»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg camelia perl6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by sorear on 25 June 2013.
colomon timotimo: en.wikipedia.org/wiki/Sierpi%C5%84ski_curve 00:00
00:02 Alina-malina joined
timotimo i was hoping i could maybe making something animated 00:05
but first i'll implement about 100000000 methods so that i can draw stuff onto gtk widgets, like the GtkDrawingArea
00:05 thou joined 00:10 dwarring left
BenGoldberg Has anyone written a grammer for C using perl6 grammars? 00:15
It would be pretty cool if you could parse the header file of some C library, and automatically generate the perl6 bindings. 00:16
00:26 virtualsue left
cognominal A grammar good enough to gather the informations necessary for an etags equivalent would be nice 00:28
00:37 thou left 00:40 hoverboard joined 00:49 daxim left 00:51 thou joined
BenGoldberg Writing it entirely in perl6 is likely to be lots of work... maybe something more like the perl5 module Convert::Binary::C, which uses an open-source C preprocessor library ... 00:51
cognominal yea, preprocessor directives don't always align with syntactic boundaries. 00:56
00:57 btyler joined
timotimo m: sub test(:$foo?) { say $foo.Bool.Int }; test; test :foo(1); test :foo<100>; 00:58
camelia rakudo-moar f1b05d: OUTPUT«0␤1␤1␤»
dalek k-simple: 3df1d25 | (Timo Paulssen)++ | lib/GTK/Simple.pm6:
GTK::Simple::Switch (gtk_switch)
 01:05
k-simple: 330b860 | (Timo Paulssen)++ | lib/GTK/Simple.pm6:
widget size request and get_allocated_width/height
k-simple: b6cb6dc | (Timo Paulssen)++ | lib/GTK/Simple.pm6:
first step towards GTK::Simple::DrawingArea
BenGoldberg Err, my point is more that it uses an already written library (ucpp), and making a perl interface, rather than writing a C grammar entirely in perl... but that's also important ;)
timotimo i'd be quite happy if people tried out the gtk-simple stuff 01:10
01:12 dayangkun joined, wingfold joined 01:13 FROGGS_ joined 01:14 wingfold_ left, Sqirrel left 01:17 FROGGS left 01:20 Su-Shee_ joined 01:23 Su-Shee left 01:25 klapperl_ joined 01:29 Sqirrel joined, klapperl left 01:31 wingfold_ joined 01:35 wingfold left 01:38 woosley joined 01:41 rindolf joined
timotimo goes down the rabbit hole of gtk 01:46
01:50 hoverboard left 01:53 hoverboard joined 01:55 zengargo1le joined
timotimo panda is still hanging in "installing" in rebootstrap.pl 01:56
"reinstalling", that is
ren1us does perl6 have any equivalent for java's "super" keyword or super() constructor call? 01:57
timotimo yeah, nextwith/nextsame/callwith/callsame 01:58
ren1us awesomeness 01:59
02:13 wingfold_ left 02:14 wingfold joined 02:18 wingfold left 02:27 SevenWolf joined 02:34 Mouq joined
dalek c/redesign: 4bcefef | Mouq++ | htmlify.p6:
Get rid of superstitious .[0]
 02:36
c/redesign: 84eabe7 | Mouq++ | template/header.html:
Move search label inside search box
c/redesign: 766740e | Mouq++ | html/css/style.css:
Get rid of space between selected menu item and sub-menu
c/redesign: 48f8964 | Mouq++ | / (3 files):
Realize that index.html isn't being generated properly

The really correct way to do it is to turn index.html into an index.pod and find some place to put it
c/redesign: 123caf2 | Mouq++ | TODO:
Add TODO file
Mouq (I was offline for a while today)
yoleaux 15 Jun 2014 11:30Z <pippo> Mouq: I found the solution for irclog.perlgeek.de/perl6/2014-06-10#i_8850807 see irclog.perlgeek.de/perl6/2014-06-15#i_8873951
Mouq \o/ pippo++ :)
02:38 thou left 02:40 virtualsue joined 02:42 thou joined
Mouq m: $¢ # -_- 02:49
camelia rakudo-moar f1b05d: OUTPUT«===SORRY!=== Error while compiling /tmp/oytTYL24nF␤Unsupported use of $¢ variable␤at /tmp/oytTYL24nF:1␤------> <BOL>⏏$¢ # -_-␤ expecting any of:␤ statement list␤ prefix or term␤ …»
Mouq std: $¢
camelia std 0f2049c: OUTPUT«ok 00:01 122m␤»
02:50 daxim joined
Mouq std: sub rand (Real $n) { $n.rand }; say rand 15 02:50
camelia std 0f2049c: OUTPUT«===SORRY!===␤Unsupported use of rand(N); in Perl 6 please use N.rand or (1..N).pick at /tmp/VwYAL8c0kw line 1:␤------> sub rand (Real $n) { $n.rand }; say rand⏏ 15␤Parse failed␤FAILED 00:01 128m␤»
03:02 daxim left 03:06 colomon left
Mouq m: =pod X<a> 03:11
camelia ( no output )
Mouq m: EVAL '=pod X<a>\n\nsay $=pod'
camelia ( no output )
Mouq m: EVAL "=pod X<a>\n\nsay $=pod"
camelia rakudo-moar f1b05d: OUTPUT«␤»
Mouq m: EVAL "=pod X<a>\n\nsay $=pod.perl"
camelia rakudo-moar f1b05d: OUTPUT«Nil␤»
Mouq hm?
m: =pod X<a>␤␤say $=pod.perl 03:12
camelia rakudo-moar f1b05d: OUTPUT«Array.new(Pod::Block::Named.new(name => "pod", config => ().hash, content => Array.new(Pod::Block::Para.new(config => ().hash, content => Array.new("", Pod::FormattingCode.new(type => "X", meta => Array.new(), config => ().hash, content => Array.new("a")),…»
Mouq That's segfaulting locally...
Wait, no, that's fine 03:13
m: =pod X<a|b,c;d,e>␤␤say $=pod.perl
camelia rakudo-moar f1b05d: OUTPUT«(signal )»
Mouq r: =pod X<a|b,c;d,e>␤␤say $=pod.perl 03:14
camelia rakudo-jvm f1b05d: OUTPUT«(timeout)»
..rakudo-moar f1b05d: OUTPUT«(signal )»
..rakudo-parrot f1b05d: OUTPUT«Array.new(Pod::Block::Named.new(name => "pod", config => ().hash, content => Array.new(Pod::Block::Para.new(config => ().hash, content => Array.new("", Pod::FormattingCode.new(type => "X", meta => Array.new(["b", "c"], ["d", "e"]), config => ().hash, con…»
Mouq parrot++
That used to work, so I'm pretty sure MoarVM's doing something wrong there 03:15
03:16 daxim joined, dayangkun left 03:20 Ben_Goldberg joined, BenGoldberg left, Ben_Goldberg is now known as BenGoldberg 03:29 dayangkun joined 03:30 thou left
ren1us if a method attempts to call another method in the same object instance, and you don't include the self. prefix, shouldn't it see that method as a candidate to call? 03:31
as in, if i do: class A = { method b = {}; method c = { b; }; }, shouldn't that be legal? 03:32
i assume that unless you provide the object instance as a prefix, it only looks for subs
03:33 kaare_ joined
ren1us woops, take out all of the = in that example 03:34
moarvm is saying b is undeclared (self.b works), and assuming that's not a bug, i would suggest that, in the absence of a matching sub, consider methods 03:39
03:47 hoverboard is now known as level_level 03:48 benabik joined 03:49 level_level is now known as hoverboard 03:52 [Sno] left 04:00 simcop2387 left, mikegrb left 04:01 simcop2387 joined, mikegrb joined 04:02 geekosaur joined 04:03 sunnavy joined 04:27 kaare_ left 04:32 thou joined 04:48 kaare_ joined 04:49 SamuraiJack joined 04:57 BenGoldberg left 05:13 hoverboard left 05:19 [Sno] joined 05:20 vendethiel left 05:39 btyler left 05:51 cognominal left, dayangkun left 05:52 cognominal joined
Mouq ren1us: If you want a shorthand by which to call "b", you can use $.b instead. If what you really want is a lexically scoped subroutine, you can just write class A { my sub b {}; method c { b; } } 05:55
Although I believe "my" is implicit for a sub
m: class A { sub b { 42 } }; say b 05:56
camelia rakudo-moar f1b05d: OUTPUT«===SORRY!=== Error while compiling /tmp/gMTNE268bS␤Undeclared routine:␤ b used at line 1␤␤»
Mouq Yeah
cognominal m: class A { sub b { 42 } }; say A.b 06:00
camelia rakudo-moar f1b05d: OUTPUT«No such method 'b' for invocant of type 'A'␤ in block at /tmp/X9IPsldoh_:1␤␤»
Mouq .tell timotimo I tried gtk-simple, but on the latest MoarVM it dies with "Attribute $!gtk_widget not declared in class GTK::Simple::Switch at lib/GTK/Simple.pm6:535", which is a bogus error... 06:12
yoleaux Mouq: I'll pass your message to timotimo.
