»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.
diakopter it's the only subrule call I can find, and subrule is in the backtrace 00:00
and the error is Don't know how to save info from node of type String
at least, I think it falls under subrule 00:01
cognominal rakudo: grammar G is HLL::Grammar { } 00:04
p6eval rakudo 8839b1: OUTPUT«===SORRY!===␤No applicable candidates found to dispatch to for 'trait_mod:<is>'. Available candidates are:␤:(Attribute $attr, Any $rw)␤:(Attribute $attr, Any $readonly)␤:(Routine $r, Any $rw)␤:(Routine $r, Any $default)␤:(Routine $r, Any $info, Any $inlinable)␤:(Pa…
00:07 cooper left
diakopter hunh. nqp (not nqpq) fails a lot of the 01-regex.t tests 00:08
I'm not much help today 00:10
jnthn diakopter: Is that maybe a side-effect of having build nqpq? (just a wild guess...) 00:12
diakopter no idea
diakopter continues commenting parts of 01-regex.t out until it compiles
hm, maybe it's in NQPHLL 00:15
diakopter looks for that
no, that's already compiled to pbc 00:16
jnthn diakopter: oh!! 00:17
my $result := $string ~~ /<$regex>/;
diakopter: OK, that looks like it's because interpolation is not yet implemented in nqpq.
diakopter: So it's same reason 49-regex-interpolation.t rails 00:18
er, fails
00:19 _dolmen_ joined
diakopter well, even commenting those two out doesn't seem to fix 00:19
jnthn diakopter: the .t file in p6regex.t is a test runner 00:20
diakopter I know 00:21
jnthn diakopter: It won't work until 49-regex-interpolation.t does.
diakopter actually, commenting both of those out makes it give a different error
00:21 _dolmen_ left
diakopter Method 'CURSOR' not found for invocant of class 'NQPMatch' 00:21
jnthn oh, that one is probably easy to fix :)
diakopter looks for NQPMatch 00:22
jnthn It's somewhere in src/QRegex/Cursor.nqp iirc 00:23
diakopter adds it
00:24 icwiener left
diakopter wonders how to impl interpolation 00:24
00:24 cooper joined 00:29 localhost left
dalek p: e1a19de | diakopter++ | src/QRegex/Cursor.nqp:
add missing CURSOR method to NQPMatch
 00:29
00:30 localhost joined 00:36 PacoLinux_ joined
jnthn 'night, #perl6 00:52
diakopter jnthn: o/
00:52 PacoLinux_ left 00:56 molaf_ joined 00:58 autin joined 01:00 molaf left 01:01 pmurias_ left
dalek ast: 41f3591 | coke++ | S05-interpolation/regex-in-variable.t:
rakudo fudge
 01:01
kudo/nom: 8c69edb | coke++ | t/spectest.data:
track failure modes/run fudged tests
 01:03
diakopter [Coke]: I think the two Multiple array tests there above your last commit are succeeding misleadingly 01:04
none of those interpolation ones will succeed 01:05
01:07 Mowah_ left 01:13 cooper left 01:14 tokuhirom joined 01:27 plutoid joined 01:41 [Sec] joined 01:54 wolfman2000 joined
colomon is too lazy to backlog -- anything exciting happen today? (Was at Christmas parade and then football game.) 02:02
02:12 pat_js left 02:18 tokuhirom left 02:19 IngisKahn left 02:21 buubot_backup left, tokuhirom joined 02:28 tokuhirom left
diakopter whee 02:32
found and fixed a bug
dalek p: fd2fd16 | diakopter++ | src/QAST/Compiler.nqp:
make conjseq handle zerowidth subtype
colomon diakopter++
02:36 aindilis left, aindilis joined 02:37 cooper joined 02:41 buubot_backup joined 02:42 risou_awy is now known as risou 02:48 am0c joined 03:00 Guest51139 left 03:05 lichtkind left
diakopter phenny: tell jnthn I fixed/finished and applied your negate/backslash patch 03:06
phenny diakopter: I'll pass that on when jnthn is around.
dalek p: 856a79c | diakopter++ | src/QRegex/P6Regex/ (2 files):
enable backslash sequences in character classes, enable negated character classes. patch mostly from jnthn++
 03:09
03:09 Logicus joined
Logicus Anyone out there? 03:10
diakopter hi
Logicus ah evening :)
03:14 Logicus_ joined 03:15 Logicus left
Logicus_ perl6: say 3; 03:15
p6eval pugs b927740, rakudo 8c69ed, niecza v11-24-gcdd4e66: OUTPUT«3␤»
Logicus_ hrm
03:15 tokuhirom joined
diakopter rakudo and niecza are the two actively developed implementations 03:16
Logicus_ ok
im completely new to perl 6
diakopter welcome :) 03:17
Logicus_ I've been learning Perl5 for a few years, but in a very heretical way
sjn That's better than most other people :)
Logicus_ had very big arguments with various PerlMonks who didn't like my ideas
sjn notes most other people don't know any Perl6 at all
Logicus_ it seems very mysterious somehow 03:18
sjn Logicus_: do your ideas make it easier for people to understand what's going on in your code? :)
Logicus_ yes.
sjn are you sure? :)
diakopter Logicus_: if your P5 ideas work for you, that's great. Maybe the monks disagreed with you if you said your ideas were the "right" way to do things. 03:19
Logicus_ to put it in a nutshell, i wrote a sort of Lispish language in Perl without even really knowing what I was doing
03:19 tokuhirom left
Logicus_ I want to start looking at Perl 6 with a view to getting my system recoded in it by the time it's ready 03:19
diakopter Logicus_: you have a choice of using rakudo, which currently runs only on the Parrot VM, and niecza, which currently runs on the CLR (both Microsoft .Net and mono) 03:20
Logicus_ well it's a long story, about 4 years ago I came to perlmonks all humble like and tried to share what I had and get feedback on how to make it faster
I'm using Ubuntu Maverick if that's any help... 03:21
sjn what kind of language constructs did you make that you thought Perl5 was missing?
Logicus_ they looked at my baby perl of the time and dismissed it
diakopter people can be dismissive :)
Logicus_ well I didn't know what was missing or what was in place or anything at the time
sjn critique can be a harsh mistress
Logicus_ I just solved the problem infront of me in what turned out to be a completely different way to how the mainstream solved it 03:22
your not wrong they made me want to completely give up on perl
diakopter everybody has a different learning path
sorear hello Logicus_
Logicus_ but I didn't, and just recently I've combined my system with PSGI/Plack
hi
sorear Logicus_: perl 6 has a lot of lisp influences 03:23
Logicus_ my learning path was through "tizag.com"
I'm good at solving problems by myself so that was all the reference I needed for a long time 03:24
sorear Logicus_: have you seen _Higher Order Perl_?
sjn half of all programming in communicating the intent of your program to the next guy (which might even be yourself in three months). Which is mainly a communication job; meaning you have to think about how you form your message so the recipient/reader has the best chance of understanding your message/code
Logicus_ yup, MJD
my system has some H.O.Pish properties that are very nice
sorear sjn: communication and memory are the same problem through special relativity-colored glasses
Logicus_ that was a big part of my problem... communicating what I had 03:25
sjn sorear: I have no idea what you just tried to say there :)
sorear heh.. communication is a large part of what we deal with every day
Logicus_ you see I came at it from a totally bizarre angle and I didn't know any of the mainstream stuff to relate it to at first
sorear we encourage many of our members to keep their heads in the clouds 03:26
sjn "mainstream
oops
Logicus_ that's cool :)
sorear and then every few days TimToady comes up with a brilliant but WTF idea and we have to figure out what he means
Logicus_ im really not a mainstream sort of person lol... the idea of getting a job doing mainstream perl and writing unit tests and things gives me day-mares
diakopter Logicus_: don't feel alone; I'm sure there are tens of thousands of Perl users like that 03:27
sorear Logicus_: I'm curious if you're out of school now. There seems to be a strong youth bias in P6 versus the broader Perl community
Logicus_ I'm 31
diakopter too
Logicus_ but I must admit, part of my brain is still about 15 03:28
:P
diakopter part of my brain is about 105 (for the worse, not for the better) :/
sjn "mainstream" means "something do people do". when it comes to languages, mainstream is really a positive thing. if many people use the same language for expressing things, more people have access to that expression. it's like choosing to write a book in a mainsteam language like "English" instead of a smaller (but perhaps more expressive) language like Northern Saami
Logicus_ well there is the whole spectrum eh, when 900 years old you reach... etc
sjn more accessibility (mainstreamy-ness) means more people can understand what you're trying to do, meaning more people can help you improve your work 03:29
diakopter has no clue how to solve rt.perl.org/rt3//Public/Bug/Displa...?id=100650
Logicus_ I've always wanted to get people on board with what I'm doing but so far I've failed to sell it very well 03:30
I'm just not a salesman at all
sjn critique = good
diakopter phenny: tell jnthn conj didn't handle zerowidth; I fixed that first
phenny diakopter: I'll pass that on when jnthn is around.
