, with code-behind files. 12:37
12:39 lue joined
im2ee b: my @tab = <a b c>; for @tab.keys { @tab.delete($_) if $_ == 'a'; }; say @tab; 12:44
p6eval b 1b7dd1: OUTPUT«Any()bc␤»
im2ee is there any function to clear array from Any()? :) 12:45
s/of/from/
tadzik not that I know of 12:46
jnthn Um, is delete meant to leave it behind?
tadzik maybe @tab.=grep($_)
well, that'll clear it from false values
jnthn *.defined
tadzik jnthn: I wanted to left that as an excercise for the reader :)
s/left/leave/
jnthn oh, sorry :P
im2ee Thanks. :) 12:47
jnthn, why *.defined? Why not just .defined? :) 12:50
.defined doesn't work, but why? :)
tadzik std: .defined
p6eval std c22b2ee: OUTPUT«ok 00:01 118m␤»
tadzik ek
im2ee: .defined doesn't construct a block 12:51
*.defined does
.grep({ .defined }) would probably work
im2ee So, clear now. :) Thanks tadzik
jnthn im2ee: { .defined } would work. 12:52
*.defined is just a bit shorter...and generates better code too, iirc.
im2ee Great! :) 12:53
masak im2ee: the reason a block is needed is that you don't want to call .defined *right now*, you want .grep to call it when it needs to. 12:58
im2ee: that whole idea is called a closure (or a callback), and I've written a blog post about it, if you're interested. strangelyconsistent.org/blog/speaki...-in-perl-6 13:00
im2ee Hmm, right, sounds logical. :) Thanks masak
masak I might be biased, but I think it's a good read ;)
im2ee masak, i will read it. 13:01
masak, i have some other problems.
masak, I want to write something like TCP server, which can work with more than 1 clients.
masak, and i must to check if there is anybody who is waiting for accept. 13:02
masak, can i do this without IO::Select ? :)
masak im2ee: sorry, I don't know enough about TCP/IP to help you. :/
im2ee: it sounds like you're sniffing around in the vicinity of asyncronous IO and event handling. Perl 6 is weak there at the moment. 13:05
13:06 Chillance joined
masak hm, it'd be a really interesting exercise to write A09. it'd have to comment a lot on the appropriate RFC's... and in order to be historically believable, it couldn't mention the @a[*-1] syntax, which was developed afterwards... 13:06
13:07 larion joined, larion left
dalek ecza: 493fbf9 | mberends++ | examples/gtk-sierpinski.pl:
[examples/gtk-sierpinski.pl] refactor and tidy up source
 13:08
tadzik masak: fwiw, metacpan.org/module/IO::Select#EXAMPLE shows the case as an example of what can be solved using select(), so maybe it's doable in nom 13:09
masak nom has select()? 13:10
tadzik more-less
github.com/tadzik/IO-Select/
masak in that case, more or less wow.
tadzik++
tadzik I wanted to provoke Mojolicious guys
masak :)
too bad you put it in a module, then! :P 13:11
tadzik they said "we'll port it once rakudo gets non-blocking IO". Few hours later I said "your turn"
still waiting for their reaction :)
13:11 drbean left
tadzik masak: I want it to be used in-field a bit before even thinking about making it core/speccing it 13:11
no one seems to have a view on how it should work like
masak nodnod 13:12
tadzik so let's just use it and see what sucks
Parrot was wondering for a few months "what interface should Select pmc have?"
13:12 kcwu left
tadzik and after some discussion we decided "let's ship it and let users criticise it" 13:12
13:13 drbean joined, f00li5h joined
tadzik IMHO, if this interface is good enough for CPAN, it's very likely it'll be good enough for us too 13:13
IO::Select is core since 5.00307 and I don't see anything that wants to replace it with something with a better interface 13:14
I didn't look around much though :)
masak I like that kind of thinking.
13:15 lue left
tadzik okay, back to sicp 13:15
13:18 risou_awy is now known as risou 13:21 pmurias left
masak tadzik: the book or the videos? 13:23
13:25 lue joined 13:26 snarkyboojum left 13:27 espadrine joined, snarkyboojum joined
cognominal ingy, no Chartreuse hangover? 13:27
13:35 lue left 13:38 pmurias joined
felher nom: state $blar; 13:39
p6eval nom 6fd077: OUTPUT«(signal SEGV)»
jnthn epic fel 13:40
felher :)
pmurias in roast/S01-perl-5-integration/basic.t a p5 routine exposed to perl6 has wantarray
masak submits rakudobug
pmurias with the way context work in Perl 6 now is there a way to make that work? 13:41
masak pmurias: doesn't feel like it.
13:41 envi left
masak or maybe Perl 6 callers always wants the same thing nowadays. 13:41
want*
13:42 envi joined
masak haven't thought about it. 13:42
pmurias it doesn't seem to be a thing that should be tested for in a basic test
masak nom: sub foo { state $blar = "OH HAI"; say $blar }; foo 13:43
p6eval nom 6fd077: OUTPUT«OH HAI␤»
jnthn masak: It's some oddity in the mainline. 13:44
pmurias should i take it out?
jnthn masak: Seen it before, evidently forgot to fix it.
masak pmurias: +1
jnthn tickets++
pmurias did blitzkost have tests of it's own? 13:45
13:45 woosley left 13:52 lue joined 13:56 zby_home_ joined, zby_home left 14:00 lue left 14:02 grondilu joined
grondilu So can't I provide arguments to a sub that has been defined as 'sub foo { }'? I mean, prototyping is mandatory now? 14:04
please don't answer all at once :) 14:05
benabik You can get around it some by… sub foo(|$c) { }, IIRc
Which declares a capture that grabs all the arguments.
grondilu well, yeah but that's 4 additional characters to type :( 14:06
masak grondilu: you must like the Perl 5 parameter handling very much ;) 14:07
benabik Someone has to. :-D
grondilu masak: I do indeed
jnthn grondilu: You have a lot of use cases where you want to take arguments and ignore them? 14:08
grondilu I like the forth-like aspect of not having to worry about argument passing.
benabik sub foo { $^these + $^are + $^parameters }
masak grondilu: well, 5 (by my count) extra characters is the "tax" you have to pay in Perl 6 so that the rest of us can have sane argument handling.
jnthn 4 14:09
grondilu For simple functions I mean. Of course, prototyping is necessary for complex stuffs.
jnthn (|$)
Don't need to name the variable you stick the capture in if you don't want it.
masak jnthn: yeahbut.
grondilu jnthn: cool
masak how will you extract the arguments from that?
benabik Why have a function that takes arguments that you can never get to?
masak exactly.
benabik grondilu: You can declare parameters inside your function. sub add { $^a + $^b } 14:10
grondilu ?