Mouq I've been having a lot of issues with Moar lately... 06:13
cognominal I tought redefining a multi function from the setting with a function with the same signature would hide it. I can't yet find the spec part that would confirm or infirm that behavoir. 06:18
Mouq cognominal: It does, are you talking about my line with rand above? 06:19
Because rand is actually a term rather than a multi 06:20
m: say rand + 15
camelia rakudo-moar f1b05d: OUTPUT«15.3216027729761␤»
cognominal no, I am trying to have a slurp test for an URI and load the file in that case. Maybe my code is wrong
06:21 igorsutton joined
Mouq m: sub grep (|anything) { say "Local, unspecific dispatch called" }; grep 1, 2, 3 06:22
camelia rakudo-moar f1b05d: OUTPUT«Local, unspecific dispatch called␤»
Mouq m: sub grep (*@a) { say "Local, unspecific dispatch called" }; grep 1, 2, 3 06:23
camelia rakudo-moar f1b05d: OUTPUT«Local, unspecific dispatch called␤»
Mouq m: multi grep (|anything) { say "Local, unspecific dispatch called" }; grep 1, 2, 3
camelia rakudo-moar f1b05d: OUTPUT«Ambiguous call to 'grep'; these signatures all match:␤:($test, *@values)␤:(Any |anything)␤ in block at /tmp/PlUfXqv05E:1␤␤»
Mouq Ahhh
cognominal: I can't really say without looking at it. I'd be happy to help you more, but I'm heading to bed 06:24
06:24 telex left 06:25 thou left 06:26 telex joined 06:27 Mouq left 06:38 kaleem joined, kaleem left, kaleem joined
cognominal r: multi sub say(\x) { $*OUT.print("I print what I want\n") }; say('hi'); 06:38
06:39 anaeem1_ joined
camelia rakudo-parrot f1b05d: OUTPUT«Ambiguous call to 'say'; these signatures all match:␤:(Any \x)␤:(Any \x)␤ in any at gen/parrot/BOOTSTRAP.nqp:1220␤ in sub say at gen/parrot/CORE.setting:13682␤ in block at /tmp/tmpfile:1␤␤» 06:39
..rakudo-jvm f1b05d: OUTPUT«Ambiguous call to 'say'; these signatures all match:␤:(Any \x)␤:(Any \x)␤ in any at gen/jvm/BOOTSTRAP.nqp:1212␤ in block at /tmp/tmpfile:1␤␤»
..rakudo-moar f1b05d: OUTPUT«Ambiguous call to 'say'; these signatures all match:␤:(Any \x)␤:(Any \x)␤ in block at /tmp/tmpfile:1␤␤»
06:40 FROGGS_ left 06:43 Grrrr joined 06:50 dmol joined 06:53 FROGGS_ joined
cognominal S13 speaks of adding cases to a amulti, not of a definition in UNIT hiding one of the setting, both having same signature. 07:01
07:03 zakharyas joined 07:16 darutoko joined 07:18 pecastro left, salv0 joined 07:24 dayangkun joined
cognominal jnthn, should it be possible or not to hide a multi method in a setting by one in UNIT? Currently it says ambiguous call. 07:25
Timbus r: multi sub say(\x) is default { $*OUT.print("I print what I want\n") }; say('hi');
camelia rakudo-{parrot,jvm,moar} f1b05d: OUTPUT«I print what I want␤»
cognominal ho, nice 07:26
Timbus should also.. 07:27
m: multi sub say(\x) is default { $*OUT.print("I print what I want\n"); nextsame; }; say('hi');
camelia rakudo-moar f1b05d: OUTPUT«I print what I want␤I print what I want␤hi␤»
Timbus kewl
jnthn cognominal: write a proto 07:28
in the scope you want to hide the outer candidates 07:29
cognominal ok, so many way to do it, and I was unable to find any :(
Timbus++, jnthn++ 07:30
07:31 denis_boyun_ joined
cognominal is not UNIT a more derived scope than SETTING? 07:33
ok, it is. That the very point jnthn used in his answer about the proto. 07:41
07:52 dayangkun left 07:54 dayangkun joined 07:55 dayangkun left 07:56 dayangkun joined, dayangkun left 07:57 dayangkun joined 08:00 krunen joined 08:02 spider-mario joined
masak antenoon, #perl6 08:05
ren1us: subs are not methods, and introducing syntax which muddles the two together does not feel like a good idea. 08:07
ren1us: that said, what you're suggesting (fall back to method lookup when sub lookup fails) is *possible*, and could probably be attempted in a module.
sergot o/ 08:08
masak ren1us: would you expect the semantics to be "look for methods in the surrounding class", or "look for methods using ordinary method lookup (possible in classes in the inheritance chain)" ?
moritz fwiw we've had such a fallback already, once. It was a bad idea. 08:13
(or was it from method to sub?)
08:19 dayangkun left
masak hm, wasn't that mediated by some marker or other? 08:28
masak seems to recall putting a '*' before a routine name in the declaration
moritz there's 'is export' these days 08:31
which exports a method as a sub
08:35 kivutar joined
masak in practice, I never use that. 08:40
methods are methods.
FROGGS_ a contract is a contract is a contract 08:41
(but only between Ferengies)
masak that seems to be the consensus on p5 CPAN as well, where the syntactic distinction between subs and methods isn't even there.
08:42 erkan left, dayangkun joined, dakkar joined 08:43 jjk joined 08:45 Su-Shee_ is now known as Su-Shee, jjk left, jjk joined 08:46 pecastro joined 08:51 cibs left 08:52 SamuraiJack left 08:59 SevenWolf left 09:07 cibs joined 09:21 igorsutton left
FROGGS_ TimToady: should Buf have a split method? 09:21
09:22 dayangkun left 09:23 oetiker left, oetiker joined 09:24 donaldh joined 09:29 dayangkun joined 09:30 fhelmberger joined
masak lizmat: can I remove S17-concurrency-OLD.pod? if not, what's blocking on doing so? 09:30
or "what is it blockng on?", rather
dalek ecs: f04f0ab | (Elizabeth Mattijsen)++ | S17-concurrency-OLD.pod:
Remove old S17
 09:33
masak \o/
lizmat++
09:42 dayangkun left 09:46 dmol left 09:47 Psyche^_ joined 09:51 Psyche^ left 09:53 Ven joined
masak r: class Conf { has @.boxes where { all(@$_) >= 0 } }; my $c = Conf.new(:boxes[ 2, 2, -1 ]); say $c.perl 09:55
camelia rakudo-{parrot,jvm,moar} f1b05d: OUTPUT«Conf.new(boxes => Array.new(2, 2, -1))␤»
masak should this be allowed?
moritz nope 09:56
Ven m: class Conf { has @.boxes where { say @$_.perl; all(@$_) >= 0 } }; my $c = Conf.new(:boxes[ 2, 2, -1 ]); say $c.perl 09:57
camelia rakudo-moar f1b05d: OUTPUT«Conf.new(boxes => Array.new(2, 2, -1))␤»
Ven masak: It doesn't even look like the `where` is evaluated
dalek p: 2657b1e | (Elizabeth Mattijsen)++ | tools/build/MOAR_REVISION:
Bump MOAR_REVISION
Ven "make: write error Failed running make install at /Users/foo/.rakudobrew/bin/rakudobrew line 23." Oh :( 09:58
moritz "write error" looks like "disc full" to me
Ven Actually, moar-HEAD doesn't contain a `bin` directory.
ChoHag If I have three attributes in a class (%.a, %.b, %.c) how can I write a method in which one of the arguments determines which attribute to process the other arguments with? 09:59
Ven ChoHag: I'm not sure I understand
moritz I think ChoHag wants indirect lookup
ChoHag method bar { eval "say \%.$^a" } achieves it, uglily.
Ven oh. $a.'arg';
lizmat feels to me you need one @.hashes attribute, and an index
moritz since the attributes are exposed with accessors, you can use self."$name"() 10:00
masak what lizmat said.
maybe binding will help here somehow, too.
ChoHag lizmat: Maybe. I'm not too happy with the design but I want to see where it goes.
moritz m: class A { has $.a = 42; has $.b = 23; method x ($which) { say self."$which"() } }; A.new.x('b') 10:01
camelia rakudo-moar f1b05d: OUTPUT«23␤»
masak r: class C { has $.x where * > 0 }; say C.new(:x(-42)).x
camelia rakudo-{parrot,jvm,moar} f1b05d: OUTPUT«-42␤»
masak submits rakudobug
ChoHag moritz: Don't self.foo and %.foo act differently if the attribute is exposed?
moritz ChoHag: well yes, the % is also a contextualizer
ChoHag: but a % contextualierizer on a % attribute is a no-op
10:02 colomon joined
Ven moritz: not a disc full, I have 200go free. 10:02
ChoHag Doesn't self.attribute restrict write access to the attribute as a method caller would?
as it would be to a consumer of the class, I mean. 10:03
Ven `.` is the getter
(and is virtual)
Ven remembers what lizmat told him 'bout this :)
ChoHag doesn't
moritz ChoHag: yes 10:04
lizmat $.foo uses the getter method to fetch the value of the attribute, and is subclassable
moritz ChoHag: you can try to experiment with ::('$!a') and stuff like that
lizmat $!foo accesses the attribute directly
moritz ChoHag: but I don't know if object-scope indirect method notation is implemented
though in the end, you can always go through the MOP 10:05
once you have the Attribute object, you can go through bind_value to set things, iirc
(or was it set_value?)
masak niecza: class C { has $.x where * > 0 }; say C.new(:x(-42)).x 10:06
camelia niecza v24-109-g48a8de3: OUTPUT«===SORRY!===␤␤Postconstraints, and shapes on variable declarators NYI at /tmp/DuTwCq0lY5 line 1:␤------> class C { has $.x where * > 0 ⏏}; say C.new(:x(-42)).x␤␤Unhandled exception: Check failed␤␤ …»
masak at least Niecza knows it can't.
ChoHag Sorry phone. 10:07
Hmm. 10:08
moritz
.oO( some people have a smart phone, some have a sorry phone )
masak sorry phones are the phones of the future, though.
ChoHag Mine's not so smart.
masak I'd rather have a phone that's sorry than one that thinks it's smart.
moritz tried to purchase a dumb phone, bought something on ebay for 15euro, and still got something that can ran opera, capture videos, and do all that annoying stuff that drains the battery 10:09
ChoHag The problem is, at a pinch, ssh on the phone is useful. 10:10
masak moritz: surely that must be a viable niche market that someone has pounced upon? "buy this phone -- it does next to *nothing*! battery time is *great*!"
ChoHag Shop where old people shop.
You will eventually anyway. 10:11
dalek kudo/nom: 6dd2e80 | (Elizabeth Mattijsen)++ | tools/build/NQP_REVISION:
Bump NQP_REVISION
cognominal moritz, in 10 years, maybe we will have trouble to shop for a dumb watch :) 10:12
moritz masak: there's a marked for mobile phones for elderly people 10:13
masak: somehow I was convinced I'm not their target market. Maybe I'm wrong.
cognominal I want a watch that only displays time? oh, you can use a watch for that? I use it to monitor my glycemy. 10:14
ChoHag Is there a standard idiom to move an entry from one hash to another? 10:16
cognominal my father has bought a smart phone. It is shared with my mother which is hermetic to all things digital so she cannot use it.
Ven Nope, rakudobrew still writeerrors, even with sudo 10:17
ChoHag In 5: $x{$foo} = delete $y{$foo}; # I think
Ven In rakudo's install.txt, the "install only one backend" examples have no --backends= 10:18
ChoHag %x{$foo} = %y{$foo}:delete; # ? 10:19
Ven Are "short argument names" implemented in rakudo ? 10:22
cognominal Ven: I think you mean by that a degenerated use of subsignature 10:27
masak moritz: old people kinda have a point in this case 10:29
cognominal m: > sub hi( :who(:w $w)) { say $w }; hi(:w<Ven>); hi(:who<Ven>) 10:32
camelia rakudo-moar f1b05d: OUTPUT«===SORRY!=== Error while compiling /tmp/lUXjIR243V␤Preceding context expects a term, but found infix > instead␤at /tmp/lUXjIR243V:1␤------> >⏏ sub hi( :who(:w $w)) { say $w }; hi(:␤»
cognominal m: sub hi( :who(:w $w)) { say $w }; hi(:w<Ven>); hi(:who<Ven>)
camelia rakudo-moar f1b05d: OUTPUT«===SORRY!=== Error while compiling /tmp/sE_pvF5tEI␤Malformed parameter␤at /tmp/sE_pvF5tEI:1␤------> sub hi( :who(⏏:w $w)) { say $w }; hi(:w<Ven>); hi(:wh␤ expecting any of:␤ statement list␤ …»
10:42 Pleiades` left 10:45 jnthn joined 10:47 Pleiades` joined 10:49 rindolf left
Ven Why do I get a write error on `make`, srsly >_> 11:00
Oh, it actually works. 11:01
tadzik ahahaha, DSON 11:02
yoleaux 14 Jun 2014 05:55Z <thou> tadzik: Do you have any feedback on github.com/tadzik/panda/pull/80 ?
tadzik xfix++
oh!
xfix .oO( nice, I get extra karma for absolutely nothing ) 11:03
Ven xfix: for dson ;)
in META.info, version has to be x.y.z or * ?
dalek nda: da6558d | (Tim Smith)++ | / (2 files):
Define make-default-ecosystem in Panda::App

Enable Panda::App to be used by other tools to set up the same ecosystem that bin/panda uses.		 11:05
nda: 035ffc6 | (Tim Smith)++ | / (3 files):
Rename '.work' to '.panda-work'
nda: 2714f9f | tadzik++ | / (4 files):
Merge pull request #80 from softmoth/master

Factor out make-default-ecosystem() for external scripts
tadzik Ven: I don't think we ever specified that 11:06
11:06 Pleiades` left
Ven tadzik: mmh. I'd like to know to make a xyz tool 11:06
tadzik sort of expected github.com/xfix/Acme-DSON/blob/master/META.info to be in DSON
xfix tadzik: wouldn't work.