Logicus_ oh im all for critique, but just not dismissive critique coming from someone who hasn't really looked at the problem properly
sorear nom: say so 'aaabaaa' ~~ /^ (.*) b/
p6eval nom 8c69ed: OUTPUT«Bool::False␤»
sorear nom: say so 'aaabaaa' ~~ /^ .* b/ 03:31
p6eval nom 8c69ed: OUTPUT«Bool::True␤»
Logicus_ I just setup a new site www.perlnights.com
sorear nom: say so 'aab' ~~ /^ (a*) b/
p6eval nom 8c69ed: OUTPUT«Bool::True␤»
sorear nom: say so 'aab' ~~ /^ (aa || a) b/
p6eval nom 8c69ed: OUTPUT«Bool::True␤»
Logicus_ I've put loads of stuff up there
sorear nom: say so 'aab' ~~ /^ (a || aa) b/
p6eval nom 8c69ed: OUTPUT«Bool::False␤»
diakopter sorear: oooo
sorear diakopter: quantifiers have nada to do with this
Logicus_ How close are we to getting Perl 6 to perform like 5 does?
sorear backtracking into parens is broken 03:32
diakopter Logicus_: kindof a long way, unfortunately
sorear diakopter: do grammars and // use the same regex engine yet?
I remember nom diverged them for a while
diakopter in nom, yes I'm pretty sure. it's reproducible in nqpq too
sorear Logicus_: depends on the aspect by which you mean
nom: my $foo = / aa /; say so 'aab' ~~ /^ <$foo> b/ 03:33
p6eval nom 8c69ed: OUTPUT«Bool::True␤»
sorear nom: my $foo = / aa || a /; say so 'aab' ~~ /^ <$foo> b/
Logicus_ well I'd really like my aXML system to be the first web-dev framework out on Perl 6
p6eval nom 8c69ed: OUTPUT«Bool::True␤»
sorear nom: my $foo = / a || aa /; say so 'aab' ~~ /^ <$foo> b/
p6eval nom 8c69ed: OUTPUT«Bool::False␤»
sorear diakopter: it's not just inline () subrules, all subrules are backtrack-broken 03:34
diakopter sorear: yup
good find
sorear nom: my $foo = /:!r a || aa /; say so 'aab' ~~ /:!r ^ <$foo> b/ # I wonder if :r is being forced on by accident
p6eval nom 8c69ed: OUTPUT«Bool::False␤»
sorear nope, not that
Logicus_: too late for that
Logicus_ ok, one of the first
03:35 cooper left, natureboy joined
Logicus_ I'm too late with Perl5 I think, everyone already has strongly held beliefs about how things should be done with it 03:35
they don't like this sort of thing at all : <count str="<sqd>str</sqd>">(db_get)blog.chapter="(sqd)chapter(/sqd)".content(/db_get)</count> 03:36
sorear :P
...what?
Logicus_ exactly!
sorear I don't understand that at all
Logicus_ that's how far outside the mainstream I am lol 03:37
sorear you need to write a freaking manual
Logicus_ working on it
sorear good
Logicus_ www.perlnights.com/action.pl?action=library/aXML
sorear if you're going to speak using special jargon, you need to define your terms 03:38
Logicus_ understood :)
brb - ciggy break 03:40
sorear you might not want to name anything in a Perl community after Merlyn 03:41
that's the screen name of Randal L. Schwartz
Logicus_ yeh I didn't know there was a merlyn when I chose that 03:45
why not btw?
I picked that name to fit in with the castle theme (as the opposite to a monastery) 03:46
Kn1ghts and M0nks... 1/0 you see 03:47
I was shocked when one of the first posts in that section was by a user called merlyn! 03:48
then I looked up the name and found out about the schwartzian transform 03:49
I thought it was impressive?
hrm.. ping? 03:59
.... tumbleweed rolls by .... 04:03
04:04 kaleem joined
Logicus_ hi 04:04
sorear Logicus_: you're going to need to learn patience if you want to fit in here. 04:12
Logicus_ I was just wondering if I had lost connection again
im reading exegesis 2 by damian conway right now 04:13
04:14 kaleem left
diakopter Logicus_: that's quite outdated 04:14
just read www.perlcabal.org/syn/S02.html instead 04:15
Logicus_ do you... ah lol
read my mind :)
diakopter notice the pretty table of contents :)
Logicus_ now I know this page will make a beep when there is activity I can be looking elsewhere between saying stuff
that's quite a large table! 04:16
diakopter it's a very long Synopsis 04:17
sorear Logicus_: be warned that a common side-effect of hanging out here is ego deflation 04:19
Logicus_ what is it with programmers and ego eh?
04:19 natureboy left
sorear #perl6 is a pretty exceptional place 04:20
the "mainstream" encourages people to be loud and egotistical
diakopter stares at src/QAST/Compiler.nqp
sorear we're not that way
Logicus_ I've never been either unless pushed into it 04:23
(like on perlmonks)
diakopter wonders how backtracking into subrules could ever work in this engine 04:25
hrm 04:27
oh. 04:28
cursor_next
Logicus_ sorear, you didn't answer my Q about randal... is there something I should know about him? 04:35
04:38 molaf_ left
Logicus_ well I'm going to bed, thanks for that link diakopter, I will read the rest of that with interest tomorrow! 04:42
night!
diakopter o/ 04:43
04:43 Logicus_ left 04:57 benabik left, benabik joined 05:04 PacoLinux_ joined 05:08 PacoLinux_ left 05:11 IngisKahn joined 05:27 jdhore1 left
diakopter sorear: I added backtracking to subrule :) I feel proud 05:27
dalek p: 035ecba | diakopter++ | src/QAST/Compiler.nqp:
add backtracking support to subrule (captures included). Tested, barely. :)
 05:30
diakopter phenny: tell jnthn I added backtracking to subrule to fix the bug you pointed out. Seems to work for the test cases I tried. No idea whether it regresses anything else. :/ 05:31
phenny diakopter: I'll pass that on when jnthn is around.
diakopter I wish there were a p6eval target for nqpq 05:33
*was
well, there's a bug 05:36
it doesn't like nested captures 05:37
Null PMC Access
phenny: tell jnthn hrm; it doesn't like backtracking into nested captures. 05:39
phenny diakopter: I'll pass that on when jnthn is around.
diakopter maybe I used $P11 instead of %*REG<cur> somewhere 05:40
b: 23:30 <+dalek> nqp: 035ecba | diakopter++ | src/QAST/Compiler.nqp: 05:42
p6eval b 1b7dd1: OUTPUT«===SORRY!===␤Confused at line 22, near "23:30 <+da"␤» 05:43
diakopter er
23:30 <+dalek> nqp: 035ecba | diakopter++ | src/QAST/Compiler.nqp:
ugh
this is troubling. from Regex to QRegex, Pm removed the capability to store a cursor on the backtracking stack 05:44
ergh 05:46
TimToady with lazy lists of cursors you don't need a backtracking stack 05:47
diakopter QRegex doesn't have that I'm preetty sure 05:53
06:00 localhost left 06:01 localhost joined 06:33 am0c left 06:35 am0c joined 06:36 colorspace joined, colorspace left, am0c left 06:59 Chillance left 07:03 sayu joined 07:29 sayu left 07:47 tokuhirom joined 07:58 daniel-s joined 08:22 birdwindupbird joined 08:23 birdwindupbird left, birdwindupbird joined
diakopter hm 08:24
maybe I need to use the cstack
tadzik good morning 08:34
diakopter o/
hey, weren't you just here
tadzik hm?