jnthn Also note that
rakudo: sub foo { say @_ }; foo; foo(1,2,3)
p6eval rakudo 6fd077: OUTPUT«␤1 2 3␤»
grondilu doesn't know about $^
jnthn If you mention @_ anywhere in your sub body, then it automatically generates a signature for you of (*@_)
masak rakudo: sub add { $^a + $^b }; say add 4, 5
p6eval rakudo 6fd077: OUTPUT«9␤»
grondilu jnthn: even cooler 14:11
masak jnthn: oh, I forgot about that! that should make grondilu very happy.
benabik jnthn: Does the same happen with %_?
jnthn Also note that if what you really want is a closure, Perl 6 doesn't require you to write "sub"
benabik: Yes
flussence rakudo: sub add { my $a = shift; my $b = shift; }; say add(4, 5)
p6eval rakudo 6fd077: OUTPUT«Too many positional parameters passed; got 2 but expected 0␤ in sub add at /tmp/Hze8d7ImSG:1␤ in <anon> at /tmp/Hze8d7ImSG:1␤ in <anon> at /tmp/Hze8d7ImSG:1␤␤»
flussence rakudo: sub add { @_; my $a = shift; my $b = shift; }; say add(4, 5)
p6eval rakudo 6fd077: OUTPUT«No applicable candidates found to dispatch to for 'shift'. Available candidates are:␤:(Positional @a)␤␤ in sub shift at src/gen/CORE.setting:4222␤ in sub add at /tmp/iRtE0BPsUH:1␤ in <anon> at /tmp/iRtE0BPsUH:1␤ in <anon> at /tmp/iRtE0BPsUH:1␤␤»
flussence oh.
jnthn flussence: No implicit argumetns.
*arguments
flussence rakudo: sub add { my $a = shift @_; my $b = shift @_; }; say add(4, 5) 14:12
p6eval rakudo 6fd077: OUTPUT«5␤»
flussence close enough
masak algorithm fail ;)
jnthn Saturating add :P
flussence rakudo: sub add { [+]@_ }; say add(4, 5)
p6eval rakudo 6fd077: OUTPUT«2␤»
flussence :(
benabik rakudo: sub foo { $_ = @_; my $a = .shift; say $a }; foo 5
p6eval rakudo 6fd077: OUTPUT«5␤»
benabik :-D
grondilu :) 14:13
flussence rakudo: sub add { [+](|@_) }; say add(4, 5)
p6eval rakudo 6fd077: OUTPUT«Could not find sub &prefix:<|>␤ in sub add at /tmp/AucDj7Fdd4:1␤ in <anon> at /tmp/AucDj7Fdd4:1␤ in <anon> at /tmp/AucDj7Fdd4:1␤␤»
benabik rakudo: sub add { [+] @_ }; say add(4, 5)
jnthn er, why on earth did [+] @_ not work...
p6eval rakudo 6fd077: OUTPUT«2␤»
flussence b: sub add { [+]@_ }; say add(4, 5)
p6eval b 1b7dd1: OUTPUT«9␤»
masak nom: sub add { [+](@_) }; say add 4, 5
p6eval nom 6fd077: OUTPUT«2␤»
benabik nombug?
masak submits rakudobug
jnthn nom: sub add(*@x) { [+] @x }; say add 4, 5
p6eval nom 6fd077: OUTPUT«9␤»
benabik A bug in the implicit @_? 14:15
grondilu raduko: sub add { [+]@_ }; say add 1, 1
rakudo: sub add { [+]@_ }; say add 1, 1
TimToady std: sub add { [+]@_ }; say add(4, 5)
p6eval rakudo 6fd077: OUTPUT«2␤»
std c22b2ee: OUTPUT«===SORRY!===␤Two terms in a row at /tmp/7tqKXg2sdR line 1:␤------> sub add { [+]⏏@_ }; say add(4, 5)␤ expecting any of:␤ POST␤ bracketed infix␤ infix or meta-infix␤ postfix␤ postfix_prefix_meta_operator␤
..prefix_circumfix_meta_o…
jnthn benabik: Yeah, that's my guess
TimToady space required after a reduce
grondilu rakudo: sub add { [+] @_ }; say add 1, 1
masak fairy nuff.
p6eval rakudo 6fd077: OUTPUT«2␤»
grondilu rakudo: sub add { [+] @_ }; say add 1, 42 14:16
p6eval rakudo 6fd077: OUTPUT«2␤»
benabik Two bugs! Parsing can @_ not working right in a reduce.
s/can/and/, wtf keyboard?
masak aye. 14:17
flussence b: sub add { [+] @_[0..*] }; say add(4, 5)
jnthn benabik: It's not reduce
benabik: It's something about the binding.
flussence er...
p6eval b 1b7dd1: OUTPUT«(timeout)»
TimToady extra capture parens?
jnthn git stash
er, wrong window
flussence rakudo: sub add { [+] @_[0..*] }; say add(4, 5)
p6eval rakudo 6fd077: OUTPUT«9␤»
jnthn TimToady: Almost certainly a failure to set a flatten flag or something like that. 14:18
masak removes all jnthn's changes and hides them
TimToady not marked as flatten
flussence well I've managed to break both rakudo branches today, that's an achievement :) 14:19
jnthn TimToady: Forget exactly how it works. Do know that we're not telling the binder something we should be by looking at the code though...
nom: ({ say [+] @^a })([1,2,3]) 14:20
p6eval nom 6fd077: OUTPUT«3␤»
jnthn nom: (sub (@a) { say [+] @a })([1,2,3]) 14:21
p6eval nom 6fd077: OUTPUT«6␤»
jnthn Yeah, same issue there.
Fixed locally, spectesting.
14:22 envi left
jnthn At least this was easier to find than whatever's wrong in my explosive multi-dispatch patches. 14:22
14:22 donri joined
grondilu rakudo: { say [+]@_ }(42, 1) 14:23
p6eval rakudo 6fd077: OUTPUT«2␤»
grondilu it does work on my machine
pmichaud good morning, #perl6 14:24
jnthn grondilu: Yes, I guess you're running from a released Rakudo.
grondilu "This is Rakudo Perl 6, version 2011.07 built on parrot 3.6.0 0"
jnthn grondilu: You just helped us find a regression in the current development branch. So, thanks! :-)
o/ pmichaud
grondilu lol, glad to have helped :) 14:25
masak grondilu++
grondilu is very enthusiastic about Perl and Perl6
I just love this language
masak we share your enthusiasm :)
flussence perl6: sub add { [+]@_ }; sub add2 { [+]@_[0..*] }; say add(4, 5); say add2(4, 5); #j/w
p6eval rakudo 6fd077: OUTPUT«2␤9␤» 14:26
..niecza v9-51-g493fbf9: OUTPUT«===SORRY!===␤␤Two terms in a row at /tmp/HGOaEiD2Ua line 1:␤------> sub add { [+]⏏@_ }; sub add2 { [+]@_[0..*] }; say add(␤␤Parse failed␤␤»
..pugs: OUTPUT«*** ␤ Unexpected "@_"␤ expecting dot, ":", "(", term postfix, operator or "}"␤ at /tmp/vWoyH1wAvI line 1, column 14␤»
flussence hm
perl6: sub add { [+] @_ }; sub add2 { [+] @_[0..*] }; say add(4, 5); say add2(4, 5);
p6eval rakudo 6fd077: OUTPUT«2␤9␤»
..niecza v9-51-g493fbf9: OUTPUT«(timeout)9␤»
..pugs: OUTPUT«9␤9␤»
flussence oh wow, I broke niecza too
dalek kudo/nom: efe3f82 | jnthn++ | src/Perl6/Actions.pm:
Fix bug with placeholder parameter binding of arrays.