Otherwise, I would do that.
tadzik I know it can't be, but that'd be hilarous nonetheless 11:07
especially after seeing the README
Ven It could be, then compiled ;) 11:09
11:11 Pleiades` joined 11:13 dmol joined
xfix tadzik: I originally planned README.md, like I do in my other projects, but I decided to make plain text file that is DSON itself, because the module itself is a joke anyway. 11:13
Ven tadzik: github.com/xfix/Acme-DSON/pull/1 11:16
xfix Ven: not valid DSON.
Object literals end with 'wow'.
Ven xfix: fixed already 11:17
xfix Merged.
Also, it seems that DSON removed 'next'. 11:19
I'll update DSON now.
Ven xfix++ :D 11:20
xfix Done. 11:22
11:26 wingfold joined 11:27 brrt joined
masak how do I mark the last statement in a routine, in the case it is a statement-modifier for loop, as being the return value? 11:29
11:29 wingfold left
masak put it in parens? 11:29
masak consults S04 on this
Ven masak: statement-modifier for loop ?
masak Ven: ... if ... for ...; 11:30
oh, parens is what doctor S04 orders.
or the 'lazy' keyword.
I think I prefer the parens in this case; a bit less obtrusive. 11:31
masak is writing a blog post
11:32 kurahaupo joined
Ven m: say foo; sub foo { (1 if True for ^5); } 11:33
camelia rakudo-moar f1b05d: OUTPUT«1 1 1 1 1␤»
Ven
.oO( That's a pretty bad way to write 1 x 5 )
 11:34
masak m: say foo; sub foo { 1 if True for ^5 } 11:35
camelia rakudo-moar f1b05d: OUTPUT«Nil␤»
masak that's what I ran into.
I'm writing fairly FP-ish code, fwiw.
so I guess I had the expectation of implicit return of the (lazy) list.
tadzik we need dsonlint.com 11:36
Ven masak: would be inconsistent with for vs do for 11:39
masak: FWIW, ruby gets away with it because it doesn't really have a `for` construct
(I mean, gets away with implicit returns everywhere and performance) 11:40
CoffeeScript doesn't get away with it and plenty of time we see people add `null` at the end of their functions not to have a `for` loop at the end of the function
masak that's the situation Perl 6 used to be in before TimToady++ changed the spec. 11:41
but (especially after what just happened) I realize that this is a damned-if-you-do-damned-if-you-don't kind of design decision.
11:43 SamuraiJack joined
Ven Most definitely. 11:46
As I said -- Ruby niftily (?) gets away with it because Array#each returns itself.
(In LS, we allow you to prefix your function sigil with ! to remove the implicit return) 11:47
11:49 igorsutton joined 11:50 dmdr joined 11:51 spider-mario_ joined, spider-mario left 11:52 spider-mario_ left 11:55 spider-mario joined 11:56 colomon left
masak just wanted to point out that the REPL currently dies after ~50 evaluations, with the error "Spesh: failed to fix up handlers (-1, 110, 110)" 12:03
lizmat better than ~13 :-) 12:04
Ven "Unhandled exception: failed to load library 'dynext/libperl6_ops_moar.dylib'" :'(
lizmat wonders whether "is cached" would make sense on a class 12:05
Ven lizmat: on a BUILD, maybe
Anytime I try to access to perl6 outside of my rakudo directory, it blows up
lizmat well, but then you would have to have a BUILD or new or bless
Ven: gist ?
Ven lizmat: gist of ? 12:06
lizmat the blowup?
Ven lizmat: "Unhandled exception: failed to load library 'dynext/libperl6_ops_moar.dylib' at <unknown>:1 (/Users/notliz/os/rakudo/perl6.moarvm::303)"
12:07 psch joined
psch hi #perl6 12:07
lizmat psch o/
Ven: sorry, doesn't ring a bell :-(
Ven I already had that last time I built rakudo on a debian.
Do I need to have the .dylib in my path or something ?
psch i found a segfault!
look at RT#69492 12:08
synopsebot Link: rt.perl.org/rt3//Public/Bug/Displa...l?id=69492
psch *looking
trying to get .name for a anon named segfaults moar
+parameter
masak yes, definitely better than ~13. 12:09
psch additionally, i think the test for the ticket isn't quite sufficient, i.e. even sub foo {} has at least one ':'; but i don't think testing for ':($' instead is much better
Ven masak: tried both 13 and 50 here, doesn't fail running
Also tried on my windows last time, definitely doesn't blow up at 13.
Here is moar, my windows has parrot.
lizmat Ven: fwiw, I have some symlinks to perl6 executables in my path, maybe that makes a difference ? 12:10
Ven lizmat: Absolutely no clue ..
I'll try and rebuild ... 12:11
nope, no luck. 12:15
my perl is 5.16, so that didn't blew up. 12:17
lizmat after the Configure, the version of perl5 should really not matter
Ven I'll reconfigure and remake.
12:18 bebus joined
FROGGS_ masak: "Spesh: failed to fix up handlers (-1, 110, 110)" this was fixed a week ago 12:18
Ven FROGGS_: hey -- is your windows build process documented somewhere ?
FROGGS_ hmmm, maybe it got reintroduced...
Ven: build what? I usually do not build windows itself :o) 12:19
12:19 bebus left
Ven FROGGS_: rakudo* 12:19
FROGGS_ ahh, star
Ven FROGGS_: wait, you mean you're not bill gates ?
FROGGS_ *g*
I have a 10 line walkthrough of the build process, hold on 12:20
Ven Nope, reconfigured, remaked. remakeinstalled. still won't work.
What am I to do if I can't p6 while I'm supposed to work ?!
FROGGS_ coffee? 12:21
Ven FROGGS_: wait, you mean you're not the coffee guy either ?
FROGGS_ and office sleep should be very healthy also
I am a coffee guy
Ven FROGGS_: wait, you mean you're not the pillow guy either ?
FROGGS_ :P
every morning it fells like I should consult my pillow more often 12:22
Ven: gist.github.com/FROGGS/b9b848250062860e61de
Ven FROGGS_: WHIPE rakudo ? Boi, what did this kid do to you ? :P 12:23
FROGGS_ Ven: rename it to 'rakudo_' :o)
Ven FROGGS_: (I was just jokign with WHIPE/WHIP) 12:24
FROGGS_ this is for rakudo@parrot star btw, in case you want to build for more just adjust step 6, and forget about step 7 and 8
hmmm, what word was I thinking of?
Ven FROGGS_: wipe ? 12:25
FROGGS_ ahh, yes! :D
masak FROGGS_: seems I have a slightly ould Rakudo built. I'll rebuild and see if the problem goes away. thank you. 12:27
Ven Don't tell me nobody ever got a dylib missing with moar ? The two times I built it, I got it :o) 12:28
FROGGS_ mine is also old, so I cannot test right now
(too many stuff I fiddle right now)
Ven: I've never seen such an error on my linux or windows box :P 12:29
so, perhaps these OSes are superior?
Ven FROGGS_: I got it on linux and osx 12:30
I'll try building jvm too. 12:31
masak also, if it were up to me, we would hide the output from declaring a sub in the REPL. 12:32
that makes no-one happy.
Ven m: sub f {} 12:33
camelia ( no output )
Ven m: sub f { say 1 }
camelia ( no output )
12:33 sqirrel_ joined
Ven masak: ? 12:33
[Coke] yawns. 12:34
masak Ven: in the REPL. 12:36
12:36 Vlavv left
masak Ven: not using camelia. 12:36
Ven will check his REPL when he gets one
Ven fixed his perl 6. 12:38
masak also, while I'm proposing things: I suggest we abolish .gist, and instead make .Str more .gist-like. 12:39
the (IMO useless) distinction between concatenating values (which uses .Str) and &say-ing values (which uses .gist) is very distracting, and leads to bugs, which leads to sadness. 12:40
lizmat $ ack 'method gist' src/core | wc -l
43
12:41 dayangkun joined
Ven Now to get panda. 12:42
12:42 guru joined, guru is now known as Guest64859
[Coke] masak: +1 from me on eliminating .gist 12:42
12:42 Guest64859 is now known as ajr_
[Coke] Good luck getting TimToady to agree, however. :) 12:43
lizmat fwiw, not many gist methods have identical .Str methods 12:44
FROGGS_ hmmmm, but .gist and .Str and a Match are very different
Ven panda has no versioning ? 12:46
FROGGS_ Ven: what do you mean exactly?
Ven oh, yeah, that story about removing META.info and just using :<ver> ?
masak maybe some .gist methods have non-identical .Str methods because one of them sucks. 12:47
note (again) that Python has *one* way to stringify things. essentially .perl, but nicer.
FROGGS_ Ven: we will always have META.info (META6.info soonish) for distributions
masak Perl 6 has *three*. (.Str .gist .perl)
moritz sometimes the .Str method sucks, because it's very often p5 heritage
masak moritz: which we have partially gotten rid of.
see boolean stringification, for example.