diakopter oh, I guess 8.5 hours ago 08:35
time flies
tadzik oh, could've had a longer sleep ;)
hmm, to silence all prints/says in a lexical scope I could change $*OUT to... what? 08:43
I can $*OUT = $some-opened-file, but what if I want to silence it completely? 08:45
besides opening /dev/null
08:50 am0c joined
diakopter wonders whether I should stay awake until jnthn returns 08:50
sorear tadzik: I guess something like temp $*OUT but= role { method print(*@_) { } }; 08:55
tadzik aye, I was thinking about that too 08:56
sorear but that won't work on niecza because &CORE::say and &CORE::print bypass $*OUT entirely
the spec is silent on exactly which $*OUT methods call which other
09:04 birdwindupbird left, birdwindupbird joined
sorear actually it won't work on niecza because but still hasn't been completely implemented :/ 09:04
diakopter still doesn't know what but means 09:05
tadzik it's like does, but doesn't modify the object, but creates a new one? I think, unsure 09:06
0 does True would modify 0, 0 but True will create a funny object
nom: 0 does True; say ?0 09:07
p6eval nom 8c69ed: OUTPUT«No applicable candidates found to dispatch to for 'infix:<does>'. Available candidates are:␤:(Mu $obj, Positional @roles)␤:(Mu $obj, Mu $role)␤␤ in block <anon> at /tmp/e3Iivej8qn:1␤ in <anon> at /tmp/e3Iivej8qn:1␤»
tadzik okay
nom: class A { }; A does role { method a { say 'b' } }; A.new.?a; my $a = A.new but role { method b { say 'c' } }; $a.?b; $a.?a; A.new.?b 09:08
p6eval nom 8c69ed: OUTPUT«Cannot change the type of a type object␤ in mixin at src/gen/Metamodel.pm:925␤ in sub infix:<does> at src/gen/CORE.setting:7748␤ in block <anon> at /tmp/i4McikgHWe:1␤ in <anon> at /tmp/i4McikgHWe:1␤»
tadzik pff
nevermind me then
sorear nom: 0 does role { method Bool { True } }; say ?0 09:15
p6eval nom 8c69ed: OUTPUT«Bool::True␤»
sorear diakopter: in Perl 5 you can use bless to change the type of an existing object. Perl 6 has a much more restricted form of the same in that you can only change to sub-types created by adding one or more roles 09:16
$foo but Bar is exactly (except possibly in cases involving overriding of clone) equivalent to given $foo.clone { $_ does Bar; $_ } 09:17
tadzik: yes, you CAN change 0.
tadzik ah, I was just syntactically wrong
sorear there's been some discussion in the past about forbidding does on objects that override WHICH
no, you tried to change the type of a type object. 09:18
which I guess 6model forbids
with good reason, perhaps.
tadzik hmm 09:19
std: Ph'nglui mglw'nafh Cobol R'lyeh wgah'nagl fhtagn. 09:38
p6eval std be1f10e: OUTPUT«===SORRY!===␤Unsupported use of . to concatenate strings; in Perl 6 please use ~ at /tmp/CkyOC_BE3_ line 1 (EOF):␤------> mglw'nafh Cobol R'lyeh wgah'nagl fhtagn.⏏<EOL>␤Undeclared routine:␤ 'fhtagn' used at line 1␤Parse failed␤FAI… 09:39
tadzik std: Ph'nglui mglw'nafh Cobol R'lyeh wgah'nagl fhtagn
p6eval std be1f10e: OUTPUT«===SORRY!===␤Undeclared names:␤ 'Cobol' used at line 1␤ 'Ph'nglui' used at line 1␤ 'R'lyeh' used at line 1␤Undeclared routines:␤ 'fhtagn' used at line 1␤ 'mglw'nafh' used at line 1␤ 'wgah'nagl' used at line 1␤Check failed␤FAILED 00:01
..119m␤»…
09:43 autin left, plutoid left 09:46 risou is now known as risou_awy
sorear sleep 09:55
10:02 sftp left 10:03 sftp joined 10:06 tokuhirom left
moritz nom: role A[::T1, ::T2] { }; class B does A[Int, Int] { } 10:13
p6eval nom 8c69ed: ( no output )
moritz nom: subset UInt of Int where * >= 0; role A[::T1, ::T2] { }; class B does A[Int, UInt] { } 10:14
p6eval nom 8c69ed: ( no output )
moritz nom: subset UInt of Int where { 0 < $_ < 100 }; role A[::T1, ::T2] { }; class B does A[Int, UInt] { } 10:15
p6eval nom 8c69ed: ( no output )
moritz hm
in the setting I get
Cannot use 'Int, UInt6' as an argument to a parametric role as its value is not known at compile time at 10:16
ah, because I misspelled UInt64
dalek kudo/Rational: 6853995 | moritz++ | src/core/Rat (2 files):
make Rational a parametric role

hangs during setting compilation :(		 10:32
diakopter stays awake waiting for jnthn 10:33
moritz has anybody updated the regex tests from ** to %? 10:47
diakopter hasn't seen it
10:52 tokuhirom joined
diakopter I can't imagine there were very many using it for separator? 10:52
moritz t/spec/S05-metasyntax/repeat.t does, at least 10:54
moritz tries to update 10:57
11:01 spine joined
jnthn o/ 11:02
phenny jnthn: 03:06Z <diakopter> tell jnthn I fixed/finished and applied your negate/backslash patch
jnthn: 03:30Z <diakopter> tell jnthn conj didn't handle zerowidth; I fixed that first
jnthn: 05:31Z <diakopter> tell jnthn I added backtracking to subrule to fix the bug you pointed out. Seems to work for the test cases I tried. No idea whether it regresses anything else. :/ 11:03
jnthn: 05:39Z <diakopter> tell jnthn hrm; it doesn't like backtracking into nested captures.
diakopter :)
jnthn Wow, somebody has been busy ;)
oh, curious...so I got the patch right and the compiler got it wrong... ;) 11:06
diakopter yeah
jnthn diakopter++ 11:07
and you improved my cclass patch some too :) 11:08
diakopter my solution for \\ was kinda hacky; probably you have a better one
jnthn Well, the backslash escape thingy makes JSON::Tiny able to use that approach rather than all the <before ...> stuff 11:13
nom: say 3.008 / 4.073
p6eval nom 8c69ed: OUTPUT«0.738521973974957␤»
jnthn diakopter: I fear that your backtrack patch has hosed the spectest though :( 11:14
diakopter eek 11:15
jnthn S05-mass\recursive # fails a bunch
S05-mass\rx # fails to compile
eek, and then one hangs
diakopter odd.
jnthn S05-match\capturing-contexts hangs
diakopter any of them that use nested rule calls and backtrack will fail, I think 11:16
dalek ast: e2b7423 | moritz++ | S05-metasyntax/repeat.t:
start to switch regex tests from ** to %
jnthn Not had enough coffee to look into that yet really
diakopter: Mind if I move that patch to a branch for now?
diakopter nope
moritz diakopter: fwiw the roast patch added a skip for a now-hanging regex test in repeat.t 11:17
jnthn sigspace test hung too
dalek ast: 1ec9f11 | moritz++ | / (6 files):
rakudo unfudges
 11:18
moritz that one involved sigspace too, iirc
dalek p: 93744c1 | jnthn++ | src/QAST/Compiler.nqp:
Revert "add backtracking support to subrule (captures included). Tested, barely. :)"

This reverts commit 035ecba0590b0dbd0a66836d5ff06db568436714.		 11:19
jnthn diakopter: It's in the branch "backtrack"
diakopter not being able to backtrack into subrule & captures is kindof a huge deal :/ I'm very surprised Pm didn't do it yet 11:20
jnthn I thought that's what he'd put the $!restart stuff in for. 11:21
I wonder if he got half way and didn't get chance to finish it.
diakopter I think $!restart is only used from cursor_next, and cursor_next wasn't called from anywhere 'til my patch 11:22
jnthn Then, sounds like "yes"
diakopter: Undoing the backtracking patch fixes all the regressions apart from S05-mass/rx.t... 11:27
So guess something in the cclass patches has done that
jnthn investigates 11:28
oh wow, it dies with the least useful error ever.
moritz try --ll-exception
well, only the BT will be better...
jnthn moritz: yeah, doing so :) 11:29
hmm...blows up in conjseq
diakopter ack
diakopter wonders how that could be 11:30
jnthn diakopter: It's not your change, by the looks of it.
diakopter: My best guess is that the cclass code sometimes makes an empty conjseq
11:30 MayDaniel joined
diakopter ah 11:31
I'll buy that for a euro
or a gyro 11:32
mmm gyros. dieting by portion control is painful. 11:33
hm 3:33 a.m.