 14:27
14:31 lue joined
grondilu got to go. Keep on the good work guys! 14:34
14:34 grondilu left 14:46 whiteknight joined 14:47 lue left 14:50 satyavvd joined
sorear good * #perl6 14:51
masak *
colomon */
14:52 im2ee left
pmurias sorear: hi 14:54
sorear pmurias: the .net platform does not come with a C# compiler, and bundling Mono.CSharp.dll would double the size of the niecza package 14:56
15:04 kaare_ left 15:06 im2ee joined
pmurias sorear: i'll work on more usefull thing for now then 15:08
sorear: does niecza work on .net?
tadzik masak: the book 'fcourse
15:10 pmurias_ joined
colomon just uttered the words "It seems like it's random whether or not it happens" and immediately launched valgrind on his $work bug from hell. 15:10
masak tadzik: nice! godspeed. 15:12
15:13 pmurias left 15:17 lue joined 15:26 lue left
tadzik seems that guile uses Rats for calculations as well 15:26
15:28 satyavvd left 15:29 MayDaniel joined
masak tadzik: Lisp people are pretty proud of their en.wikipedia.org/wiki/Numerical_tower 15:30
15:30 pmurias joined 15:31 MayDaniel left
tadzik I ended up writing my (define (numify x)) :) 15:32
15:32 birdwindupbird joined, birdwindupbird left 15:33 pmurias_ left 15:36 mj41 joined
pmichaud general note: I'm still tuit deprived. Yesterday ended up not being a very good day. :-/ 15:36
s/not being/being not/ 15:37
mberends :(
colomon :(
15:38 lue joined
colomon has found his $work bug, which appears to be something colossally stupid he did nearly a decade ago... 15:44
dalek kudo/nom: c4c2e1d | jnthn++ | src/binder/bind. (2 files):
Move/share a struct definition.
 15:45
kudo/nom: 03afd02 | jnthn++ | src/binder/ (2 files):
Switch junctional arguments check in multi-dispatcher over to being able to cope with natively typed arguments.
kudo/nom: 9e831e0 | jnthn++ | src/binder/multidispatch.c:
Tweak the multi-dispatch candidate check to try and make it aware of native types.
pmichaud afk again
15:46 huf joined
masak pmichaud: :( 15:48
[Coke] yawns. 15:50
mberends masak: this was around the time I'd suggested for pugs and haskell exploration, primarily because I thought it stood a good chance of being possible for others too. I could also make other times. I'm now writing a GUI solution for your matchboxes puzzle, but am happy to look at pugs any time. 15:55
masak mberends: cool. 15:56
mberends masak: shall we look at pugs in, say, an hour?
15:57 pmurias_ joined
masak mberends: hm. I was planning to go out for a run around that time. 15:57
mberends masak: ok, what time suits you?
masak but sometime during the evening I think I could have some time to download Pugs and try to get it to run. 15:58
mberends: mostly I'd like to see you and [Coke] agree on a time that works for both of you.
right now I'm going AFK for ~30 minutes.
mberends ok, me too
15:59 pmurias left 16:03 lue left
japhb has a lot to backlog ... anyone happen to know if my question last night about how to assign $?USAGE properly in the Rakudo setting got answered? (By the time I finish backlogging, the people who could answer will probably be gone :-/ ( 16:10
16:13 pmurias joined 16:14 espadrine left, im2ee left, im2ee joined 16:16 pmurias_ left 16:19 lue joined, envi joined
dalek kudo/nom: 8b7a2a5 | jnthn++ | src/binder/multidispatch.c:
Get multi-dispatch cache to handle natively typed arguments.
 16:20
japhb jnthn++ is awake! 16:21
jnthn japhb: I didn't answer it...but I probably could ;)
japhb jnthn, Once I have the message string, how do I create the $?USAGE variable in Rakudo, and set it to my newly-minted message? (In the setting, of course)
jnthn nom: say $?USAGE
p6eval nom 9e831e: OUTPUT«===SORRY!===␤Symbol '$?USAGE' not predeclared in <anonymous> (/tmp/7eqQXlAlF_:1)␤»
japhb jnthn, thx
jnthn japhb: Is it actually meant to live in the setting?
Not just in the mainline of the user's program?
jnthn had assumed it was the latter 16:22
japhb IIUC, the main helper generates it, and sets it ... in *some* scope ... such that the user's MAIN and USAGE can see it. But I don't know what that scope should be. 16:23
I meant "assigning it from within code that exists in the setting"
jnthn ah, here's the bit of S06... 16:24
This usage message is automatically generated from the signature (or
signatures) of C<MAIN>. This message is generated at compile time,
and hence is available at any later time as C<$?USAGE>.
16:24 Maddingue joined
japhb has been updating MAIN_HELPER() in src/core/Main.pm. 16:25
jnthn oh
it does
my $m = callframe(1).my<&MAIN>;
japhb right
jnthn In theory
16:26 lue left
jnthn callframe(1).my<$?USAGE> = $the_message 16:26
However, you need to make sure a $?USAGE variable exists to stick the thing in
That can probably be done by editing Actions.pm
There's a
if $unit.symbol('&MAIN') {
Which adds the call to &MAIN_HELPER 16:27
japhb The current flow in my version is: 1. Fetch MAIN, 2. process cmdline args, 3. generate usage message, 4. attempt to dispatch to matching &MAIN, 5. Failing that, attempt to call &USAGE, 6. Failing that, print generated usage message and exit.
jnthn It could also add code to declare the variable.
japhb OK, understood so far.
jnthn Thing is, the spec implies that we figure out the usage message at compile time.
Heh 16:28
I wonder what happens if we add a .usage() method into Routine ;)
So you can generate usage for anything <grin>
japhb chuckles
jnthn Anyway, the beneift would be that the compiler could do
my $message = $*ST.find_symbol('&MAIN').usage(); 16:29
Then create a constant
And bind it (again, at compile time) to $?USAGE
:)
(Just using the same static lexpad mechanism we use to install all other declarational things).