12:48 Vlavv joined
masak used to be 0 and 1. no-one seems to mind that it's now False and True :) 12:48
FROGGS_ masak: propose something for Match
masak FROGGS_: let me think.
m: "foo" ~~ / f (o)* /; say $/.Str; say $/.gist; say $/.perl 12:49
camelia rakudo-moar 6dd2e8: OUTPUT«foo␤｢foo｣␤ 0 => ｢o｣␤ 0 => ｢o｣␤␤Match.new(orig => "foo", from => 0, to => 3, ast => Any, list => ([Match.new(orig => "foo", from => 1, to => 2, ast => Any, list => ().list, hash => EnumMap.new()), Match.new(orig => "foo", from => 2, to…»
FROGGS_ hmmm, perhaps I could live with .Str and .perl for Match
betterworld masak: python has at least str() and repr() :)
masak betterworld: oh, troo
moritz IMHO it would be fine if Match.Str was used on default output, and the .gist still be available
masak betterworld: but they are often equal, at least for the built-ins.
moritz for manual calling
FROGGS_ Match.perl is not very readable though, and Match.gist is the opposite 12:50
m: say Int.Str; say Int.gist; say Int.perl
camelia rakudo-moar 6dd2e8: OUTPUT«use of uninitialized value of type Int in string context␤␤(Int)␤Int␤» 12:51
FROGGS_ I want to keep all three :/
I would be fine if say would use .Str instead of .gist though
12:53 dmdr left
moritz note that 'Int' is available more idiomatically as .^name 12:59
Ven `$version ~~ / $<major>=\d '.' $<minor>=\d /; my %h = ($_ => $/{$_}.Int for $/.keys);` isn't there a more elegant way of doing this ? 13:01
13:02 erkan joined
lizmat make it a Version object? 13:02
FROGGS_ moritz: yes, but when I do 'say $something.^name', I don't see the value if it isnt a type object
Ven That seems really overkill
I don't have any checking with Version, either 13:04
lizmat you were asking more "elegant", that could be overkill :-) 13:06
awwaiid if version is so fundamental, maybe we should have $something.^version :) 13:07
Ven and I can't modify Version.parts.
lizmat hmmm... I'm not sure how TimToady would feel about making it mutable 13:10
13:10 molaf joined
Ven I just resorted to split. 13:11
13:11 jnap joined
moritz FROGGS_: well, .^name is the shortest way to get the type name, even if it's an instance 13:12
FROGGS_ moritz: I know
but want to know from $something is what is inside, so .gist gives me a value if it has one or the typename 13:13
that is, I don't want the type name if it has a value
moritz well, it really depends on the use case 13:14
there are lots of situations when I'm interested in the type
like, when a dispatch failed
masak r: say [2, 0] eqv (2, 0) 13:21
FROGGS_ False
masak this just surprised me. expected True.
camelia rakudo-jvm 6dd2e8: OUTPUT«(timeout)»
..rakudo-{parrot,moar} 6dd2e8: OUTPUT«False␤»
FROGGS_ I guess it is because the LHS is an item
Ven github.com/Nami-Doc/xyz.pm6 It works \o/ 13:22
./xyz --patch will update your file, commit the change, tag the new version and push the tag
FROGGS_ such magic!
:D
13:23 flaviusb joined
lizmat shouldn't that be META6.json ? 13:25
as the default I mean?
Ven Oh, I broke my perl 6 again. 13:28
dalek kudo/nom: 4dd2cc4 | (Elizabeth Mattijsen)++ | tools/build/Makefile-JVM.in:
Make sure S17-supply/watch_path.t actually compiles
Ven fixed it again 13:29
Not sure how I can go with both a .pl and one with a shebang, though :(
13:29 sqirrel_ left
Ven perl6 is cool for shell scripts, it is. 13:30
But I should be working at
lizmat [Coke]: I'm seeing S04-phasers/in-loop.t fail reliably on jvm now: gist.github.com/lizmat/2bbcb4f8799f689da1ff 13:31
is this something you've seen before? 13:32
timotimo masak: fwiw, python has two ways to stringify things; str and repr
yoleaux 06:12Z <Mouq> timotimo: I tried gtk-simple, but on the latest MoarVM it dies with "Attribute $!gtk_widget not declared in class GTK::Simple::Switch at lib/GTK/Simple.pm6:535", which is a bogus error...
donaldh lizmat: I've seen that for a while now.
timotimo oh, someone else told you that already
lizmat ah, ok... it;s just that since a long time, I wasn't able to complete the spectest on jvm 13:33
13:33 FROGGS_ left
Ven Oh, project euler is down 13:33
timotimo oh, interesting error
donaldh lizmat: the in-loop tests are all marked todo but exit ungracefully on JVM so a better fudge might be required. 13:34
13:35 noganex joined
dalek k-simple: 5b28644 | (Timo Paulssen)++ | lib/GTK/Simple.pm6:
fix GTK::Simple::Switch, hopefully
 13:35
13:36 bluescreen10 joined 13:42 FROGGS_ joined
FROGGS_ lizmat: you are right of course # META6.json 13:43
13:44 psch left
masak Ven: did you know that you can indent the heredocs in Perl 6, so that they follow the indent of the rest of the code? 13:46
Ven masak: yes
13:46 psch joined
masak timotimo: yes, betterworld++ pointed that out. that would be like us having .Str and .perl, I guess. 13:47
13:47 kaleem left
masak Ven: did you also know that a usage message is automatically generated for you? (and there are probably ways to trigger it in your xyz.pl program) 13:48
Ven masak: --help can't be generated with %*
masak you might be better off overriding USAGE, in any case.
Ven %* catches --help
so no, I can't 13:49
FROGGS_ the normal way would be to add a MAIN(:$help!), MAIN(:$minor!), MAIN(:$major!) and MAIN(:$patch!), 13:51
masak well, this one seems to be able to do combinations. is that a feature? 13:52
13:57 anaeem1_ left 13:58 btyler joined
masak m: my $s = set; $s (|)= 42; say $s 14:05
camelia rakudo-moar 6dd2e8: OUTPUT«===SORRY!=== Error while compiling /tmp/BfwH5vp3JD␤Two terms in a row␤at /tmp/BfwH5vp3JD:1␤------> my $s = set; $s ⏏(|)= 42; say $s␤ expecting any of:␤ postfix␤ infix stopper␤ inf…»
masak m: my $s = set; $s = $s (|) 42; say $s
camelia rakudo-moar 6dd2e8: OUTPUT«set(42)␤»
14:05 oetiker left
masak something is wrong with the assignment variants of the set operators. 14:05
14:05 oetiker joined
masak m: my $s = set; $s ∪= 42; say $s 14:06
camelia rakudo-moar 6dd2e8: OUTPUT«===SORRY!=== Error while compiling /tmp/eDMlNVCBBN␤Two terms in a row␤at /tmp/eDMlNVCBBN:1␤------> my $s = set; $s ⏏∪= 42; say $s␤ expecting any of:␤ postfix␤ infix stopper␤ inf…»
masak m: my $s = set; $s = $s ∪ 42; say $s
camelia rakudo-moar 6dd2e8: OUTPUT«set(42)␤»
masak submits rakudobug
14:08 thou joined, pdcawley joined
lizmat could it be some internal attribute of the set operators is not correct ? 14:09
masak no idea.
so, FIRST doesn't run if there are no iterations. 14:16
jnthn And nor does LAST any more.
masak which phaser do I use if I want to run even when there are no iterations?
moritz INIT?
masak (I want to do an unconditional once-only take in a 'gather for') 14:17
moritz: trying.
m: say (gather for 1..3 { take $_ })
camelia rakudo-moar 6dd2e8: OUTPUT«1 2 3␤»
masak m: say (gather for 1..3 { INIT take "OH HAI"; take $_ }) 14:18
camelia ( no output )
masak o.O
I...
masak submits rakudobug
r: say (gather for 1..3 { INIT take "OH HAI"; take $_ })
camelia ( no output )
..rakudo-parrot 6dd2e8: OUTPUT«take without gather␤ in block at gen/parrot/CORE.setting:533␤ in block at gen/parrot/CORE.setting:569␤ in block at /tmp/tmpfile:1␤␤»
masak oh.
Parrot may actually have a point.
moritz masak: maybe you want gather { take "bla"; for 1..3 { ... } } 14:19
masak moritz: yes, that'd work.
moritz: I was just wondering if there is a phaser that does this.
moritz masak: on 2015-04-01 I'll propose a NORMAL phaser that runs as part of the normal code execution path
masak: or maybe I'll call it BORING 14:20
lizmat r: unless do for (1) { FIRST say "first" } { say "none" }
camelia rakudo-{parrot,jvm,moar} 6dd2e8: OUTPUT«first␤none␤»
lizmat surprised to see the "non" in this case
r: unless do for () { FIRST say "first" } { say "none" }
camelia rakudo-{parrot,jvm,moar} 6dd2e8: OUTPUT«none␤»
lizmat actually, I think a NONE phaser would make sense here 14:21
for () { NONE say "none" }
masak moritz: :)
moritz masak: and the companion phaser BORAT fires mostly at random, and must be written with an additional trailing { 14:23
[Coke] lizmat: Trying to find a git grep that will tell me the answer to your question, but sounds like you already found it. 14:26
donaldh: any more suggestions on preventing the jvm daily roast from failing using the eval server?
masak moritz: does it still adhere to the '}'-at-end-of-line-ends-statement rule?
14:27 colomon joined 14:28 kaare_ left
moritz masak: well no, otherwise the trailing { would be a statement on its own 14:29
[Coke] donaldh: have you any idea how to bundle perl6 in a jar we could invoke from eclipse? 14:31
donaldh [Coke]: no, I tried to reproduce your problem, but without any success so far. I may need to create a VM with limited memory to force the same behaviour. 14:32
[Coke] donaldh: we can probably get you a login on the roast box to play with.
have you tried using the perl6-roast-data runners to duplicate the issue? 14:33
donaldh [Coke]: Bundling perl6 in a jar would be a fun project. Eclipse lets you deploy plugins that expand on the filesystem so would support the current module loading. A more sophisticated approach would involve module loading using a ClassLoader 14:35
I could have a go at embedding perl6 for eclipse. Do you have a use case in mind? 14:36
[Coke] ... and I just realized I don't need the jar bundled in eclipse, I could do what I wanted as a standalone thing to start. 14:37
donaldh: I do coldfusion development for a living. All the editors suck in various ways. I want to make a pretty printer/linter using a perl6 grammar.
so then folks can use whatever they want to edit, but we can all agree on a standard formatting. (by which I mean, I agree, and they conform. :)
I could use something like ANTLR, but perl6 seems much nicer. 14:38
14:38 amirite left
donaldh yeah, I agree 14:38
14:38 [Sno] left 14:39 fhelmberger left
[Coke] I have some crappy tests now written in coldfusion, but they can't handle things like embedded HTML/JS/CSS/SQL 14:39
(which grammars would also be good for.)
donaldh This gist was the result of me running the perl6-roast-data runner: gist.github.com/donaldh/57020f86459600eedaa5
[Coke] ok, so you're not getting any unexpected failures. 14:40
donaldh My next step is to repeat the perl6-roast-data run inside a memory limited vm.
donaldh was thinking of embedding perl6 inside the perl epic editor. 14:41
colomon perlcabal.org down?
lizmat works for me
however, it is bad need of an update 14:42
colomon huh
14:42 treehug88 joined
colomon looking at github.com/perl6/specs/blob/master...ary/IO.pod at the moment, and it seems pretty okay (and I suspect completely up to date). 14:42
lizmat colomon: hmmm...perlcabal.org gives the pugscode.org website 14:43
some VirtualHost mixup ?