nom: say 'foo' ~~ /f & \w & \D/ 11:36
p6eval nom 8c69ed: OUTPUT«=> <f>␤␤»
diakopter nom: say 'foo' ~~ /(?:f & \w & \D)/ 11:37
p6eval nom 8c69ed: OUTPUT«===SORRY!===␤Quantifier quantifies nothing at line 1, near "f & \\w & \\"␤»
diakopter std: say 'foo' ~~ /(?:f & \w & \D)/
p6eval std be1f10e: OUTPUT«===SORRY!===␤Quantifier quantifies nothing at /tmp/vNaO1tx0KX line 1:␤------> say 'foo' ~~ /(?:⏏f & \w & \D)/␤Check failed␤FAILED 00:01 123m␤»
jnthn diakopter: oh
'bbccdd' ~~ /<-[d..b]>/ 11:38
diakopter (?: needs a better error
jnthn It's that one
One that should die anyway.
iirc
std: 'bbccdd' ~~ /<-[d..b]>/
p6eval std be1f10e: OUTPUT«ok 00:01 123m␤»
jnthn hm
Is that death or just "never match"?
nom: say 'bbccdd' ~~ /<-[d..b]>/ 11:39
p6eval nom 8c69ed: OUTPUT«=> <b>␤␤»
diakopter nom: say 'bbccdd' ~~ /<-[d..b]>+/
p6eval nom 8c69ed: OUTPUT«=> <bbccdd>␤␤»
diakopter niecza: say 'bbccdd' ~~ /<-[d..b]>+/
p6eval niecza v11-24-gcdd4e66: OUTPUT«#<match from(0) to(6) text(bbccdd) pos([].list) named({}.hash)>␤»
jnthn hm :) 11:40
here's what std says
A reversed range is illegal. In directly compiled code it's a compile-time
error to say
<[ z .. a ]> / # Reversed range is not allowed
diakopter heh 11:41
masak aye.
jnthn So we should die just...not so badly :)
lolitsmasak!
And the test should be dies_ok
diakopter why do I always read lolitamasak
jnthn ouch! 11:42
"keys right next to each other" is usually an input problem :P
11:45 danishman joined
diakopter jnthn: how long does your nqpq take to rebuild after changing .nqp 11:45
11:45 whiteknight joined
dalek ast: 1d2cac2 | jnthn++ | S05-mass/rx.t:
Fix test for reversed character ranges.
 11:45
11:46 whiteknight is now known as Guest43790 11:47 danishman left
jnthn diakopter: Well, here it goes and builds the entire nqp bootstrap again...so at least 30s... 11:47
diakopter hm, mine takes a full minute just to compile QRegex.nqp 11:48
jnthn diakopter: Seems that the negated escaped hyphen tests somehow got broken too
diakopter ah, that probably needs a special case too
jnthn nom: say '---x--' ~~ /<-[\-]>/
p6eval nom 8c69ed: OUTPUT«=> <x>␤␤»
jnthn > '---x--' ~~ /<-[\-]>/
=> <->
d'oh :)
diakopter I bet I can fix that 11:49
jnthn ok 11:50
jnthn is just adding in a decent error for the reversed char ranges
dalek p: 1b061e8 | jnthn++ | src/QRegex/P6Regex/Actions.nqp:
Catch illegally reversed character ranges and die with a good error.
 11:53
p: af28c73 | diakopter++ | src/QRegex/P6Regex/ (2 files):
allow escaped hyphens too
 11:55
p: 6ad13cc | diakopter++ | src/QRegex/P6Regex/Actions.nqp:
Merge branch 'master' of github.com:perl6/nqp
diakopter hm how do I do the rebase thing
jnthn git pull --rebase
diakopter that'll avoid the merge branch commit? 11:56
jnthn yes
diakopter: testing that patch :)
diakopter thanks 11:57
jnthn: I think \- accidentally worked previously 12:00
jnthn oh :)
I can believe that :)
diakopter that is, the \ was interpreted as one of the possible (or not possible) chars
as well as the \ or -
er
jnthn yes, seems that your patch helped \o/
diakopter you know what I mean
jnthn doing full spectest run then will bump NQP_REVISION. 12:01
moritz: With the bump I'm about to do, the JSON::Tiny::Grammar's str rule can start using the commented out "<-["\\\t\n]>+" case 12:03
(rather than all the <before ...>
12:03 Logicus joined
jnthn moritz: It's...rather faster ;) 12:04
diakopter jnthn: in your sleep did you think about regex interpolation? 12:08
jnthn diakopter: I realized that we already have it working in Rakudo and that it actually doesn't need any methods in the common role or engine changes. It just needs a couple of methods writing in NQPCursor. 12:09
diakopter: I can take a crack at that today.
diakopter oh..
Logicus I woke up to a dream about Perl 6 this morning lol
that's what happens when you read about something last thing at night
diakopter bed& 12:10
masak diakopter: 'night. dream of interpolation in regexes :)
jnthn diakopter: sleep well o/ 12:11
dalek kudo/nom: b312db8 | jnthn++ | tools/build/NQP_REVISION:
Bump to an NQP with support for escape sequences in character classes, plus other fixes and improvements. diakopter++.
masak Logicus: two questions: was it a happy dream, and will it eventually lead to a spec change? :P 12:12
12:12 Psyche^ joined
Logicus two answers: yes it was very pleasant, and I don't know enough about Perl6 yet to say anything meaningful about specs 12:13
masak :) 12:14
fsergot nom: say "a::b::::d".split("::").perl
p6eval nom 8c69ed: OUTPUT«("a", "b", "", "d").list␤»
Logicus in the dream I was learning how to put some sort of Perl6 statement together, and I was using this interactive graphical learning console which told me I had the "moon" character in the wrong place
whatever that means.
masak sometimes the effect newcomers have is to set off a chain reaction that leads to a spec change.
Logicus: oh no, the "moon" character in the wrong place! o.O
Logicus as long as it's not a chain reaction which destroys the station that's all good
masak "that's not a moon character!" 12:15
Logicus It's too big to be a moon charachter
masak .u ⺝
phenny U+2E9D CJK RADICAL MOON (⺝)
12:16 Patterner left, Psyche^ is now known as Patterner
masak .u 月 12:17
phenny U+6708 CJK UNIFIED IDEOGRAPH-6708 (月)
masak phenny: "月"?
phenny masak: "Month" (zh-CN to en, translate.google.com)
masak close enough :)
phenny: "月亮"?
phenny masak: "Moon" (zh-CN to en, translate.google.com)
masak ah, of course.
jnthn moritz: Please can I have a commit bit for JSON::Tiny?
moritz jnthn: sure 12:18
hugme: add jnthn to json
hugme hugs jnthn. Welcome to json!
masak rule in Mandarin: the word is usually two characters, even if the first character would've been enough in Classical Chinese.
Logicus and the argonauts!
moritz jnthn: feel free to commit the simplification to the regex rule
and I guess .trans can be replaced by .subst or so 12:19
or even s///
masak Logicus: best way to learn how to avoid putting the "moon" character in the wrong place is to learn about the most basic rule of Perl 6 parsing: that of two-terms-in-a-row being disallowed.
moritz now that that works
masak Logicus: in some languages you can write 'obj.method 1, 2, 3' and it means 'obj.method(1, 2, 3)'. not so in Perl 6. 12:20
jnthn moritz: Also the .item workaround in the actions is no longer needed :)
moritz jnthn: \o/
masak Logicus: because both '$obj.method' and '1' are terms, and they can't follow one another without an intervening operator.
moritz jnthn: anything that simplifies JSON::Tiny is appreciated, because it also simplifies the book :-) 12:21
masak Logicus: it's like the Perl 6 parser is always in a mode where it expects a term or expects an operator, and it flips back and forth between these two.
jnthn moritz: It's not so much a simplification, but it turns out special-casing the string action method for the "single string with no escapes" case is a > 10% win. 12:24
Logicus ok 12:25
jnthn Even without that, from-json on projects.json now runs in half the time that it did yesterday 12:26
felher When you define a function in common lisp and some default value for an argument X , you can specify another variable (x-supplied-p mostly) that is set to NIL if the argument X was user-supplied, T if X was set to its default value. Is something like that possible in Perl6? 12:27
jnthn moritz: method value:sym<number>($/) { make eval $/.Str }
moritz: Is +$/.Str not good enough these days?
arnsholt felher: I'd probably just do that with two different multis, one with and one without the argument 12:28
masak Logicus: so something like 'say $*OUT "OH HAI"' is also disallowed in Perl 6. both '$*OUT' and '"OH HAI"' are terms. 12:29
12:29 rindolf joined
rindolf Hi all 12:29
masak hi rindolf
rindolf was finally able to checkout Rakudo nom from github. There were networking problems. 12:30
masak: hi, what's up?
masak rindolf: a bit too many things at the moment :) it will get better soonish, though. 12:31
rindolf: how're you?
felher arnsholt: good point, but that may lead to quite a duplication of code sometimes :)
rindolf masak: I'm fine. Yesterday I went over my RSS feeds. 12:32
masak: and I worked on some CPAN modules. 12:33
Logicus so everything is expressed as either a term or an operator?
rindolf masak: I blogged about the upcoming Tel Aviv Perl Mongers meeting, and have some blog posts in the pipe.