That was, if somebody writes
sub USAGE() { 16:30
say "OH LOOK HERE'S SOME HELP";
say $?USAGE
}
Then it'll all work out.
s/was/way/
(Essentially, you're running the code to generate the usage message at BEGIN time.)
japhb Ah, I thought all that was necessary for *that* to work was to "declare" $?USAGE in the compiler, and have it *assigned* in MAIN_HELPER ... because USAGE is called from MAIN_HELPER. Though I suppose 16:31
jnthn Well, we could do that too.
japhb the mainline code could have called USAGE manually, and the user might be surprised it's not there, because MAIN_HELPER never got called
jnthn Right.
Also
We may need to twiddle the way we find &MAIN
As this is also meant to work: 16:32
class Foo {
sub MAIN() { ... }
}
e.g. MAIN is allowed to be somewhere other than in the mainline.
tadzik wow, that's specced? 16:33
jnthn I think the easiest way to do that is to have a :my $*MAIN; declared in comp_unit in the grammar
japhb oh. Hmmm. So how would the compiler know which MAIN to generate $?USAGE from at compile time?
jnthn And routine_def looks for if the thing is called MAIN when doing symbol installation and shoves a reference to it in $*MAIN
tadzik: I thought so.
tadzik: Or I seem to recall TimToady++ expecting it would work, at least.
tadzik I see
jnthn I guess that 16:34
class A { sub MAIN() { } }
japhb Would the compiler need to generate a $?USAGE for each comp_unit, and have them all separate?
jnthn class B { sub MAIN() { } }
Would be an error.
(Ambiguous MAIN)
japhb: comp_unit is kinda top-level :)
japhb jnthn, I'm more worried about
...hold, child hurt
jnthn ouch
japhb ah, more hurt pride than body 16:35
jnthn ah, phew
japhb quite
So: My reading of the spec was that every module might define a MAIN for self-testing or magic behavior, Python-style, but they also might be 'use'd into another module or even the mainline program, all of which might define their own MAIN. So it's only after all compiling is done and runtime begins (or according to spec, when runtime falls off the end of the "main" programs mainline, that MAIN is checked for and called. 16:37
16:38 mj41 left
japhb But ... the mainline code could call USAGE manually. So when is the decision made about which MAIN the $?USAGE is set from 16:38
?
jnthn I don't think there's a problem (more)
If $?USAGE is a lexical in the mainline, and there's one MAIN per compilation unit too, then MAIN_HELPER is going to see the correct lexical scope. 16:39
e.g. it gets called from the lexical scope of the program; callframe looks 1 frame down, which will be the mainline with the correct $?USAGE in. 16:40
japhb Ah, OK. So each comp_unit gets a $?USAGE lexical. I can see that making perfect sense
16:40 uasi left
jnthn Yes, I think so. 16:40
japhb It also has the advantage that all modules will know the $?USAGE of their MAIN, even if they aren't the mainline. 16:41
jnthn Yes. 16:42
oh noes, I'm out of curry paste 16:44
shop, bbl &
japhb OK, so lessee. (thinking aloud) If as you suggest we add a .usage to Routine, I could do two branches in the code: if it's a proto, pull .usage from each candidate, and join with 'or' or somesuch; if its not a proto, generate the calling info from the signature. Hmmm, I wonder if I need to return back something more than a simple string, because
jnthn oh, can wait a moment actually... :)
japhb It may be desirable to have a top section of the summary and a bottom section explaining aliases and such.
jnthn Hmm...I was thinking a string would be enough for now.
japhb The current MAIN_HELPER code tries to do something like that, but I'm not sure yet whether to drop that and just return a string, or keep supporting it. 16:45
jnthn If people want something much fancier that's what USAGE is for.
japhb jnthn, sure ... but the question is, would .usage need to return more complex information that USAGE needs?
I guess we can play that by ear. :-)
jnthn Yeah, I think so.
(play that by ear, that is) 16:46
japhb thx for all the answers
jnthn Welcome. Thanks for hacking on this :) 16:47
japhb afk for a few & 16:50
jnthn really going to shop &
16:55 Trashlord left 16:56 wolfman2000 left 17:01 replore left 17:08 wamba left, risou is now known as risou_awy, risou_awy is now known as risou 17:10 ZapZ_ left 17:14 carlin left 17:15 carlin joined, carlin left 17:16 carlin joined, carlin left 17:17 lue joined, carlin joined 17:21 lue left 17:24 risou is now known as risou_awy 17:32 envi left 17:36 daniel-s left 17:51 lue joined
mberends GUI solution for masak's matchboxes puzzle on Niecza: gist.github.com/1240895 17:53
17:54 kaare_ joined
colomon mberends++ 17:54
mberends sorear++ 17:56
17:56 lue left 18:11 cognominal left
masak mberends: wow -- looks cool! I wish I could run it. 18:11
18:12 cognominal joined
mberends masak: try installing another Ubuntu or Debian guest VM, when you have time 18:12
18:16 cognominal_ joined 18:19 cognominal left 18:20 colomon left 18:21 lue joined 18:24 soh_cah_toa joined 18:27 colomon joined
masak mberends: will do. 18:32
mberends masak: ok. I looked into audreyt's github repo list, and there is a pugs repo that looks unchanged since 2006 and a Pugs.hs that seems to be actively maintained. We probably need the latter. 18:34
18:37 cognominal_ left, cognominal_ joined
masak nod. 18:38
I concluded the same.
Tene jnthn: you might also consider changing the spec, and divorcing "Run this when the module is loaded" from "Run this when this file is passed as the first argument to a perl interpreter" 18:41
Perhaps both of those are aliased to the same code by default.
MAIN vs LOAD or something 18:42
18:43 pmurias_ joined 18:45 pmurias left
jnthn Tene: Hmm, interesting thought. 18:55
jlaire compiling pugs on this laptop is taking a while... :) 19:11
Pugs.hs on github seems to match the pugs package on hackage 19:14
so it's definitely the right one
benabik Looks like audreyt tried compiling it yesterday and discovered some updates were needed. 19:15
jlaire hmm, I should update my ghc 19:16
benabik Looks like the Haskell Platform release has been somewhat delayed. 19:17
19:17 thou joined
jlaire I've never tried the platform 19:18
benabik It's GHC + cabal + useful libraries. Something like: Platform : GHC :: Star : Rakudo
mberends according to pugs/README ghc 6.10 is sufficient, I'm using Ubuntu's 6.12.3 but not getting very far yet. 19:19
benabik The webpage says the next release is July 2011, so I'm guessing there's a delay, but I don't see any mailing list or discussion links.
jlaire yeah, but installing ghc and libraries separately has been very painless for me
I have GHC 7.0.3 19:20
benabik It's been convenient for me, lacking a proper package manager. :-/
jlaire :/
fortunately for me, haskell is very well supported on arch linux
19:21 pmurias joined
mberends 'cabal install HTTP zlib' went very well, now trying 'cabal install Pugs' 19:22
19:23 pmurias_ left 19:26 wallberg joined 19:30 furura joined
mberends hmm, Pugs install failed because terminfo failed to install, and terminfo failed because curses.h was not found. Installing libncurses5-dev in hope.. 19:31
benabik Down the dependency rabbit hole.