[Coke] here's more info than I've gotten before: gist.github.com/coke/27f00d0bed4929b8ec1b
^^ donaldh 14:44
donaldh [Coke]: I spent a bit of time over the weekend trying to figure out why the JVM retains so much on the heap during a spectest run. Unfortunately the profiler crashes when I try to use it so I'm having to use blunter tools.
colomon lizmat: you're right. perlcabal.org is coming out Pugs, and perlcabal.org/syn/ (which is always where I go) is… err, it's back now. :) 14:47
also seems to be completely up-to-date?
moritz s/Pugs/Mu 14:54
/
donaldh [Coke]: Yeah, you got a hotspot VM dump that basically says a native alloc failed. 14:58
lizmat FROGGS_: is there any reason why CompUnitRepo::Distribution has a .Hash instead of a .hash ? 14:59
FROGGS_ lizmat: hmmm, no 15:00
15:00 erdic joined
[Coke] donaldh: was hoping there was something more interesting in there than that. ah well. :) 15:02
masak is having a lot of fun solving a mathematical problem in Perl 6 today
15:04 anaeem1 joined
Ven masak: project euler is down tho :( 15:05
15:06 pecastro left
masak found this problem on Twitter. 15:07
15:07 anaeem1 left
masak it's been sloshing around in my brain for two days. 15:07
15:07 anaeem1 joined 15:08 anaeem1 left 15:09 denis_boyun_ left
dalek p/pmc2c_orig: 1aa8cd9 | rurban++ | src/vm/parrot/pmc/ (7 files):
pmc2c_orig: optimize 6model pmcs for the new pmc2c

Annotate several PMC methods for unneeded or manual GC write barriers. This requires the new pmc2c from parrot 6.5.0
This is 2-4% faster for the nqp testsuite.
  $ perf stat -r2 prove -r --exec "./nqp-p --vmlibs=nqp_dyncall_ops" t/nqp t/hll t/qregex t/p5regex t/qast t/serialization		 15:13
lizmat rurban++ for not giving up on rakudo on parrot 15:14
[Coke] rurban++ # agreed. 15:19
masak rurban++ # the more backends, the better 15:21
lizmat FROGGS_: I'm considering renaming CompUnitRepo::Distribution to just Distribution 15:24
as the Distribution itself does not necessarily have anything to do with CompUnitRepo's just yet 15:25
and we might have Distributions without any installed CompUnit's, just consisting of data files et all
FROGGS_ (or anybody else): thoughts? 15:26
FROGGS_ +1 15:27
15:31 Ven left 15:32 igorsutton left 15:34 rindolf joined
masak +1 15:35
15:39 zakharyas left 15:40 brazuca joined
masak what's the current way to define something to intercept MyType($a, $b, $c) ? 15:41
lizmat .wrap ? 15:42
masak never mind, I'll just use a sub with a lowercase letter instead. 15:43
lizmat or do you mean PARAMETERIZE_TYPE ?
masak yes, that seems to be what I wanted. lizmat++ 15:47
but I think my current approach is nicer.
lizmat PARAMETERIZE_TYPE is not specced, afaik, and a temp hack at best (was my impression) 15:48
15:51 stux joined 15:52 xinming joined
masak m: class C { method Str { "overridden" } }; say C.new.gist 15:53
camelia rakudo-moar 4dd2cc: OUTPUT«C.new()␤»
15:53 dmol left, kaare_ joined
masak m: class C { method gist { "overridden" } }; say C.new.Str 15:53
camelia rakudo-moar 4dd2cc: OUTPUT«C<140282296048608>␤»
15:55 xinming_ left
dalek ecs: 480e2ef | (Elizabeth Mattijsen)++ | S22-package-format.pod:
Elaborate some more on S22:

Describe:
  - CompUnitRepo specification (e.g. "inst:name<work>:/installed_modules")
  - Distribution class
  - %*CUSTOM_LIB
  - %*INC
and other various tweaks, fixes and elaborations		 16:00
lizmat masak: you're getting the .Str.gist
timotimo what am i doing wrong if my blog gets most of its hits on each monday before i even posted the weekly post? >_> 16:02
lizmat not posting early enough?
actually, I think this is because it's mentioned in Gabor's Perl Weekly mail
16:03 brrt left
timotimo heh :) 16:03
could be 16:04
lizmat under Weekly collections 16:05
Beginning of inlining, some GSoC work materializes, DSON encoder/decoder
timotimo ah, that's last week's
masak yes, it's got to be because of szabgab++'s newsletter. 16:10
segomos_ or people are trying to get back into the swing of work 16:14
and not working
16:15 denis_boyun_ joined
donaldh timotimo: You said you'd post every Monday. I check during my day but it's usually not posted. 16:15
timotimo right 16:17
16:17 brazuca left
betterworld once the blog post is there, people stop refreshing their browsers 16:17
timotimo i should write "monday evening"
donaldh what TZ are you in? 16:23
timotimo GMT+1 i believe; CEST 16:24
16:24 molaf_ joined 16:25 brrt joined
donaldh Oh okay, evening must stretch pretty late for you ;-) I am in GMT and don't usually see your post until Tuesday. 16:26
donaldh makes mental note to check for post on Tuesdays
16:26 denis_boyun___ joined 16:27 molaf left
donaldh is running rakudo.jvm.sh on a VM with 3GB RAM for [Coke]++ 16:27
16:27 denis_boyun_ left
timotimo p6weekly.wordpress.com/?p=180&s...f1b2a1ed81 - feedback much appreciated 16:28
especially mentions of things i've forgotten
16:29 erkan left 16:33 denis_boyun___ left
FROGGS_ ohh, inline got merged? 16:33
timotimo aye 16:35
and the revisions got bumped, too
16:35 brrt left
btyler "I'ven't", while logical, would probably get you a slap on the wrist from a grammar teacher :) 16:36
timotimo i'dn't've decided against it, even if you told me that.
btyler any news from sergot++'s gsoc work? 16:39
16:39 jnap left
timotimo no new blog post yet 16:40
[Coke] donaldh++
16:42 dmol joined, hoverboard joined, ajr_ left 16:43 guru joined, guru is now known as Guest62139 16:44 Guest62139 is now known as ajr_, Rotwang joined 16:45 SamuraiJack left
lizmat timotimo: maybe explain a little more about what is so special about the JIT? 16:53
that it changes the "return 12 - 6" to machine code (is the way I understand it)
timotimo: maybe add "of the past years" to the mention of dwarring++ 's work ? 16:54
timotimo can do, thanks 16:55
16:55 amirite joined 16:56 dakkar left 16:57 donaldh left
rurban_ timotimo: you could add that the new parrot release 6.5.0 includes Chirag's work already. "Released with 6.5.0" 16:59
and all roast spectests pass now (again) on parrot 17:00
timotimo ah, great!
tadzik timotimo:
oh, I was going to mention I'ven't too
timotimo why the hate for i'ven't? 17:01
rurban_ parrot added also namealiases for control chars, which went away with ICU 5.2
lizmat rurban_: does that mean we need to bump nqp's PARROT_REVISION ?
tadzik it looks like something between Elvish and Cthulu fhtagn
however you write the latter :)
timotimo rurban_: we do 100% of the spectests on parrot with 6.5.0?
rurban_ not yet. not yet released officially. tomorrow
I pushed a roast pull request to FROGGS, with that it passes. 17:02
lizmat ok, will test after release then :-)
timotimo oke! :)
17:02 SamuraiJack joined 17:03 jnap joined
rurban_ I will remove the namealias skips on the spectests also. Currently disabled because of ICU madness 17:03
17:04 flaviusb left, brrt joined, brrt left
timotimo going to publish now 17:04
lizmat timotimo++ 17:05
dalek kudo/nom: fc9dcee | (Elizabeth Mattijsen)++ | src/core/CompUnitRepo/Local (3 files):
Introduce CompUnitRepo.short-id
 17:07
17:07 ajr_ left
timotimo FROGGS_: it'd be pretty excellent if there'd be a guide/example/blogpost about the thing where scripts installed via panda (or something) can differentiate between different installed versions and stuff 17:08
you know what i mean, right?
even though i fail at expressing it %)
dalek kudo-star-daily: 9ff98d1 | coke++ | log/ (13 files):
today (automated commit)
 17:10
kudo-star-daily: 30f02b2 | coke++ | log/ (14 files):
today (automated commit)
rl6-roast-data: 9a69a12 | coke++ | / (6 files):
today (automated commit)
rl6-roast-data: da46d70 | coke++ | / (6 files):
today (automated commit)
kudo/nom: 7e66d3e | (Elizabeth Mattijsen)++ | / (5 files):
Part 1 CUR::Distribution -> Distribution migration
 17:14
17:14 SamuraiJack left
ChoHag How can I create a class (or role?) which calls a method when an instance of the class is used in a string/numeric/etc (the Cool types) context? 17:15
timotimo well, at least .Str can just be used as a method 17:16
otherwise there's .Numeric, .Real, .Num, .Int and .Bool
lizmat method Numeric for + context
method Bool for ? context
timotimo ah, numeric is the right one?
lizmat m: say $*USER; say +$*USER; say ~$*USER 17:17
camelia rakudo-moar 4dd2cc: OUTPUT«p6eval (1005)␤1005␤p6eval␤»
lizmat Numeric is the one I use in src/core/Process.pm
17:18 denis_boyun_ joined
psch m: for 1,2,3, { say 3 } # RT#66776; this throws Comp::Group - i've replaced the one Comp::AdHoc with a narrower type (which sits in a PR now) but i'm wondering if it should just not throw Comp::Group? 17:21
synopsebot Link: rt.perl.org/rt3//Public/Bug/Displa...l?id=66776
camelia rakudo-moar 4dd2cc: OUTPUT«===SORRY!===␤Expression needs parens to avoid taking the block␤at /tmp/dpdNaCFi7L:1␤------> for 1,2,3, { say 3 }⏏ # RT#66776; this throws Comp::Group - i␤Missing block (apparently taken by expression)␤at /tm…»
synopsebot Link: rt.perl.org/rt3//Public/Bug/Displa...l?id=66776
ChoHag So just provide those few functions which call (variations on) my foo method?
psch m: class A { method Numeric { 5 }; method Str { "five" }; method Bool { False } }; my A $a .= new; say +$a; say ~$a; say so $a # practical example for ChoHag
camelia rakudo-moar 4dd2cc: OUTPUT«5␤five␤False␤»
ChoHag thankyou
dalek kudo/nom: 6ca805d | (Elizabeth Mattijsen)++ | src/core/Distribution.pm:
Part 2 CUR::Distribution -> Distribution migration
 17:22
17:28 jnap left, jnap joined
psch i think it does need the Comp::Group for the missing block case at least, if only for the cursor position for pointing to where the block is missing and which expression took it... 17:31
17:32 denis_boyun_ left
lizmat ah, ok 17:32
merging
dalek kudo/nom: ec27142 | (Pepe Schwarz)++ | src/ (2 files):
Added X::Syntax::BlockTaken and use it from the Grammar.
kudo/nom: 09134a7 | lizmat++ | src/ (2 files):
Merge pull request #280 from peschwa/nom

Added X::Syntax::BlockTaken and use it from the Grammar.
lizmat psch++
psch lizmat: i had $.what as 'Expression' and no ternary, but that means the caller has to hand in "Function" for the other case 17:33
lizmat yeah, got that now :-)
I had misread :-)
psch alright, thanks for merging
i'll push the tests
dalek kudo/nom: aaa82dd | (Elizabeth Mattijsen)++ | src/core/IO.pm:
No need for two calls to .print
 17:34
17:37 [Sno] joined 17:44 denis_boyun joined 17:47 carlin joined
dalek ast: d6b8fac | (Pepe Schwarz)++ | S32-exceptions/misc.t:
Added tests for PR 260/RT#66776
 17:49
synopsebot Link: rt.perl.org/rt3//Public/Bug/Displa...l?id=66776
17:59 denis_boyun left 18:02 vendethiel joined
lizmat cycling& 18:04
dalek kudo/nom: 6e0ac15 | (Elizabeth Mattijsen)++ | src/core/IO.pm:
Don't bother calling .gist for Strings
 18:05
18:18 hoverboard is now known as moistcherry, rindolf left 18:22 rindolf joined 18:31 Mouq joined
psch oh shucks, i didn't even mention the right PR in my roast commit :( 18:38
that aside, anyone who can tell me something about this? gist.github.com/peschwa/863b4d839e0024388052
in case it matters, i'm running a pae enabled 32bit debian vm 18:39
xfix It's a bug.