Logicus I'm guessing we select $*OUT first, then say to it ?
masak literal: you could do it as $*OUT.say("OH HAI") or $*OUT.say: "OH HAI" or say $*OUT: "OH HAI" 12:34
rindolf masak: one of them is about the rt.cpan.org anti-pattern of someone filing a bug on one of my modules, I'm asking for some input, I wait many weeks, they don't reply, and I close the bug due to inactivity.
It happened to me many times.
felher Hm, maybe somthing like: 'role IsDefault {}; ... ... ... sub foo($a = 3 but isDefault) { ... $a ~~ isDefault ...}'. Is that a legitimate way for using roles?
masak Logicus: depending on what you mean by "everything", then yes, everything is either a term or an operator :) 12:35
Logicus: of course, the picture gets more detailed the closer you look. for example, prefix operators are OK in term position. things like -42 or ?$value
Logicus ok, everythings a fractal really... 12:36
masak yes :)
Logicus: also, things that look like terms (such as "say") are really subroutine listops (list operators), and they throw the parser back into expect-term mode. that's why "say 42" isn't a parse error. 12:37
s/things/some things/
Logicus so because it's expecting a term, 42 by itself without quotes is taken to be a term 12:38
implicitly 12:39
masak 42 is a term, period.
but "say" isn't, and that's the point.
so it's not two terms in a row.
Logicus but it must be preceded by a listop
? 12:40
masak right. the listop "resets" the parser mode.
Logicus: also, sometimes the parser "cheats" and accepts TTIAR because it's part of a special form in the language. for example C< if 42 { say "OH HAI" } > would've been TTIAR (the 42 and the block are both terms), if it weren't for "if" which is a special form.
Logicus TTIAR? 12:41
masak two terms in a row :)
Logicus Term Term...
masak right.
Logicus Two Terms In A... ok
masak this concludes my light introduction to parsing in Perl 6.
I like this stuff. I once wrote a Perl 6 parser in Perl 6. 12:42
Logicus good stuff thanks :)
I love parsers :)
I've even been known to be sat playing around with parsing on xmas morning :P 12:43
12:43 MayDaniel left 12:47 tokuhirom left
masak xmas morning is a surprisingly good time for hacking. 12:48
Logicus perhaps today is a good day to hack! 12:53
jnthn nom: my $x = 42; my $y = "$x"; say $y.WHAT
p6eval nom b312db: OUTPUT«Int()␤»
jnthn b: my $x = 42; my $y = "$x"; say $y.WHAT
p6eval b 1b7dd1: OUTPUT«Str()␤»
Logicus nom: my $x = 42; my $y = $x; say $y.WHAT 12:54
p6eval nom b312db: OUTPUT«Int()␤»
Logicus so the quotes didn't make any difference 12:55
jnthn Logicus: Yeah, it's a bug... 12:56
Logicus so in the correct final version the quotes are required?
jnthn well, an ex-bug given I've patched it locally
masak Logicus: no, both $x and "$x" are fine.
Logicus: but "$x" is a Str, not an Int. 12:57
moritz ... at least it should be
Logicus ah right of course, so putting the quotes should coerce it
masak in Pure, Ideal Perl 6 :)
moritz let's call that PIP6 :-)
masak Logicus: yes, it's a circumfix coercion :)
jnthn Oddly, found it when I was optimizing match object construction :)
12:57 Trashlord left
masak moritz: deal :) 12:57
Logicus circumfix... hrm sounds dodgy... like circumfixtion or something 12:58
masak jnthn: because no sane person would use that way to coerce Int->Str? :P
Log
Logicus im a long way from sane :)
masak Logicus: well, that would be being nailed to a pair of matching parentheses or something :)
jnthn masak: Nah, 'tis just a bug :)
Logicus no I was thinking more along the lines of a certain popular jewish operation 12:59
12:59 Trashlord joined
masak jnthn: right, just trying to motivate why it's been long unfound. 12:59
Logicus: ah, that would be stripping of the outermost pair of matching parentheses, then :P
Logicus ouch!
masak could be wrose. the outermost parens don't actually do much. 13:01
worse*
rindolf Can anyone put a tarball of git://github.com/perl6/roast.git on a fast HTTP server, so I can download it quickly? github is SLOW.
Logicus true, but I think they would be best left exactly where they are.
jnthn nom: say 2.205 / 6.117
p6eval nom b312db: OUTPUT«0.360470819028936␤»
masak they're just for grouping :P
jnthn tadzik: projects.json should now get parsed in about a third of the time :) 13:03
13:04 Trashlord left, Trashlord joined
moritz rindolf: once? or regularly? 13:06
rindolf moritz: once.
13:08 Trashlord left 13:09 Trashlord joined
rindolf Though regularly would also be a good idea. 13:10
moritz rindolf: moritz.faui2k3.org/tmp/roast.git.tar
dalek kudo/nom: 0786d96 | jnthn++ | src/Perl6/Actions.pm:
Fix an interpolation bug.
 13:11
kudo/nom: 0ad267e | jnthn++ | src/core/Cursor.pm:
Optimize match object construction somewhat.
jnthn moritz: Interesting. from-json spends less than 20% of its time actually parsing. The other 80% is constructing the match objects and running the actions. 13:13
rindolf moritz: thanks, but git checkout there didn't work.
13:15 Trashlord left
arnsholt rindolf: It looks like a tar archive of the repo 13:18
rindolf arnsholt: no, it's a tar archive of the dot-git (".git") directory. 13:19
13:19 Trashlord joined
rindolf arnsholt: I don't know how to get the working copy. 13:19
13:19 Trashlord is now known as Guest56160
arnsholt Put it in an empty directory, rename to .git and then git checkout master should do the trick I think 13:20
rindolf Got it working now. 13:25
13:29 dpress joined, Guest56160 left 13:32 Guest56160 joined 13:36 Lothar left, yath left, fhelmberger left, Woodi left, japhb left, mls left, zamolxes left, risou_awy left, jevin left, revdiablo left 13:37 ilogger2 joined, ChanServ sets mode: +v ilogger2, snarkyboojum_ joined 13:38 stepnem joined 13:41 jevin joined, Guest56160 joined
rindolf Hi all. I'm getting these make spectest failures - paste.debian.net/146373/ 13:42
13:50 Guest56160 left, Guest56160 joined 13:51 Mowah_ joined
jnthn rindolf: The regex stuff is having a load of work done at the moment. Maybe you've just got a slightly inconsistent Rakudo and spectest version. 13:52
rindolf jnthn: OK. 13:54
13:56 Guest56160 left
dalek atures: be7c71f | jnthn++ | features.json:
Rakudo handles backslash escapes in char classes now.
 13:56
atures: f387311 | jnthn++ | features.json:
Rakudo nom also has some Buf support.
 13:58
13:58 dorlamm joined 14:01 nightlord joined 14:02 Guest56160 joined 14:07 fsergot joined
spine hi there. Is there anything simple in perl6 to Serialize my data structures to disk, like the Storable module from perl5? 14:11
Or do I have to write my own serialization code with something like Simple::Redis?
14:13 mj41 joined 14:17 Guest56160 left, kaare_ joined
masak spine: there's .perl, which may or may not be enough for your needs. 14:18
14:18 vivek joined, vivek left
masak rakudo: say [1, 2, { a => [5, 42] }].perl 14:19
p6eval rakudo 0ad267: OUTPUT«[1, 2, {"a" => [5, 42]}]␤» 14:20
14:20 Guest56160 joined
dalek ast: 3f97578 | jnthn++ | S05-mass/rx.t:
Unfudge more lookahead/lookbehind tests.
 14:21
spine masak: wow that looks great thanks a lot. So I can just store that to disk and later read it in with eval :) 14:22
arnsholt nom: q{"}.perl.say 14:24
p6eval nom 0ad267: OUTPUT«"\""␤»
dalek ast: 7dbc061 | jnthn++ | S05-mass/rx.t:
Another bunch of unfudges.