jlaire real 35m22.509s 19:34
works for me
mberends jlaire: is that with cabal or make? 19:35
jlaire mberends: cabal
I ran 'cabal install' in Pugs.hs/Pugs
and have a working pugs repl now 19:36
mberends: what platform are you on?
mberends jlaire: here too, Ubuntu 11.04 x86. cabal seems to ignore the current directory and build everything in ~/.cabal, which is unsuitable for hacking.
jlaire :S 19:37
masak I was just wondering whether that might happen. 19:38
19:38 abercrombie joined
masak cabal is nice and all... for end users. 19:38
19:39 packetknife joined
mberends error [37 of 89] ... src/Pugs/AST/Internals/Instances.hs:216:9 No instance for (Typeable Unique) ... 19:39
masak aww
mberends since Pugs obvious builds on some platforms, I hereby deem this platform unusable. :( 19:40
masak wait, but that's a *compile* error.
jlaire might be differences between old/new ghc or base libraries
masak shouldn't that depend more on the compiler than on the platform?
jlaire yep 19:41
but installing up-to-date haskell environment is easier on some platforms than others
mberends base libs, most likely. There was a 'base 3.0 deprecated' warning during the building of deps.
benabik Pugs.hs master says "build-depends: base >= 4" 19:42
mberends thinks 'cabal upgrade base' 19:43
jlaire why does it start compiling with unsatisfied dependencies :?
masak mberends++ # explorer 19:44
benabik That is what's on github. Perhaps it's not what cabal is using. 19:51
jlaire cabal uses *.cabal in the current directory
masak I thought that was the point of the Pugs.hs repo. 19:52
jlaire if you say just 'cabal install'
benabik I said perhaps… Haven't used cabal myself.
mberends cabal download Pugs 6.2.13-20110925, that is only a few hours old 19:53
19:53 Chat4236 joined
dalek Heuristic branch merge: pushed 121 commits to rakudo/optimizer by jnthn 19:54
Chat4236 hey
masak Chat4236: hi!
Chat4236: I see you're a Chatmosphere user. welcome.
Chat4236 hey 19:55
masak Chat4236: this is a very friendly, accomodating channel. we like visitors. but I should warn you: it's about Perl 6.
Chat4236 thnx
masak there, you have been warned.
Chat4236 perl 6? 19:56
wats dat?
masak Chat4236: Chatmosphere users often miss this fact, and embarass themselves by asking "where are the chicks" and similar things.
Chat4236: it's a programming language, like Python or C.
Chat4236 kBoy gets into bed :* with a girl and notice her clean shaved vagina:O .n asks 'baal kyon nahi hain'.=-? she replies-'tu sex karne aaya hai ya kangi ??':x
hehe 19:57
masak ok, that would be an example of inappropriate behavior.
mberends Chat4236: we don't want that kind of chat here
Chat4236 k
19:57 Chat4236 left
flussence good thing you got the first word in. My first instinct would be to disembowel such people with a blunt object... 19:57
benabik Why do Chatmosphere users keep showing up here?
tadzik a frying pan
benabik: a Chatmosphere dev hates us
masak tadzik: oh? 19:58
tadzik masak: just a guess
jnthn Good theory :)
masak I thought it was just randomizing channels on freenode.
jnthn "How can I make Perl 6 take even longer?" "oooh...I know..."
flussence more likely it was written by someone with no clue how IRC works, so they just have a "random server/channel" option
masak right.
benabik flussence: The chatmosphere.org website has a big "Random selection of chatrooms" on the front page, so it seems you're likely correct. 19:59
"On the home screen on the app there is a random list of popular chatrooms."
benabik sighs.
20:00 SHODAN left
flussence ) # fixing that quit message 20:01
20:02 saaki left
flussence
.oO( and now I've broken it for people who hide quit messages... )
 20:02
masak "Hi Chatmosphere. We'd like you to exclude our channel from your random channel selection. Neither we nor your users enjoy the others' company." 20:04
benabik masak++ 20:05
soh_cah_toa ok, i have to ask b/c it's bothering the heck outta me: why do i always see people using .oO(blah blah blah)? what in the world is that? 20:09
masak soh_cah_toa: thought bubbles.
soh_cah_toa: like in comics.
mberends has no options left except 'cabal install cabal-install', to *upgrade* from the existing Ubuntu-supplied cabal-install 20:10
masak :)
20:11 mj41 joined 20:12 fridim_ joined
soh_cah_toa ah, i see 20:13
mberends failed. 'cabal install base' dies with Distribution/Client/Dependency/TopDown.hs:169:37-73: Non-exhaustive patterns in lambda, and 'cabal upgrade base' dies with fromFlag NoFlag. Use fromFlagOrDefault, about which StackOverflow basically said 'ur Linux is doin it rong'. This is the end of the road for GHC 6.x on Ubuntu. 20:15
20:15 lue left
pmurias is Ada95 a sensible language to learn? 20:17
benabik Ada?
tadzik nqp: my @a := []; @a.push(2); @a.push(7); say(@a[-1]) # intentional?
p6eval nqp: OUTPUT«7␤»
tadzik that's not really Perl6-y
benabik 95?
pmurias we have a parrallel ai algorithm course in uni which uses Ada95
soh_cah_toa wow 20:18
i would expect common lisp or scheme for a course like that
jlaire if the focus is on algorithms, ada is fine imo
weird choice tho
soh_cah_toa nopaste.snit.ch/81994 20:21
i could use some advice on that. i'm trying to test that my module exports it's subroutines properly
mberends pmurias: gnat, the GNU Ada compiler, is very actively maintained, which suggests a lot of use as well, so it seems a good idea. Ada support real-time processing too. Main Ada site: www.adaic.org/ 20:22
benabik I _think_ that use and export just make the names visible, not add them to the namespace.
*current namespace
20:22 lue joined
im2ee Good night! :) 20:23
ingy cognominal_: strangely no Chartreuse hangover!
20:23 im2ee left
cognominal_ :) 20:24
20:24 packetknife left 20:25 soh_cah_toa_ joined
soh_cah_toa_ hates his laptop battery 20:25
anyway...
20:26 soh_cah_toa left 20:27 mkramer1 joined, soh_cah_toa_ is now known as soh_cah_toa
soh_cah_toa so 'use' doesn't import subs into the current namespace? that's...weird 20:28
then i guess the keyword 'export' is kinda inappropriate as it does not actually "export" anything
jnthn Import in Perl 6 is lexical.
benabik It exports it into the lexical scope, but not the class it's in. I think. 20:29
20:29 mkramer left
soh_cah_toa hm... 20:29
jnthn sub foo is export { } # adds foo to the export list
benabik Which is really useful, because you don't want your classes sprouting a bunch of methods just because your library decided to export some stuff.
jnthn use module_with_foo_in; # imports everything in the default export list into the current lexpad.
benabik jnthn: But if you're inside a class, that doesn't mean they're imported into the class, right? 20:30
jnthn benabik: I don't get "imported into the class"
benabik: Note that writing a sub in a class doesn't put it in the method list. Importing something inside a class body doesn't make it a method either. 20:31
benabik jnthn: From soh_cah_toa's example: class A { use Foo::Bar; #`{exports baz} }
jnthn: Is there an A::baz after that?