Perl 6 should never SIGSEGV, unless you call C functions.
[Coke] p6: my $f = 5; $f.isa(Scalar)
camelia rakudo-jvm 6e0ac1: OUTPUT«(timeout)» 18:40
( no output )
[Coke] seems fine here. what version of perl6 are you running?
18:40 moistcherry is now known as hovercraft, colomon left
psch This is perl6 version 2014.05-197-gba608c6 built on MoarVM version 2014.05-123-g87221ba 18:40
fwict it only happens with passing Scalar to .isa 18:41
Scalar.^mro dies as well i just found out 18:44
18:44 labster joined, denis_boyun_ joined
psch .^name works; not sure what else i should/could try 18:47
[Coke] psch: please try updating your copy of moar. it's over a month old. :) 18:49
psch huh, that's curious; i ran Configure.pl after make realclean 18:50
18:50 colomon joined 18:51 dwarring joined
flussence are you doing the "use rakudo's Configure.pl to build everything" method? 18:51
psch flussence: yes, is there a better one?
flussence that's fine, but you might need to also `make clean` in the nqp/moar subdirs separately.
psch oh, i didn't; i'll do that 18:52
thanks
cognominal is there a way to use HLL::Grammar from rakudo? I have asked the question a few month ago and it was no. But things may have move since then. 19:00
FROGGS_ m: use NQPHLL:from<NQP> 19:04
camelia ( no output )
FROGGS_ cognominal: but this is a dangerous path...
cognominal: what do you want to achieve?
19:04 dmol1 joined 19:05 dmol left
cognominal FROGGS_, I want to use the EXPR rule 19:05
FROGGS_ hmmmm, I've got a Perl 6 version here...
19:05 denis_boyun_ left
cognominal May be I need to translated HLL::Grammar from nqp to rakudo 19:05
ho, that would be nice 19:06
19:06 rindolf left
FROGGS_ github.com/rakudo-p5/v5/blob/nqp_t...ar.pm#L632 19:06
19:07 denis_boyun joined 19:08 hovercraft is now known as moistcherry 19:11 itz joined
dalek c/redesign: 3d00cff | Mouq++ | / (2 files):
Detect multiple subkinds on routines + note if a subkind can't be found

This is useful for those routines defined as both a sub and method, of which Perl 6 has many		 19:11
19:19 darutoko left 19:23 [particle] left 19:24 [particle] joined 19:26 Psyche^_ left, Psyche^ joined 19:30 treehug8_ joined
dalek c/redesign: f019036 | Mouq++ | / (3 files):
Update handling of the now listy Documentable.subkind
 19:30
19:32 treehug__ joined 19:33 treehug88 left 19:34 treehug8_ left, moistcherry is now known as LurkingStalker 19:37 denis_boyun___ joined 19:38 denis_boyun left, LurkingStalker is now known as hoverboard 19:39 brrt joined 19:43 jnap left
masak m: sub infix:«~~>»($premise, $conclusion) is looser(&infix:<max>) { !$premise || $conclusion } 19:46
camelia rakudo-moar 6e0ac1: OUTPUT«===SORRY!=== Error while compiling /tmp/Z2H_EHRDi2␤Can't use unknown trait 'is looser' in a sub+{precedence} declaration.␤at /tmp/Z2H_EHRDi2:1␤------> ␤ expecting any of:␤ rw parcel hidden_from_backtrace␤ pure…»
masak "unknown trait"?
the error report is cut off early, but "looser" is in the list of the traits it expects.
lizmat I'm not sure "is looser" is really implemented 19:49
ah, I see what the problem is 19:50
sub +[precedence] !~~ Routine
masak m: say &infix:<max>
camelia rakudo-moar 6e0ac1: OUTPUT«Sub+{<anon>}.new()␤»
dalek c/redesign: 3b8e741 | Mouq++ | / (2 files):
Rename .subkind to .subkinds
 19:51
c/redesign: 57ef3c5 | Mouq++ | htmlify.p6:
Abstract index writing process
masak lizmat: why wouldn't it match Routine. it's a subclass, no?
moritz known bug
my: my &c = &die;
m: my &c = &die;
camelia rakudo-moar 6e0ac1: OUTPUT«Type check failed in assignment to '&c'; expected 'Callable' but got 'Sub+{<anon>}'␤ in block at /tmp/KTRbI9uBLb:1␤␤»
moritz same bug 19:52
(already submitted)
lizmat masak: &die has "is hidden_from_backtrace"
which makes it not mach
*match 19:53
masak ah :/
ok, anyone know an operator I can use that's just looser than || ?
er, rephrasing.
ok, anyone know how I can write to make my operator just looser than || ?
the one who cracks this gets a favorable mention in the blog post I'm writing... :) 19:54
segomos_ looser how?
[Coke] precedence wise.
19:55 jnap joined
[Coke] so if you have an expression with || and his new operator… 19:55
segomos_ ah didn 19:56
ah didn't see his example earlier
lizmat perhaps the answer lies in src/core/Precedence.pm
masak segomos_: it's Perl 6 terminology. tighter/looser instead of higher/lower.
ooh, 'is prec' :)
lizmat++
m: sub infix:«~~>»($premise, $conclusion) is prec<j=> { !$premise || $conclusion } 19:59
camelia rakudo-moar 6e0ac1: OUTPUT«===SORRY!=== Error while compiling /tmp/ChSx1DYSym␤Can't use unknown trait 'is prec' in a sub+{precedence} declaration.␤at /tmp/ChSx1DYSym:1␤------> ␤ expecting any of:␤ rw parcel hidden_from_backtrace␤ pure d…»
masak doesn't understand where the error comes from
Mouq m: sub infix:<ll> ($a,$b) is looser(&[||]) { "$a ll $b" }; say 1 ll 2
camelia rakudo-moar 6e0ac1: OUTPUT«===SORRY!=== Error while compiling /tmp/UQKd4RUvPd␤Routine given to looser does not appear to be an operator␤at /tmp/UQKd4RUvPd:1␤------> ␤»
masak surely the sub isn't a sub+{precedence} before 'is prec' has been applied!? 20:00
Mouq m: sub infix:<ll> ($a,$b) is looser(&infix:<||>) { "$a ll $b" }; say 1 ll 2
camelia rakudo-moar 6e0ac1: OUTPUT«===SORRY!=== Error while compiling /tmp/cczdqQvn11␤Routine given to looser does not appear to be an operator␤at /tmp/cczdqQvn11:1␤------> ␤»
masak Mouq: infix:<||> is too special. I already tried that.
Mouq Huh
segomos_ those were Ls
masak arguably, it *should* work. but it doesn't.
segomos_: no, in the 'is looser'.
segomos_: doesn't matter what the op being defined is called. 20:01
masak compromises for now and uses parens :(
20:02 SevenWolf joined
lizmat masak: using is prec or is looser doesn't make a difference as they use the same basic mechanism to find what to apply to 20:02
and that's børked atm
masak in conclusion, I cannot set precedence right now? 20:04
lizmat it appears like that, on any operator you define 20:05
would be my conclusion
FROGGS_ isn't there a test for that?
20:06 SevenWolf left
masak I saw tests in roast with operators defined with precedence, yes. 20:07
20:07 SevenWolf joined
masak don't know if they are in spectest.data, though. 20:07
lizmat t/spec/integration/advent2012-day19.t
for instance
20:08 mr-foobar left
dalek ast: 0a4db36 | (Elizabeth Mattijsen)++ | integration/advent2012-day19.t:
Fix typo
 20:10
jnthn S06-traits/precedence.t is in spectest.data 20:13
Passes 9 out of 10 tests 20:14
lizmat jnthn: is the a performance difference between using $x and \x as a parameter to a method ?
*there
jnthn Potentially. 20:15
lizmat + multi method print(IO::Handle:D: Str:D \x) {
+ nqp::printfh($!PIO, nqp::unbox_s(x));
would that make sense ?
jnthn Str !~~ Iterable
lizmat (rather than $value, which is what the param is called now)
jnthn So I think it optimizes away the need for a Scalar container there.
lizmat ok, then I'll commit 20:16
jnthn So $x is going to perform as well as \x
In that case.
lizmat ah, so it *doesn't* make sense
git reset --hard # so much for that opt :-) 20:17
moritz git reset --that-hurt 20:22
Mouq This doc work is soo boring :P It takes an eternity to re-gen the docs. I really need to implement caching of some sort.. 20:23
lizmat "is cached" to the rescue?