 14:25
14:25 Guest56160 left
rindolf jnthn: now I'm getting this after an update - paste.debian.net/146381/ 14:33
14:43 kfo joined 14:46 Guest56160 joined
Logicus is eval going to be as slow in p6 as in p5? 14:46
Woodi Logicus: probably eval will be fast becouse it's noop :) 14:47
masak eh? 14:48
Woodi non-existen i mean :)
masak nom: eval "say 'OH HAI'"
p6eval nom 0ad267: OUTPUT«OH HAI␤»
jnthn rindolf: hM, INTERESTING.
oops :)
Woodi ok, maybe i am wrong
masak maybe... :)
Woodi i thinked grammars replace it
jnthn rindolf: Please could you run t/spec/S03-operators/range.rakudo and give me the output? 14:50
14:50 Moukeddar joined, benabik joined
rindolf jnthn: yes. 14:53
jnthn: paste.debian.net/146383/ 14:54
dalek ast: d957740 | jnthn++ | S02-literals/ (2 files):
Untodo some tests we now pass; spotted by rindolf++.
 14:55
rindolf Heh, thanks for the ++ 14:56
:-)
jnthn: :-)
jnthn rindolf: I can reproduce that range fail here too
rindolf jnthn: OK. Good.
dalek p: a6a4c80 | jnthn++ | src/QRegex/P6Regex/Grammar.nqp:
Tweak an error message to have a missing space.
 14:58
15:02 Guest56160 left 15:04 Guest56160 joined
dalek kudo/nom: ac24a6d | jnthn++ | src/core/Cursor.pm:
Eliminate a pir:: and a method call.
 15:07
kudo/nom: f3467e0 | jnthn++ | tools/build/NQP_REVISION:
Get corrected error.
ast: 6242705 | jnthn++ | S05-metasyntax/charset.t:
Correct and unfudge a test for old char range syntax.
15:08 kaare_ left 15:10 kaare joined 15:11 kaare is now known as Guest40733 15:13 Guest40733 left 15:14 nightlord left 15:15 Guest56160 left
tadzik 'afternoon 15:18
15:27 dorlamm left 15:29 Moukeddar left 15:34 Guest56160 joined 15:38 Guest56160 left 15:46 MayDaniel joined 15:48 lloh joined, lloh left, jdhore joined
jnthn o/ tadzik 15:52
15:55 kaare joined 15:56 kaare is now known as Guest2721
tadzik t/spec/S06-other/main-usage.rakudo ........................ Failed 6/17 subtests 16:01
jnthn tadzik: That test file has failed for me for a while; I figured it was win32-specific stuff 16:04
tadzik well, I don't think it is, since it fails for me :) 16:06
jnthn Does 16:07
perl6 -e "sub MAIN($bar) { }" --help
give any output for you?
tadzik Usage: 16:08
-e '...'
jnthn oddness
here it gives nothing at all
oh, wait, what...
What about just perl6 -e "sub MAIN($bar) { }" 16:09
tadzik nothing
jnthn Great, we're getting opposite results...
tadzik :>
how about perl6 -e "sub MAIN($bar) { say $bar }"? 16:10
it prints a blank line here
jnthn prints
Usage: -e '...' <bar>
er, line break after Usage:
tadzik yeah
jnthn so, not a blank line 16:11
hmm, wtf...
wolverian I hope you're not running that on a shell that interpolates $ variables inside double quotes
wolverian captain obvious
jnthn er
well, mine doesn't 16:12
tadzik? :)
tadzik haha
should've used ''
now perl6 -e 'sub MAIN($bar) { }' --help doesn't print anything
jnthn ok
that's at least one of the fails then 16:13
nom: say "--help" ~~ /^ ( '--' | '-' | ':' ) ('/'?) (<-[0..9\.]> .*) $/ 16:15
p6eval nom f3467e: OUTPUT«#<failed match>␤»
jnthn nom: say "help" ~~ /^ (<-[0..9\.]> .*) $/
p6eval nom f3467e: OUTPUT«#<failed match>␤»
jnthn nom: say "help" ~~ /^ (<-[0..9]> .*) $/
p6eval nom f3467e: OUTPUT«=> <help>␤ 0 => <help>␤␤»
16:21 MayDaniel left
cognominal jnthn, how to inherit form HLL::Grammar in Perl6? 16:34
nom: class A is HLL::Grammar { }; 16:35
p6eval nom f3467e: OUTPUT«===SORRY!===␤No applicable candidates found to dispatch to for 'trait_mod:<is>'. Available candidates are:␤:(Attribute $attr, Any $rw)␤:(Attribute $attr, Any $readonly)␤:(Routine $r, Any $rw)␤:(Routine $r, Any $default)␤:(Routine $r, Any $info, Any $inlinable)␤:(Param…
cognominal nom: grammar A is HLL::Grammar { };
p6eval nom f3467e: OUTPUT«===SORRY!===␤No applicable candidates found to dispatch to for 'trait_mod:<is>'. Available candidates are:␤:(Attribute $attr, Any $rw)␤:(Attribute $attr, Any $readonly)␤:(Routine $r, Any $rw)␤:(Routine $r, Any $default)␤:(Routine $r, Any $info, Any $inlinable)␤:(Param…
cognominal I meant the second 16:36
jnthn cognominal: Not sure...you'd have to have some kind of "use" statement, but we don't have a :from<nqp> thingy in place yet. 16:37
cognominal ok
jnthn tadzik: Think I've got a fix for the main-usage.t issue 16:39
cognominal this means EXPR and the like are unvailable from perl6
jnthn, meaning somehow encasulating a nqp object to use it from Perl 6? 16:42
jnthn cognominal: I think it's - at least primarily - a module loading issue. 16:45
masak nom: class A is B {}
p6eval nom f3467e: OUTPUT«===SORRY!===␤No applicable candidates found to dispatch to for 'trait_mod:<is>'. Available candidates are:␤:(Attribute $attr, Any $rw)␤:(Attribute $attr, Any $readonly)␤:(Routine $r, Any $rw)␤:(Routine $r, Any $default)␤:(Routine $r, Any $info, Any $inlinable)␤:(Param…
masak hm, that error message could probably be more awesome...
jnthn masak: Not sure about the best way to do it without making other things worse... 16:46
masak: I guess some kinda catch-all-ish multi candidate may help 16:47
cognominal the error displays neither the place of the error nor the signature of the sub. That makes difficult to find the problem 16:51
by place, I mean file and line number. 16:52
jnthn signature of...what sub?
dalek p: 98ef1a1 | jnthn++ | src/QAST/Compiler.nqp:
Compilation of literal nodes ignored negate flag; fix this.
 16:53
p: cc3918c | jnthn++ | src/QRegex/P6Regex/ (2 files):
Turns out that fixing the negation of literals renders the backlit stuff unrequired; remove it.
jnthn tadzik: That should fix your main-usage issue.
cognominal jnthn, here, the trait_mod:<is> signature should involves grammar types here?
jnthn No
tadzik nice 16:54
jnthn It's the (Mu:U, Mu:U) case that would catch it were the type defined
tadzik: And I think I figured out some aspect on the windows problem too
tadzik: It's the :args thing in tests
cognominal sorry, I should have said capture. I mix signature and capture, just the way I do for arguments and parameters :( 16:56
jnthn cognominal: ah, I see what you mean...
cognominal: I'll try adding a candidate to catch this case 16:57
tadzik: heh. The reason it doesn't work is, of all things, missing :i support
dalek kudo/nom: ca161f7 | jnthn++ | tools/build/NQP_REVISION:
Bump to NQP version with fix for \. in the char class.