20:31 kaare_ left
jnthn No, because the import is lexical (my scoped) 20:31
There's a way to make it import into the package instead, however.
We don't implement that in Rakudo yet; S11 does have it. I forget the exact syntax. 20:32
benabik our use?
jnthn No, I don't think it's that. :)
Though could be interesting
has use Foo; # adds all the imported things as methods...omgz :)
Yeah, maybe too "interesting" :)
benabik perlcabal.org/syn/S11.html#Compile-...mportation 20:33
jnthn Anyway, the syntax is something you write after use.
benabik use Sense :EXPORT
jnthn No, that's not going to give an A::baz, iiuc 20:34
use Foo::Baz :OUR<&baz> maybe
But you'd think there would be a way to say "bring 'em all in as our-scoped things"
benabik Oh, that would cause A to export baz. Buh, confusing.
soh_cah_toa unless i'm misunderstanding something, doesn't that fact that 'use' is lexical imply that `package A { use Foo::Bar }` should have foobar() imported into it?
package A creates a lexical scope, right?
jnthn Every pair of curlies implies a lexical scope 20:35
soh_cah_toa package A { ... } yes
jnthn To be clear though
soh_cah_toa right
jnthn rakudo: module Foo { sub x() { say 42 } }; Foo::x()
p6eval rakudo 8b7a2a: OUTPUT«Could not find symbol 'Foo::&x'␤ in <anon> at /tmp/MI25wdfyl3:1␤ in <anon> at /tmp/MI25wdfyl3:1␤␤»
jnthn rakudo: module Foo { our sub x() { say 42 } }; Foo::x()
p6eval rakudo 8b7a2a: OUTPUT«42␤»
jnthn In the first case, x is my-scoped (lexical). In the second, it's also installed into the package Foo 20:36
soh_cah_toa so i need to change the sub definition to 'our sub foobar is export { ... }' then?
jnthn Not unless you want to be able to import it also.
sorear good * #perl6 20:37
soh_cah_toa ugh, this is confusing
benabik o/ sorear
jnthn If you just want it visibile thorugh a fully qualified name (like Foo::x) you just need to makr it "our"
soh_cah_toa what i want is Foo::Bar to export a bunch of subs but i also want to be able to test that it's doing so
benabik soh_cah_toa: { use Foo::Bar; #`{ do tests here} } 20:38
Instead of { use Foo::Bar } #`{tests here}
mberends emerges from the cabal dependency rabbit-hole, ragged and defeated 20:39
20:39 lue left
soh_cah_toa how do i fully qualify a sub from the "default" namespace though? w/o a fqn, it's interpreted as a call 20:41
i don't know the default namespace name
::foobar perhaps?
sorear ping jnthn 20:42
jnthn Do you mean
sorear soh_cah_toa: GLOBAL::foobar
soh_cah_toa ok
jnthn rakudo: module Foo { our sub x() { } }; &Foo::x # if you want a ref, not a call
p6eval rakudo 8b7a2a: ( no output )
jnthn sorear: pong
benabik soh_cah_toa: Following on jnthn's example { use Foo; &x } (if x was exported) 20:43
soh_cah_toa nopaste.snit.ch/81995
that does not pass
sorear jnthn: I am attempting to figure out how role parametricity is implemented in nom
jnthn soh_cah_toa: Sorry if I'm misunderstanding a bunch of your questions. I'm a little tiredish today...
soh_cah_toa :) that's alright
jnthn sorear: OK.
sorear jnthn: in particular, in role Foo[$x] { method bar() { $x } }, how does &bar come to close over $x?
I've run out of obvious places to check 20:44
jnthn sorear: It's just a closure clone.
sorear: See specialize_with in...ParametricRoleHOW iirc.
benabik soh_cah_toa: Try ok(defined &eq_irc)
sorear was looking for RoleHOW
jnthn src/Perl6/Metamodel 20:45
sorear: No, there's not one "RoleHOW".
sorear looking
jnthn Been there, done that. :)
sorear: See *RoleHOW.pm
Also GenericHOW.pm
soh_cah_toa benabik: yeah, that works. thanks
jnthn Also *Applier.pm
20:47 mberends left
sorear grumbles at the unachievable elegance of this 20:47
jnthn sorear: There's 7 or 8 things in Metamodel:: that are related to making roles work in all their glory. If I can offer you any "don't make the same mistakes I've done over the years" style advice, it'd be don't conflate things that are different. 20:48
By that I mean 20:49
role Foo[::T] { }; class C does Foo { }
So, how many forms can this show up in?
Foo - well, clearly this isn't actually the "real role" itself because we can also declare a role Foo[::T1, ::T2] { } 20:50
So it must be some kinda "grouping" thing
benabik jnthn: Does that make T ~~ C in Foo?
jnthn But we have to be able to know which one we're talking about composition time.
benabik: No
So we're already up to two "forms" 20:51
Now consider that when we compose it into the class we need a specialized form.
felher Whats the preferred way to get the indicies of elements in an array that match some condition? Like grep, just for indicies. @array.kv.map: {$^k if $^v ~~ condition};?
jnthn With the generic bits - including ::?CLASS - reified.
That's a third
Then, just to ice the cake, Foo[Int] is also something we can talk about first class.
But it's not the same as the fully specialized thing because we haven't got the information we need to specialize ::?CLASS yet. 20:52
So it's a forth form.
I call these four ParametricGroup, Parametric, Concrete and Curried. 20:53
benabik If you have a Foo[::T], what does class Bar does Foo mean? I'd think you'd have to give it a parameter.
jnthn benabik: er, darn
:)
benabik: It means "compile time error"
I mean class Bar does foo[Int] { } or so 20:54
Sorry.
*Foo[Int]
benabik jnthn: That was why I was asking about T ~~ C in your example. ;-) Glad to know I hadn't missed something.
jnthn nom: role R[::T] { }; class C does R { }
p6eval nom 8b7a2a: OUTPUT«===SORRY!===␤None of the parametric role variants for 'R' matched the arguments supplied.␤No applicable candidates found to dispatch to for '_block1377'. Available candidates are:␤:(Mu, Mu)␤␤»
jnthn Hmm...some error awezomization needed in the second part of that.
But the first line of the eror is kinda spot on. 20:55
OK, time for me to get a beer and hack on the optimizer for a little bit.