20:23 guru joined
Mouq lizmat: Unfortunately, I mean cacheing between runs 20:24
PerlJam Mouq: continuous deployment :)
yoleaux 15 Jun 2014 11:32Z <pippo> PerlJam: I found the solution for irclog.perlgeek.de/perl6/2014-06-10#i_8850807 see irclog.perlgeek.de/perl6/2014-06-15#i_8873951
20:24 guru is now known as Guest32892, Guest32892 is now known as ajr_
lizmat no one is stopping you from implementing "is persistent" :-) 20:24
Mouq :) 20:25
jnthn Mouq: Thanks for reporting the Moar SEGV earlier; fixed in 462402be7f. 20:26
dalek c/redesign: d7a28e7 | Mouq++ | htmlify.p6:
Write out routine sub-indices
 20:27
Mouq jnthn: \o/ Thank you :)
jnthn gist.github.com/peschwa/863b4d839e0024388052 also doesn't SEGV with that fix for me 20:28
dalek kudo/nom: 09a1674 | (Elizabeth Mattijsen)++ | src/core/IO.pm:
Small opt on IO.print: not sure
kudo/nom: ccc5193 | (Elizabeth Mattijsen)++ | src/core/Deprecations.pm:
Fix obsolete variable deprecations
jnthn Please could somebody with a Panda commit bit apply this Windows fix: gist.github.com/jnthn/8b51ba6d9757d84a4957 20:32
20:32 psch left, virtualsue left 20:33 maelus joined
ChoHag Does anyone know of any documentation describing how to define and use new traits on class attributes? 20:33
20:34 itz left
dalek ast: 19bd159 | (David Warring [email@hidden.address] | integration/advent2011-day11.t:
removed guff
 20:34
ast: 68b1ae8 | (David Warring [email@hidden.address] | integration/advent2011-day15.t:
adding advent 2011 day 15
PerlJam ChoHag: There's probably an advent calendar entry about it, but I don't think there's "documentation" as such. (I do think that would make a good blog post though to help the next guy) 20:35
dalek c/redesign: 4494168 | Mouq++ | / (2 files):
Use placeholder= rather than value= in search bar

This way, "Search" disappears when focus is put in the textarea
ChoHag Yeah I've given up (in the best of ways) expecting full on documentation for much of perl 6. 20:36
Mouq PerlJam: Or, you know, we could add some info to perl6/doc about it :P
20:37 itz joined, kaare_ left
PerlJam Mouq++ yeah, that too :) 20:37
jnthn r: multi trait_mod:<is>(Mu $what_is_it, :$curious!) { say $what_is_it.WHAT }; class A { has $.x is curious; } # I really don't know what this does :)
camelia rakudo-{parrot,jvm,moar} 6e0ac1: OUTPUT«===SORRY!=== Error while compiling /tmp/tmpfile␤Can't use unknown trait 'is curious' in an attribute declaration.␤at /tmp/tmpfile:1␤------> ␤ expecting any of:␤ rw␤ readonly␤ box_target␤» 20:38
jnthn oops
m: multi trait_mod:<is>(Mu $what_is_it, :$curious!) { say $what_is_it.WHAT }; class A { my $.x is curious; } # I really don't know what this does :)
camelia rakudo-moar 6e0ac1: OUTPUT«===SORRY!=== Error while compiling /tmp/fPr8ahe8pz␤Can't use unknown trait 'is curious' in a variable declaration.␤at /tmp/fPr8ahe8pz:1␤------> s_it.WHAT }; class A { my $.x is curious⏏; } # I really don't know what…»
jnthn Hmm
Oh...dammit
PerlJam See?!? Even *jnthn* doesn't know how to do it! ;)
jnthn I bet the "helpful" error message hides the candidate I'm after.
ChoHag Well if this place doesn't know I don't see what hope I have... 20:39
jnthn Oh, but it tells me what I wanted to know.
m: multi trait_mod:<is>(Variable:D $var, :$curious!) { say "in trait mod" }; class A { my $.x is curious; }
dalek kudo/nom: 3911c1c | (Elizabeth Mattijsen)++ | t/spectest.data:
Add advent2011-day15 tests
camelia rakudo-moar 6e0ac1: OUTPUT«in trait mod␤»
jnthn ChoHag: Well, that's how you can declare a trait that gets applied there.
ChoHag What's the :D ? 20:40
jnthn "Definite" (not a type object)
Mostly me being paranoid, 'cus I'm used to defining traits in core. :)
masak how do I sort two arrays (of known equal length) in lexicographic order, i.e. comparing elements pairwise and sorting on the first differing element? 20:47
the shorter the solution, the better.
preferably, I'd like to just use some metaop, but I don't really see how. 20:48
dalek c/redesign: ff624ef | Mouq++ | htmlify.p6:
Operators are routines

This gets as far as generating the index files, but dies trying to hack together the urls for the operators. On a positive note, some of the steps involved to make this work are also steps toward generalizing the documentation of types, routines, etc.
jnthn masak: Hm, example? 20:49
lizmat masak: would the result still be 2 arrays ?
Mouq sub infix:<sort> ($a, $b) { sort $a, $b }; @a Zsort @b # ??
Don't we have a zip :by(&sort), actually? 20:50
m: say zip (1,2,3), (4,5,6)
camelia rakudo-moar 6e0ac1: OUTPUT«1 4 2 5 3 6␤»
Mouq m: say zip (1,2,3), (4,5,6), :by(&sort)
camelia rakudo-moar 6e0ac1: OUTPUT«1 4 2 5 3 6␤»
Mouq m: say zip (1,2,3,9), (4,5,6,7), :by(&sort)
camelia rakudo-moar 6e0ac1: OUTPUT«1 4 2 5 3 6 9 7␤» 20:51
Mouq Hm :(
masak lizmat, jnthn: I currently have what I want as `sub array-comp(@xs, @ys) { [||] @xs »cmp« @ys }`
looking for better/shorted suggestions.
shorter* 20:53
jnthn That's already quite short :P
20:53 stux left
Mouq any @xs Zcmp@ys :P 20:53
jnthn Zcmp may allow it to short-circuit more effectively... 20:54
masak yes, good idea.
`any` doesn't do what I want, tho'
20:57 stux joined
dalek p: 9bb18da | (Elizabeth Mattijsen)++ | tools/build/MOAR_REVISION:
Bump MOAR_REVISION
 20:57
vendethiel Isn't there a neater way to write ` my @parts = $version.split('.'); for @els.kv -> $k, $v { @parts[$k]++ if $v; } @parts.join('.');` ? I feel like I'm missing something 20:58
20:59 hoverboard left
dalek c/redesign: 8569817 | Mouq++ | htmlify.p6:
Make sure we sort sub-index output
 20:59
lizmat testing bumped NQP_REVISION now 21:01
Mouq "@parts = $version.split('.') Z+ ?<<@els; @parts.join('.')", though that's not exactly readable
Technically it should be <<+<< insetad of Z+ 21:02
jnthn Rakudo's startup time seems to have got quite a bit worse since the last relesae 21:05
21:05 pecastro joined
vendethiel $version.split('.').map(* + @els[(state $)++].Int).join('.'); 21:06
Okay, now I think I'm doing perl 6, not "ruby/python/php in perl 6"
lizmat jnthn: this last MOAR version bump also seems to have a negative effetc :-(
running spectest a 2nd time
vendethiel Mouq: thanks, I went with ^ but yours is funny too P
jnthn Well, since the last MOAR_REVISION bump there was only 462402b 21:08
21:09 Rotwang left
jnthn lizmat: Which is just an extra NULL check to avoid a SEGV. So seems...ulikely 21:09
lizmat indeed, was noise on my side
dalek kudo/nom: 54e264e | (Elizabeth Mattijsen)++ | tools/build/NQP_REVISION:
Bump NQP_VERSION
 21:10
21:10 bluescreen10 left 21:11 Psyche^ left, araujo left
jnthn lizmat: OK, phew :) 21:11
21:11 Psyche^ joined
jnthn lizmat: Is tehre any new/extra setup work of note since the last relesae? 21:11
lizmat is looking at the log
21:12 sivoais joined
jnthn lizmat: Also this: 21:12
Mouq vendethiel: Cool :)
jnthn - $out.print(x.gist);
- $out.print("\n");
+ $out.print(x.gist, "\n");
Will be slower
'cus it hits the slurpy candidate of print
Which is gonna be WAY slower than the Str-taking one.
lizmat even if you're calling it twice?
jnthn Yes.
lizmat and have an extra alias?
jnthn Because iterating the slurpy is some method calls :) 21:13
21:13 araujo joined
lizmat ok, will rewrite 21:13
jnthn Furthermore, the Moar inliner, when we get it, will be able to take a good shot at inlining the two simple .print calls
lizmat but creating a candidate for Str:D does make sense, no? 21:14
jnthn Uh, well, when the inliner can inline Rakudo methods properly
lizmat so we don't have to do the .gist ?
jnthn 6e0ac15e5 makes a lot of sense
aaa82dd9 doesn't
You are timing these before/after, I guess? :)
vendethiel Mouq: it's actually far more complicated haha :'( 21:15
lizmat only globally CPU in the spectest
jnthn Oh
lizmat but it's more or less in noise levels in the spectest
jnthn But...these things are very benchmarkable. 21:16
Right.
But a loop doing printing will show it up
Not to mention that it could make a good perl6-bench contribution; I think we've no IO tests tehre. 21:17
s/tests/benchmarks/
21:19 hoverboard joined
vendethiel Mouq: well... I ended up with this 21:21
Mouq: $version.split('.').map({any(@els[0..(state $)++]) ?? ( (state $)++ ?? 0 !! $_ ) !! $_}).join('.'); 21:22
jnthn lizmat: Taking a read through CORE.setting to see if I spot anything that way...
timotimo i'm glad to see that commandline argument passing with utf8 stuff works out of the box in rakudo perl6 21:23
lizmat jnthn: wow, what a difference:
$ time 6 'for ^100000 { say "foo" }' >/dev/null
real0m4.296s
user0m4.090s
after: $ time 6 'for ^100000 { say "foo" }' >/dev/null 21:24
real0m1.269s
user0m1.096s
sys0m0.163s
jnthn lizmat: With what change?
timotimo what the hell did you do? o_O
lizmat undo my "optimization"
jnthn And this is why to benchmark :P
lizmat hehe, wish we had a Banchmark module
vendethiel
.oO( and a banchee )
jnthn Comitting optimizations without benchmarking them is like drinking from a random bottle you find in the shed. It might be nicely matured homebrew you forgot about, or it could are arsenic. 21:25
*be
lizmat
.oO( "This could go either way" )
lizmat wonders how many people get the 3rd rock reference
vendethiel Now I really wish state++ was allowed. 21:26
lizmat as opposed to? 21:27
vendethiel (state $)++
m: sub a { state++ }; say a; say a;
camelia rakudo-moar 6e0ac1: OUTPUT«===SORRY!=== Error while compiling /tmp/_qr9KjsKH3␤Malformed state␤at /tmp/_qr9KjsKH3:1␤------> sub a { state⏏++ }; say a; say a;␤ expecting any of:␤ scoped declarator␤»
Mouq std: state$++
camelia std 0f2049c: OUTPUT«===SORRY!===␤Malformed state at /tmp/_6tsot63i8 line 1:␤------> state$⏏++␤ expecting twigil␤Parse failed␤FAILED 00:00 122m␤»
Mouq std: ++state$
camelia std 0f2049c: OUTPUT«ok 00:00 122m␤»
lizmat std: state++
camelia std 0f2049c: OUTPUT«===SORRY!===␤Malformed state at /tmp/8A9psL5di1 line 1:␤------> state⏏++␤ expecting any of:␤ name␤ scoped declarator␤ statement end␤ statement list␤Parse failed␤FAILED 00:01 121m␤»
dalek osystem: dddcfd6 | segomos++ | META.list:
Update META.list

adding HTTP::Server::Async		 21:28
vendethiel m: say (++state$)
camelia rakudo-moar 6e0ac1: OUTPUT«1␤»
vendethiel ooh.
timotimo yay!
segomos_++
segomos_ super basic - time to build out some middlewares for it :-)
vendethiel too bad that ... 21:29
m: say [++state$];
camelia rakudo-moar 6e0ac1: OUTPUT«===SORRY!=== Error while compiling /tmp/uj5DPj5b8p␤Unsupported use of $] variable; in Perl 6 please use $*PERL_VERSION␤at /tmp/uj5DPj5b8p:1␤------> say [++state$]⏏;␤»
vendethiel maybe TimToady++ will want to "fix" this one too.
lizmat wonder if we could define ++state and state++ as syntactic sugar for (++state$) and (state $)++ respectively
segomos_ m: say state$
camelia rakudo-moar 6e0ac1: OUTPUT«(Any)␤»
vendethiel lizmat: I'd like it ;-) 21:30
segomos_ what is that supposed to do?