 17:00
17:10 sayu joined 17:19 sayu left, sayu_ joined 17:24 Guest2721 left 17:27 Moukeddar joined 17:32 orafu joined, pat_js joined
pat_js hi 17:33
What is the recursion max in Rakudo?
jnthn pat_js: Think it's set to around a thousand frames or so 17:38
pat_js jnthn: hmmm, then i might need another algorithm, thanks 17:39
felher nom: sub correct($x) { return True; }; sub func($x where correct) { say $x; }
p6eval nom f3467e: OUTPUT«===SORRY!===␤CHECK FAILED:␤Calling 'correct' will never work with no arguments (line 1)␤ Expected: :(Any $x)␤»
felher Shouldn't that be possible in theory? 17:40
(btw: jnthn++ for merging the protoregex branch :) )
rindolf pat_js: you can use a dedicated stack for recursion.
jnthn felher: I think Rakudo is right here
wolverian 'where &correct'?
jnthn You need...what wolverian said :D 17:41
felher wolverian: jnthn: ah, i see, thnx :)
jnthn wolverian: The RHS of the where acts kinda like a smartmatch.
gah
that was aimed at felher :)
felher jnthn: ah, and you can't match against functions because the function would be evaluated first? So you need the '&'? 17:43
jnthn felher: correct
You'd be evaluating against the return value of the function there.
wolverian felher: otherwise you couldn't match against the return value of a function...
jnthn felher: BTW, Rakudo caught that one at compile time for you ;-)
17:44 skangas joined
felher jnthn: yeah, i know. I watched the optimizer quite a bit, lately. Great work, btw :) 17:44
wolverian: jnthn: thnx for explaining :)
wolverian you're welcome
moritz jnthn++ # JSON::Tiny fixes 17:45
jnthn moritz: If you can find a way to improve string handling there's at least another 25% to be shaved off 17:46
moritz: For example, it's a bit ugly, but checking for the "exactly one match and it was <str>" case and just grabbing and using its ast rather than doing the join stuff gets you about that. 17:47
moritz: But I didn't want to ugly up your nice code ;) 17:48
moritz you mean for strings which don't contain any backslash escapes? 17:50
jnthn moritz: correct 17:51
moritz: It's the join and map that noms up time
moritz jnthn: feel free to commit it anyway
for the book I'll nicify the code in a branch
jnthn moritz: ah, that works too 17:52
:)
17:52 icwiener joined
moritz but for real world usage, 25% speed win is enough to warrant a single 'if' :-) 17:52
jnthn nom: say 1.574 / 2.205 17:58
p6eval nom ca161f: OUTPUT«0.713832199546485␤»
jnthn More like 30%
:)
nom: say 1.574 / 6.117
p6eval nom ca161f: OUTPUT«0.257315677619748␤»
jnthn Runs in a quarter of the time it did when I first got it to work at all :)
masak \o/ 17:59
18:00 sayu joined
jnthn moritz: pushed :) 18:01
18:01 sftp_ joined
moritz \o/ indeed 18:02
jnthn wonders how "fast" b ran it :)
18:03 sayu_ left
[Coke] would be nice to know for something like a blog post when starnom comes out 18:05
18:06 Guest2721 joined
jnthn [Coke]: I'm planning to make some comparrisons for that :) 18:06
moritz jnthn: for the JSON emitter, I was thinking of something like 18:07
[Coke] jnthn: ah, right, not like we can't just rebuild b then. ;)
moritz proto sub to-json(..) is export {*} 18:08
jnthn moritz: Yes, that will work
moritz good
jnthn moritz: I did that locally once
moritz: Then it blocked on trans
moritz great
jnthn hmm...seems I tossed the patch. d'oph.
moritz ah well, probably not hard to redo
jnthn no, it took me about a minute... :) 18:09
moritz jnthn: feel free to add yourself to the CREDITS in json/README 18:12
tadzik++ should do that too 18:13
dalek kudo/nom: 4cff8e5 | jnthn++ | NOMMAP.markdown:
Remove done stuff, de-dupe etc. nommap.
tadzik I don't think I did commit to json
moritz tadzik: you had the "speed up string parsing by a factor of 100" commit (or similar) 18:14
github.com/moritz/json/commit/bca5...24a5e9e4a2 18:15
tadzik oh, that was jnthn's idea :) 18:18
at mberends++' place
moritz ... and implemented by tadzik++
18:24 molaf joined
colomon tadzik: how's panda coming? ;) 18:25
tadzik colomon: it works :) 18:30
colomon \o/
tadzik moritz: no, I just retyped jnthn++'s code :)
colomon tadzik: have you pushed your fixes?
tadzik colomon: si
colomon \o/
tadzik colomon: look at gist.github.com/1379175
it already gave results :) 18:31
but now, I need to hack on minix abit
colomon tadzik: okay, I'll take a look at getting it working with niecza. OH! you've got useful results for nom already! Callooh! Callay! 18:32
benabik Minix?
tadzik aye 18:33
colomon: I updated the wiki too
colomon Woah, we hacked on minix in class in 1991...
tadzik I hack on it on my classes now :) 18:34
18:37 benabik left
colomon tadzik: wiki? 18:39
wait, we've got a Rational branch now? 18:40
apparently everything has happened while I've been busy this weekend, and I'm going to keep on being too busy to track it in all likelihood. :( 18:41
tadzik colomon: perl6/ecosystem wiki 18:42
18:43 edkolev joined
colomon tadzik++ 18:43
tadzik that's not new :) 18:44
18:45 vmspb joined
colomon I didn't know about it, that's good enough for me. 18:45
tadzik Like 18:46
18:46 Chillance joined 18:48 molaf left
jnthn I blogged: 6guts.wordpress.com/2011/11/20/raku...kudo-star/ 19:10
19:12 snarkyboojum_ is now known as snarkyboojum
masak jnth++ 19:15
gah, jnth-- jnthn++
jnthn ;) 19:19
19:19 MayDaniel joined 19:26 icwiener left 19:28 MayDaniel left
felher jnthn++ for blogpost. Great to see so many things improving :) 19:30
19:36 icwiener joined
colomon tadzik: install of panda fails on my 64-bit Linux machine. 19:46
t/04-roundtrip.t .. No applicable candidates found to dispatch to for ''. Available candidates are: 19:48
in block <anon> at t/04-roundtrip.t:37
That's in JSON::Tiny 19:49
djanatyn hmm, I'm getting a weird error which I think might be because something isn't implemented yet, or because of a bug 19:51
or, because my code sucks :)
colomon Those three cases do cover a lot of the possibilities. :) 19:52
djanatyn gist.github.com/1380783 19:53
I'm not sure how self-explanatory my code is, but I can explain what I'm doing if you like 19:54
...hmm, wait a sec'
oh, I didn 19:55
*I didn't initialize anonymous arrays to push to -_-
never mind, fixing
thanks for your help ;) 19:56
tadzik colomon: I know that :/ 19:58
bootstrap won't work
due to JSON and Test::Mock tests failing
you can either install it via ufo, or... tactically remove JSON and Test::Mock tests before they get run
20:02 MayDaniel joined, MayDaniel left
snarkyboojum with ./bootsrap.sh --no-tests or --force? :D 20:04
tadzik (: 20:06
colomon djanatyn: sorry to disappear there, surprise bout of 3-year-old watching on my part. 20:16
20:17 imarcusthis joined
sorear good * #perl6 20:19
colomon o/
20:20 edkolev left 20:30 benabik joined 20:33 ksi joined 20:34 sayu left 20:35 MayDaniel joined
djanatyn does perl5 have auto-vivification? 20:36
tadzik yes, I think 20:38
moritz yes
20:38 MayDaniel left
moritz though with slightly different semantics than in p6 20:38
in particular, read accessess never autoviv in p6 20:39
but in p5, they do (except in the very last level)
jnthn++ # blog 20:40
sorear in p6, autoviv is done at the point of assignment, and it chases backward to autoviv all the necessary containers
benabik last level?
sorear in p5, autoviv is done at the point of the indexing operation
$foo->{'bar'}; # this will do the equivalent of $foo //= {} first in p5 20:41
(it is my opinion that the p5 method works better in practice) 20:42
20:45 PacoLinux_ joined
djanatyn is there a way I can look up all the methods for hashes? 20:47
actually, I'll just ask:
sorear rakudo: say Hash.^methods(:all)».name 20:48
djanatyn I want to figure out how many keys exist for a hash. I know that I can do %foo.keys.elems - is there a cleaner or more readable way?
p6eval rakudo 4cff8e: OUTPUT«too many named arguments: 1 passed, 0 used␤ in methods at src/gen/Metamodel.pm:370␤ in method dispatch:<.^> at src/gen/CORE.setting:726␤ in block <anon> at /tmp/AVKyseFeOg:1␤ in <anon> at /tmp/AVKyseFeOg:1␤»
sorear rakudo: say Hash.^methods».name
p6eval rakudo 4cff8e: OUTPUT«new at_key STORE_AT_KEY STORE delete push _push_construct perl elems exists iterator list keys kv values pairs invert at_key STORE_AT_KEY Capture FLATTENABLE_LIST FLATTENABLE_HASH fmt Bool ACCEPTS perl elems infinite item fmt Int Num chrs Numeric Str abs conjugate …
djanatyn ooh, %foo.elems, thanks 20:49
flussence rakudo: my %a = (a=>1, b=>2, c=>3); say +%a
p6eval rakudo 4cff8e: OUTPUT«3␤»
djanatyn is refactoring and using some cool perl6 tricks
20:51 tokuhirom joined
djanatyn what order does %hash.keys put the keys in the list? order added? 20:52
sorear random order 20:54
benabik hash order, I think.
djanatyn :D
sorear in niecza %hash.keys returns in order-added for all hashes up to 7 elements 20:55
djanatyn anyway, I have a surprise for you guys that's almost finished
benabik sorear: 7?
djanatyn which isn't really that interesting, but kind of cool, I suppose
sorear above 7 elements it switches to hash order
benabik: actually 11 looking at the code again 20:57
benabik sorear: Faster to iterate over a small loop than to bother allocating a hash?