This will probably screw up my sleep tonight, but given I slept 12 hours last night I'm done for anyway. :/ 20:56
sorear heh same here ... 20:57
benabik nom: role R[::T] { }; class C does R[int] { } # curious... 20:58
p6eval nom 8b7a2a: ( no output )
benabik \o/
jnthn Oh my :)
It's not going to handle that so cleverly as you may wish though, I suspect.
benabik jnthn: Fair enough. :-D 20:59
jnthn I mean, it'll work
But under the hood it's probably going to box/unbox in places it maybe shouldn't.
Fixing that is a...research problem.:)
jnthn holds out a research problem to sorear++ to see if he bites :)
I wonder if we can do anything insane on the CLR to get its generics support to do some of the lifting on that front... 21:00
Figuring out the details makes my brain want to run off and hide though. :) 21:01
And on the JVM it's back to square one since their generics are type erased.
benabik And handle native types via boxing anyway 21:02
jnthn Well, that's what passing a native type to a role will do.
In Rakudo today I mean.
It'd be great fun to improve/fix that some day. 21:03
21:03 Kivutarrr joined 21:06 newsbad_com left 21:07 newsbad_com joined 21:11 Kivutarrr left
sorear jnthn: What problem are you holding out? R[int]? 21:15
jnthn: What is type pretence 21:17
jnthn: What is the difference between new_type and new? 21:18
jnthn sorear: R[int], yes :) 21:21
sorear: Type pretence is how a role behaves like it's Any
Even thought that doesn't really "make sense"
new - what you expect. A new meta-object.
new_type - makes a type object (and thus an s-table)
We pretty much never call .new from the outside. 21:22
Always new_type
21:34 pmurias left
soh_cah_toa perl6: sub foobar(Str:D $str) { say "$str is defined" }; foobar('qux') 21:37
p6eval niecza v9-51-g493fbf9: OUTPUT«===SORRY!===␤␤In parameter declaration, typename 'Str:D' must be predeclared (or marked as declarative with :: prefix) at /tmp/hRdGoq8gOh line 1:␤------> sub foobar(Str:D⏏ $str) { say "$str is defined" }; foobar␤␤Parse f…
..pugs, rakudo 8b7a2a: OUTPUT«qux is defined␤»
sorear pugs? really? 21:38
soh_cah_toa aw darn, 2011.07 gives me "Invalid typename in parameter declaration" when i use :D or :U
nom is taking too long, i'm getting antsy ;) 21:39
jnthn soh_cah_toa: Yes, implementing that in nom :)
21:39 zby_home_ left
jnthn *implemented 21:39
soh_cah_toa: Yeah, I know...we're getting there.
21:39 mj41 left
soh_cah_toa it's getting annoying b/c that also means i need to stay w/ parrot 3.6.0 21:40
if i just install nom/HEAD, can i use 3.8.0? 21:41
benabik Parrot 3.8.0 works on Rakudo HEAD. Parrot HEAD works with nqp/kill-useless-defaults and rakudo/HEAD (last I tried) 21:42
soh_cah_toa new rakudo branch = old nom branch, right? 21:43
i know something like that got changed a while ago
jnthn soh_cah_toa: nom is the current branch
soh_cah_toa: If you git clone Rakudo today, you get the nom branch
(master is just a convention)
soh_cah_toa so what branch am i on if i 'git checkout nom'? 21:44
jnthn ...nom? :)
soh_cah_toa and if i 'git checkout master' i'm also on nom?
jnthn soh_cah_toa: No, you get "branch doesn't exist" or some such
21:45 cognominal___ joined
soh_cah_toa nope, i can go to master just fine 21:45
jnthn Oh?
:S
soh_cah_toa git branch -v
master d7f45ff Cool.eval
jnthn huh.
That sounds...kinda odd.
soh_cah_toa and that's after rebasing
jnthn I wonder if somebody accidentally pushed to master. 21:46
benabik soh_cah_toa: Did you just clone it, or have you had it for a while?
sorear jnthn: git pull won't delete branches
jnthn sorear: oh.
soh_cah_toa i've had it for a while
jnthn ah
OK.
soh_cah_toa i gotta re-clone?
jnthn sorear: Makes sense...
benabik Also, master appears to still exist in rakudo.it
git
sorear soh_cah_toa: no, just don't check out master anymore
check out b or nom
21:48 cognominal_ left 21:49 espadrine joined
soh_cah_toa bleh...this "nom = master but master = DNE except for soh_cah_toa's machine" business is crazy 21:49
i'm checking out nom
installing
i dont' care
:P
benabik 1) Your local branches don't have to correspond to anything remote. 2) master does exist in rakudo.git, but it's not the default branch. 21:51
soh_cah_toa ok
benabik nom = HEAD
sorear jnthn: now I'm wondering a bit about ze archetypes business... 21:52
jnthn sorear: heh, I shoulda gotten around to writing the blog post about those...
sorear: See slides on my talk from YAPC this year
ingy seen TimToady
aloha TimToady was last seen in #perl6 7 hours 34 mins ago saying "not marked as flatten".
jnthn sorear: I put the best description I did so far for archetypes in there. 21:53
sorear jnthn: link?
jnthn jnthn.net/papers/2011-yapceu-rakudo.pdf
sorear thanks
benabik jnthn: The book from that presentation is one of my textbooks this year. 21:54
masak "inheritalizable".
benabik (Types and Programming Languages)
jnthn benabik: It's very nice. :)
Enjoy :)
masak benabik: it was the book that made audreyt begin Pugs.
jnthn masak: Derivational morphology. :p 21:55
masak jnthn: "yesish" :)
jnthn Oh stop yesishing. 21:56
masak ok, so. which implementation will be the first to support infix:<%> in regexes?
Niecza, or Rakudo?
tadzik yapsi
jnthn yapsi
masak hah!
tadzik jnthn: high five! 21:57
Erm, high six i mean
jnthn \o/
masak yapsi doesn't even have regexes yet.
but I'm glad you still remember Yapsi. it hasn't had a release in a while.
dangerous with implementations with low bus numbers...
sorear nqp: BEGIN { say(5) } 22:12
p6eval nqp: OUTPUT«Confused at line 1, near "BEGIN { sa"␤current instr.: 'nqp;HLL;Grammar;panic' pc 23698 (src/stage2/gen/NQPHLL.pir:6757) (src/stage2/gen/NQPHLL.pm:329)␤»
masak confused from the very beginning!
sorear holds the most important difference of niecza to be the lack of a nqp/rakudo like split 22:13
tadzik . o O ( sorear uses confusion! It's super-effective! )
masak .oO( 9 out of every 10 sorears recommend confusion ) 22:14
jnthn Not sure nqp is going to survive it out the year without learning about BEGIN, fwiw. 22:15
(I'm not thrilled about making it do so, but I suspect I may need it at some point.) 22:16
We'll see.
sorear jnthn: if nqp does learn BEGIN, how will box operators know what type to use?
jnthn sorear: Confused :) 22:17
sorear: I think I'm missing something.
sorear niecza's kernel keeps a bunch of globals like Kernel.StrMO; this seems the main obstacle to a more nom-like runtime/compile split
jnthn: does/will nqp have anything like perl6_set_parcel_type ?
will nqp have any ops that need to find NQPMu or something like that? 22:18
jnthn sorear: I'm really not sure yet. Note that NQP on Parrot uses Parrot PMCs to break some circularities. 22:19
sorear: And when I did NQP on CLR, I ran into...fun...on things like that.