ChoHag What does "Returns the package (class/grammar/role) to which this attribute belongs." mean in the documentation of Attribute.package?
timotimo segomos_: does the server parse the complete body of the request before calling the registered callables?
segomos_ timotimo: yea - i should add that to the readme
ChoHag I get something back, but every time I use it perl just says "Method <something, like gist> not found for invocant of class 'Foo'" 21:31
timotimo segomos_: i don't like that :)
ChoHag I want to be able to adjust the class at compile time with an attribute trait.
lizmat "adjust the class" as in re-bless (p5 speak?)
ChoHag No. In this case, adjust a class (not instance) attribute it holds. 21:32
segomos_ timotimo: what should it do/
ChoHag But I expect that adding (meta-)attributes or methods should follow roughly the same lines.
timotimo segomos_: consider websockets; IMO it should call the callables as soon as the headers are finished being parsed
btyler segomos_: woah, this is super cool. will play with it this evening :) 21:33
vendethiel "github.com/Nami-Doc/xyz.pm6/commit...55dc1131", eww but works.
segomos_ timotimo: couldn't that lead to problems if you need the POST data (or whatever) to be completed?
ChoHag class foo { my @bar; has $baz is bingly; }, where bingly does something to @bar.
segomos_ btyler: thanks :-)
lizmat ChoHag: S12:793
synopsebot Link: perlcabal.org/syn/S12.html#line_793
ChoHag bingly being defined as: multi trait_mod:<is>(Attribute $a, :$bingly!) is export { ... }
timotimo segomos_: of course the data should be retrievable in a blocking manner if the user wants it 21:34
segomos_ timotimo: ahh i got you, call registered subs and *if* post data is requested then block until it's complete
jnthn ChoHag: .package is the class the attribute is declared in
timotimo IMO the registered subs ought to block by themselves
ChoHag jnthn: So what can I do with it? 21:35
segomos_ brb
ChoHag $a.package.foo complains as above for just about any foo.
jnthn ChoHag: Call a meta-method on it typically
21:36 isBEKaml joined
ChoHag $a.package.^foo? 21:36
That complained, but I only tried one thing...
jnthn I'm not sure the .^ shortcut works because the default parent wasn't yet added
But you can get at it with .HOW
Which always works.
timotimo segomos_: also, the registered handlers could say "i'm not interested" ASAP if only the headers are needed to decide
so you can reach the correct handler with less delay
ChoHag Ah.
Now where was that documentation I found on class meta methods... 21:37
vendethiel Golf ! {any(@els[^++state$]) ?? ( (state $)++ ?? 0 !! $_ + 1 ) !! $_}
dalek kudo/nom: 354c5c5 | (Elizabeth Mattijsen)++ | src/core/IO.pm:
Properly optimize say/note, jnthn++

Committing optimizations without benchmarking them is like drinking from a random bottle you find in the shed. It might be nicely matured homebrew you forgot about, or it could are arsenic.
jnthn m: multi trait_mod:<is>(Attribute:D $attr, :$foo!) { my $pkg = $attr.package; $pkg.HOW.add_method($pkg, 'foo', method () { say 'look a foo' }) }; class A { has $.x is foo }; A.foo 21:38
camelia rakudo-moar 3911c1: OUTPUT«look a foo␤»
jnthn ChoHag: ^^ may help
ChoHag It does. Thanks.
There's some in S12 here too. 21:39
timotimo lizmat: so are say and note now faster than they used to be a week ago or is it back to the way it was recently?
lizmat they should be faster for a single Str
because there is no .gist done on them
but that's the only difference 21:40
timotimo ah, that's good!
21:41 donaldh joined
vendethiel the $] kills me :'( 21:41
segomos_ timotimo: awesome, gonna implement that -
donaldh [Coke]: Here's the perl6-roast-data results from earlier: gist.github.com/donaldh/c37b5fdf96da8f5e6e7f 21:42
dalek ecs: 309a882 | (Elizabeth Mattijsen)++ | S22-package-format.pod:
Some more S22 tweaks, introducing .parse-spec
 21:43
21:43 rurban joined 21:44 dmol1 left
timotimo segomos_: in general, websocket support would be pretty nifty 21:45
segomos_ i might try to tackle that next, i've used them but never read about how they work 21:46
21:46 dmol joined
dalek c/redesign: 4a477cb | Mouq++ | htmlify.p6:
Get operators-as-routines to work
 21:47
c/redesign: ec0abda | Mouq++ | sketch/ (22 files):
Remove sketch dir

No longer needed
btyler segomos_: hmm, I seem to be segfaulting r-m with great regularity on the echo example (just curl-ing localhost:8080) -- have you noticed the same?
segomos_ btyler: moar has a bug where it seg faults, it only works with java right now 21:48
i think there is a race condition or something on the moar implementation 21:49
btyler segomos_: ok, glad it isn't just me
segomos_ because it will sometimes respond and sometimes won't, and sometimes it will segfault depending on .. something :-) 21:50
lizmat segomos_: is this still on the latest and greatest after the Moar/NQP version bumps I did today? 21:51
segomos_ lizmat: i'll check it out
btyler lizmat: I tried with master/master/nom as of 20 minutes ago and still found the segfault 21:52
lizmat :-(
21:52 SevenWolf left
masak lol! I blogged! strangelyconsistent.org/blog/boxes-and-pebbles 21:52
lizmat I guess jnthn / nwc10 still have some heavy debugging ahaed
masak (long post. but hopefully worthwhile.) 21:53
ChoHag What does 'will' do (attribute trait)?
21:53 SevenWolf joined
ChoHag Set the default value? 21:53
[Coke] donaldh: looks much better than what I get. :)
lizmat ChoHag: set a phaser for the variable 21:54
[Coke] I only have 1GB of swap
ChoHag Oh it's always somethi (new and) weird...
donaldh We will need to look at the exact config on the roast server.
lizmat my $dbh will leave( { .disconnect };
ChoHag So what's a phaser then?
donaldh Maybe increasing the swap would be a good start
ChoHag I read something about that I didn't understand. 21:55
[Coke] wrong box.
ChoHag Or remember.
[Coke] one sec.
lizmat S04:1387
synopsebot Link: perlcabal.org/syn/S04.html#line_1387
ChoHag Ah yes those.
[Coke] 3954MB Mem, 0MB Swap (per htop)
ChoHag Hmm. 21:56
jnthn lizmat: Result. 21:57
donaldh Can you add some swap?
jnthn lizmat: src/core/Inc.pm takes nearly 0.1s to execute
lizmat: Which accounts for the startup time regression. 21:58
21:58 colomon left
lizmat well, that's code that came from the eleven branch merge 21:59
I'm actually refactoring a lot of that
jnthn OK
Well, anything to make it not be a huge startup time regression would be really great.
lizmat I must also point out that maybe a lot of that is because a lot of directories are being tested 22:02
btyler pretty wild watching the JVM JIT warm up: I ran wrk against segomos_'s async http server. first round was ~6 requests/second, 2nd: 13, 3rd round: ~40 22:03
22:03 colomon joined 22:04 dmol left
btyler 40 requests a second is well into "servicable for a wide variety of webapps" territory, so that's pretty neat :) 22:04
granted it's an empty echo, but still
jnthn lizmat: I don't really have the energy to dig through the code, I'm afraid. I can just say that of all the code we run in CORE.setting's mainline at startup, Inc.pm accounts for over two thirds of the time. 22:05
lizmat ok, will see what I can do about that
jnthn lizmat++
tadzik btyler: I read that as "MoarVM JIT" and my heart stopped a little :)
22:05 dmol joined
btyler seems to be holding steady around 45-50 rps, with the occasional uncaught exception taking the server down 22:06
lizmat tadzik: be patient :-)
btyler tadzik: heh, sorry :P
[Coke] donaldh, moritz: can we add swap to host07?
22:07 brrt left
jnthn btyler: Got a link for me to the script that SEGVs on Moar? 22:07
segomos_ github.com/tony-o/perl6-http-serve...r/examples
either of those jnthn
btyler jnthn: github.com/tony-o/perl6-http-serve...server.pl6
ran it and then curl localhost:8080 22:08
jnthn When I run it, it just says "closed" 22:10
And exits immediately.
dalek c/redesign: ee92b0b | Mouq++ | / (2 files):
Generate operator index
 22:11
22:11 treehug__ left
jnthn oh, it's 'cus port 8080 was in use 22:12
lizmat jnthn: seems that at least .06 of startup time is spent setting up CUSTOM_LIB 22:15
and testing whether directories exist
dalek c/redesign: c0bdd05 | Mouq++ | htmlify.p6:
Make sure infixes in lib/Type work and stop generating html/op dirs
lizmat I'll refactor it so that @*INC will contain strings initially only 22:16
22:16 amirite left
lizmat to be turned into actual objects (or removed from @*INC) when they're actually needed 22:16
jnthn: it feels like overkill to turn @*INC into an iterator 22:17
masak 'night, #perl6 22:21
timotimo masak: nice post!
lizmat gnight masak 22:22
jnthn 'night, masak 22:25
lizmat: yeah, think so
22:29 isBEKaml left, rurban left 22:30 spider-mario left 22:34 denis_boyun___ left, colomon left 22:35 kivutar left, rurban joined 22:36 colomon joined
jnthn btyler, segomos_: I'm wondering if the issue is that .write is async, but it seems there's nothing being done with the Promise that is returned 22:38
So .close happens on the socket before the writing is done
It's SEGVing 'cus of a missing "did we close the handle" check
I'll fix that tomorrow, but I think that needs addressing in the module. 22:39
Is socket writing on an async socket actually async on the JVM?
timotimo maybe it isn't and that's the cause of the problem :) 22:40
jnthn Well, maybe it isn't and that's why the module works there :P 22:41
22:42 Araq joined
timotimo er, i mean: the cause of the lack of the problem 22:42
Araq timotimo: !
timotimo hi there! 22:43
22:44 carlin left 22:47 xragnar_ joined, xragnar is now known as Guest14585, Guest14585 left, xragnar_ is now known as xragnar 22:49 ren1us left 22:50 btyler left, telex left 22:52 telex joined 22:56 colomon left 22:59 hoverboard left
jnthn sleep & 23:02
timotimo gnite jnthn 23:03
kurahaupo watches jnthn perform "sleep" in the background while continuing other tasks :)
g'nite 23:04
23:05 itz left
lizmat gnight jnthn 23:06
23:06 Alina-malina left 23:07 Alina-malina joined
lizmat also calls it a day 23:11
gnight, #perl6!
timotimo gnite lizmat 23:12
23:14 donaldh left 23:18 ajr_ left
segomos_ jnthn: so is moar representative of what that async socket stuff is actually doing? 23:19
or should be doing? is the thought that the jvm is in a workaround mode for async soxkets?
23:19 dmol left 23:22 hoverboard joined 23:31 hoverboard left 23:37 BenGoldberg joined 23:43 BenGoldberg left, BenGoldberg joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!