20:57 cooper joined
sorear benabik: exactly 20:57
niecza: say 'a' .. 'z' 20:59
p6eval niecza v11-24-gcdd4e66: OUTPUT«"a".."z"␤»
sorear niecza: say ('a' .. 'z').pick 21:00
p6eval niecza v11-24-gcdd4e66: OUTPUT«c␤»
sorear niecza: say ('a' .. 'z').pick(5)
p6eval niecza v11-24-gcdd4e66: OUTPUT«s f g k p␤»
sorear niecza: my %hash; for ('a' .. 'z').pick(11) { %hash{$_}++; say $_; }; say %hash.keys 21:01
p6eval niecza v11-24-gcdd4e66: OUTPUT«t␤k␤j␤p␤l␤e␤r␤u␤v␤h␤i␤t k j p l e r u v h i␤»
sorear niecza: my %hash; for ('a' .. 'z').pick(12) { %hash{$_}++; say $_; }; say %hash.keys
p6eval niecza v11-24-gcdd4e66: OUTPUT«b␤o␤w␤e␤z␤q␤m␤s␤a␤h␤y␤i␤i y h a s m q z e w o b␤»
sorear \o/
benabik sorear: Is that something you did, or the way the CLR HashMap works? 21:02
benabik is just curious.
sorear benabik: something I did
21:02 HarryS joined
sorear it's kind of sad that someone with my qualifications can write a significantly faster HashMap than the builtin one 21:02
Moukeddar it's funny 21:04
21:04 pat_js left
benabik You have excellent qualifications. You're the author of one of the leading implementations of Perl 6. :-) 21:07
Moukeddar guys, how do you benchmark you collections? 21:09
sorear I have a script in tools/ that runs a sub a million times, averages the time, and makes a crude attempt to compensate for systematic error 21:12
diakopter benabik: is there a VTABLE_push_PMC or how is it spelled 21:13
jnthn diakopter: VTABLE_push_pmc
diakopter jnthn: or you
oh
sorear I was using a benchmark like sub () { my %hash; %hash<a> = 1; %hash<b> = 2; %hash<c> = 3 }
then I confirmed that it did, in fact, make a measurable speedup on niecza's self-compilatin time 21:14
this was a few versions ago when niecza's STD was still using @*MEMOS[$/.pos]<L> to implement 'lineof', which requires 200,000 one-element hashes when compiling a big file like CORE.setting or STD.pm6 21:15
jnthn sorear: Do you still have @*MEMOS at all? 21:16
sorear I've since changed it to something more like @*LINEMEMOS[$/.pos], reducing the need for performant one-element hashes
jnthn: yes
moritz nom: say so '"' ~~ /<[\c32 .. \c126]>/
p6eval nom 4cff8e: OUTPUT«Bool::False␤»
moritz nom: say so '"' ~~ /<-[\c32 .. \c126]>/
p6eval nom 4cff8e: OUTPUT«Bool::True␤»
moritz nom: say ord '"'
p6eval nom 4cff8e: OUTPUT«34␤»
moritz nom: say ord '\'
p6eval nom 4cff8e: OUTPUT«===SORRY!===␤Confused at line 1, near "say ord '\\"␤» 21:17
moritz nom: say ord '\\'
p6eval nom 4cff8e: OUTPUT«92␤»
jnthn moritz: Ah...the case of backslash sequences and range may not be implemented yet...
diakopter: ^^ if you're bored ;-) 21:18
moritz nom: say so '"' ~~ /\ ..~/ 21:19
p6eval nom 4cff8e: OUTPUT«===SORRY!===␤Confused at line 1, near "say so '\"'"␤»
Moukeddar interesting
moritz nom: say so '"' ~~ /<-[\ ..~]>/
p6eval nom 4cff8e: OUTPUT«Bool::True␤»
Moukeddar thanks for the tips ;)
moritz erm 21:20
that's not correct either, is it?
jnthn er, how does that one parse... P)
:)
moritz: Looking at it - the .. handling certainly only supports having an actual character each side. 21:21
moritz char class from space (codepoint 32) to ~ (codepoint 126)
nom: say chr 35
p6eval nom 4cff8e: OUTPUT«#␤»
jnthn moritz: yes, that's how it should be parsed :)
moritz: I'm saying I don't think it's happening that way.
sorear jnthn: @*MEMOS is useful! but having a memo on _every_ character position is just wrong. 21:22
djanatyn okay, let's see if this works
moritz commits a patch that makes JSON::Tiny not use .trans... but it comes out wrong anyway :( 21:24
jnthn sorear: Yeah, Pm didn't do an @*MEMOS, so was curious if Niecza kept it :)
djanatyn here we go! :D 21:27
djanatyn.github.com/2011/11/19/pars...hdays.html
a blog post about how I used perl6 21:28
21:28 fsergot left
djanatyn grr, gotta fix a few things 21:30
there, now the example of the ICS file has it's own seperate lines 21:32
jnthn djanatyn++
djanatyn jnthn: ?
why are you incrementing me? o_o
diakopter karma djanatyn 21:33
aloha djanatyn has karma of 4.
djanatyn :D
hooray
djanatyn takes a shower, brb
jnthn djanatyn: It's a #perl6 way of people showing they like something somebody else did :)
djanatyn well, thank you :) 21:34
#perl6++
diakopter karma perl6 21:35
aloha perl6 has karma of 9.
jnthn karma diakopter
aloha diakopter has karma of 486.
jnthn :)
diakopter karma jnthn
aloha jnthn has karma of 2187.
diakopter karma moritz 21:36
aloha moritz has karma of 2732.
diakopter karma TimToady
aloha TimToady has karma of 250.
diakopter karma lwall
aloha lwall has karma of 0.
diakopter oh
21:37 Guest2721 left
tadzik karma tadzik 21:42
aloha tadzik has karma of 832.
tadzik wow, it was 666 not that long ago :)
jnthn Clearly you've been hacking like a beast since then :) 21:43
masak .oO( hacking like baest... )
vmspb Hi 21:44
This is correct for Ideal Perl 6?
rakudo: say "\x4_0", "\c6_4", "\o1_0_0"
p6eval rakudo 4cff8e: OUTPUT«@@@␤»
vmspb I think the current STD grammar finds "\c6_4" as "\c6" ~ "_4", in contrast to "\x4_0" and "\o1_0_0" 21:45
May need to be replaced s/\d+/<decint>/ in STD.pm6#L415
21:45 mj41 left
djanatyn is there anything new on perl6 + google code in yet? 21:51
ah, it opens tommorow :D 21:52
22:07 fsergot joined 22:14 icwiener left 22:16 MayDaniel joined 22:17 mj41 joined 22:18 amkrankruleuen joined 22:19 PacoLinux_ is now known as PacoLinux, Moukeddar left 22:23 Trashlord joined
sorear o/ MASAK 22:27
22:27 PacoLinux left 22:28 cooper left
sorear or masak, if your highligher is case sensistive. 22:31
colomon :) 22:35
22:41 MayDaniel left 22:46 icwiener joined
masak 'night, #Perl6 22:46
22:46 daniel-s joined
sorear :/ 22:52
I wanted to finish the discussion on promoting niecza we started the other day
22:54 mj41 left, icwiener_ joined, icwiener left, PacoLinux joined
diakopter ditches gedit for scite 23:01
23:02 icwiener_ left 23:08 Moukeddar joined 23:12 cooper joined, vmspb left 23:17 PacoLinux left
jnthn 'night, #perl6 23:18
23:29 Moukeddar left 23:53 woosley joined, woosley 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!