I can't remember exactly what happened.
I must have had to deal with it somehow.
Ask me again when I get somewhere with the nqpclr update/refactor. :) 22:20
sorear What does $method.insantiate_generic call? 22:21
jnthn Routine's instantiate_generic method.
sorear What file is RoutineHOW defined in?
jnthn ? 22:22
22:22 wallberg left
jnthn There's no RoutineHOW 22:22
Routine is just a class.
Routine is partially set up in BOOTSTRAP.pm
And partially in src/core/Routine.pm. 22:23
masak routines aren't type-y things like classes and roles and enums. that's why there's no RoutineHOW. 22:24
jnthn That's a good way of looking at it. 22:25
22:32 PacoLinux_ left 22:35 static_perl joined
sorear jnthn: is create_lexical_capture_fixup responsible for role Foo[$x] { method bar { $x } } working? 22:36
22:36 aloha joined
jnthn sorear: Yeah. 22:36
sorear: er
wait
I think actually no
No, it's not that. It's fixing up the role's outer pointer at fixup time.
*role body's outer pointer 22:37
sorear jnthn: I'm still trying to find what *is* responsible...
jnthn sorear: instantiate_generic on Routine 22:38
sorear: It calls self.clone() or some such
sorear: That includes capturing the outer pointer.
Since we just ran the body block, those were all capture_lex'd correctly. 22:39
sorear still thinks that is the most eewsome thing in Parrot 22:40
dalek kudo/optimizer: 8db01ef | jnthn++ | src/binder/multidispatch. (2 files):
First sketch of algorithm for considering multiple dispatches at compile time.
 22:41
benabik sorear: capture_lex?
22:42 replore_ joined
masak good night, channel. 22:42
tadzik gnightmasak 22:43
ha, made it!
masak :P
jnthn night masak!
lol me too
tadzik bet he's waiting for us today
masak tadzik: and no, I don't backlog to see who replied. not until the next morning :P
tadzik ;P
jnthn bwaha :)
22:43 masak left
jnthn ok, how the heck do I write this analyser up... 22:43
tadzik haha, watch this
22:43 replore_ left
tadzik GOOD MORNING MASAK! 22:43
jnthn s/write/wire 22:44
flussence sorear: did you see this earlier? irclog.perlgeek.de/perl6/2011-09-25#i_4477224 22:47
sorear flussence: yes 22:48
flussence I'm guessing that's just some NYI laziness?
sorear it's building a list of @_[0] through @_[Inf] 22:50
jnthn That's gonna take a while, even at Niecza's speed :) 22:51
oh noes, my beer bottle is empty :( 22:52
benabik The horror!
jnthn Thankfully the fridge can help. 22:54
sorear I find x ** -100..100 to be a very troubling pattern
jnthn sorear: Context? 22:55
(sorry if it should be obvious...)
sorear jnthn: Regexes
jnthn Oh.
:)
sorear I'm reading the S05 patches because of masak
jnthn Er, -100? :)
sorear Yes, -100
it's in S05! 22:56
benabik Apparently has been for a while.
flussence ow, I think I vaguely remember what negative numbers mean and it's not pretty... 22:57
flussence reads spec
jnthn Oh. My. Whoa.
Yeah, I didn't remember that was there :)
sorear I think it's bad for exactly the same reason p5ish @foo[-1] is bad
it creates a semantic discontinuity 22:58
niecza: say ?/ ^ x++ <[xy]> / for < y xy xxx > 23:00
p6eval niecza v9-51-g493fbf9: OUTPUT«===SORRY!===␤␤STD parses + as a quantmod but there is nothing at all in S05 to explain what it should _do_ at /tmp/ov8M5SKk2r line 1:␤------> say ?/ ^ x++⏏ <[xy]> / for < y xy xxx >␤␤Unhandled exception: Check failed␤␤ …
sorear yay fixed
jnthn sorear++ for underlying do 23:01
:D
*underlining
sorear I think that's your client's doing. 23:03
jnthn oh. :)
irssi
What did you do?
sorear prefix and suffix _
flussence dwfm :/
jnthn Oh.
I thought you'd done something curious with ANSI escape codes :P 23:04
sorear rakudo: say '_' ~ 'foo' ~ '_'
p6eval rakudo 8b7a2a: OUTPUT«_foo_␤»
flussence I think it's probably an option somewhere in irssi...
jnthn That doesn't do it.
sorear rakudo: say ' _' ~ 'foo' ~ '_ '
p6eval rakudo 8b7a2a: OUTPUT« _foo_ ␤»
jnthn That does.
flussence (I think there's an ANSI code to do it, too...)
(an irc client-compatible one, that is)
jnthn Yeah, I learned a little about ANSI codes when writing Grammar::Debugger.
I had to go and find a patched version of the Windows terminal that supported them. 23:05
23:05 molaf_ joined
jnthn So I could actually run the demo. 23:05
tadzik was Term::ANSIColor not sufficient>?
jnthn tadzik: I used Term::ANSIColor
tadzik I know, that's why I'm asking :)
jnthn tadzik: The problem is that the default terminal program Windows ships with does not support ANSI color codes
tadzik: It's not the module's fault.
tadzik ah, true
yeah, I'm aware of that
jnthn I looked around for a way to fix it but failed. 23:06
However, there *must* be one
tadzik how did you solve that? Different terminal emulator?
jnthn Yeah.
But latest version of msys git gives me the colors in the out-the-box terminal.
I dunno what they're doing.
Unless they're using various Windows-specific console APIs.
flussence the thing I don't like about terminal colour codes is that they're customisable, and there's no way for a program to figure out what palette you're using. 23:07
jnthn (msysgit folks)++ anyway...they've given me an almost completely trouble-free git experience on Windows.
23:08 static_perl left
sorear phernny: tell TimToady My current reading of S05 tells me that % is a quantmod, and since you may only have one quantmod, this means there is no shorter way to say [ :r x+ % ',' ]. Is that correct? 23:08
23:08 Trashlord joined
sorear phenny: tell TimToady My current reading of S05 tells me that % is a quantmod, and since you may only have one quantmod, this means there is no shorter way to say [ :r x+ % ',' ]. Is that correct? 23:08
phenny sorear: I'll pass that on when TimToady is around.
23:09 molaf left, bonifatio joined 23:10 aloha left 23:11 aloha joined 23:22 Chillance left 23:26 bonifatio left 23:27 espadrine left 23:30 soh_cah_toa left 23:36 soh_cah_toa joined 23:52 soh_cah_toa_ joined 23:54 soh_cah_toa 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!