»ö« | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, alpha:, pugs:, std:, or /msg p6eval perl6: ... | irclog: irc.pugscode.org/ | UTF-8 is our friend!
Set by moritz_ on 25 June 2010.
TimToady rakudo: say (1...10).map(* * *).map(* * 10) 00:00
p6eval rakudo e58b16: OUTPUT«20120300560900␤»
TimToady rakudo: say ~(1...10).map(* * *).map(* * 10)
p6eval rakudo e58b16: OUTPUT«20 120 300 560 900␤»
TimToady that's also quite wrong 00:01
oh, way
*wait
rakudo: say ~(1...10).map(* ** 2).map(* * 10)
p6eval rakudo e58b16: OUTPUT«10 40 90 160 250 360 490 640 810 1000␤»
TimToady no, it's fine, it's my brane is rwong 00:02
00:02 bjarneh joined
jnthn I think it's more a bug in the iterator impl 00:02
Likely oversight/unchecked condition than anything fundemental.
But too tired to trace it now
00:03 eiro left
TimToady rakudo: say (1 ... 10).grep({ $_ %% 2}).grep(*) 00:03
p6eval rakudo e58b16: ( no output )
TimToady rakudo: say ~(1 ... 10).grep(*).grep(*)
p6eval rakudo e58b16: ( no output ) 00:04
TimToady rakudo: say ~(1 .. 10).grep(*).grep(*)
sorear is my *$a := foo; the best way to get the head of a lazy list in P6?
00:04 eiro joined
p6eval rakudo e58b16: ( no output ) 00:04
TimToady well, if you don't want the tail, maybe
also, probably fails with too many args 00:05
sorear I'm thinking about the correct equivalent code for if (my ($C) = $C->foo) { .. }
TimToady that looks vaguely familar :) 00:06
how 'bout foo[0]
hmm, might be heavy 00:07
well, my ($a) = foo might just still work
or possibly my ($a,*@) = foo will do 00:08
since the my makes a sig in p6 00:09
jnthn I hope you wanted a := :-)
But yes, I think that'd work
slurpies are lazy
TimToady ((my $a),*) = foo to force nonsig
though probably nyi
jnthn my ($a,*@) = foo will parse it as a sig but then whine that it's too complex to be turned into a plain old list 00:10
TimToady rakudo: say (my $a) = 1,2,3; say $a
p6eval rakudo e58b16: OUTPUT«Any()␤»
jnthn huh.
TimToady thats...bogus
jnthn That's very odd
TimToady rakudo: my $a; say ($a) = 1,2,3; say $a 00:11
lue only one say got interpreted, it seems.
p6eval rakudo e58b16: OUTPUT«Any()␤»
TimToady rakudo: my $a; ($a) = 1,2,3; say $a
lue [timtoady's first try]
p6eval rakudo e58b16: OUTPUT«1 2 3␤»
TimToady rakudo: (my $a) = 1,2,3; say $a
p6eval rakudo e58b16: OUTPUT«1 2 3␤»
TimToady rakudo: ((my $a),*) = 1,2,3; say $a 00:12
p6eval rakudo e58b16: OUTPUT«1␤»
TimToady there you go
rakudo: (my $a,*) = 1,2,3; say $a
p6eval rakudo e58b16: OUTPUT«1␤»
sorear TimToady: not planning to actually generate Perl 6 for input to Rakudo
instead I'm plotting niecza's regex compiler 00:13
TimToady well, tmtowtdi
ingy rakudo: class C { has $.v is rw }; my $o = C.new; $o.v = "a\nb\n"; my $s = $o.v; $s ~~ s/\n$// 00:16
00:16 Guest48289 left, ispeak_ joined
p6eval rakudo e58b16: ( no output ) 00:16
ingy rakudo: class C { has $.v is rw }; my $o = C.new; $o.v = "a\nb\n"; my $s = $o.v; $s ~~ s/\n$//; say $s; 00:17
p6eval rakudo e58b16: OUTPUT«a␤b␤»
TimToady try with $$ instead 00:18
unless, of course, that's not what you want :) 00:19
00:20 ispeak_ left
ingy jnthn: have you seen: Null PMC access in find_method('new') 00:20
jnthn Not for a long time.
00:20 chitragupt left, [Coke] left, gfldex left, LionMadeOfLions left, rokoteko left, cosimo left
ingy jnthn: IIIII 00:22
sorry
TimToady this is #perlIIIIII
ingy jnthn: I get it on a $s ~~ /...//
TimToady: :P 00:23
sorear Perl IX
00:23 eiro left
TimToady is that like Ice IX? 00:23
turns the whole world into a Perl cube? 00:24
00:24 frew joined 00:25 [Coke] joined, rokoteko joined
ingy rakudo: my $s = *; $s ~~ s/...//; 00:26
p6eval rakudo e58b16: ( no output )
ingy rakudo: my $s = *; $s ~~ s/...//; say $s 00:27
p6eval rakudo e58b16: OUTPUT«!whatever_dispatch_helper␤»
ingy rakudo: my $s = *; say $s
p6eval rakudo e58b16: OUTPUT«!whatever_dispatch_helper␤»
TimToady DIHWIDT
ingy rakudo: say *
p6eval rakudo e58b16: OUTPUT«!whatever_dispatch_helper␤»
00:27 chitragupt joined
jnthn So it's doing *.Str which gets curried 00:27
TimToady .Str isn't supposed to curry * 00:28
.Stringy is
sorear rakudo: say *.Str
p6eval rakudo e58b16: OUTPUT«_block65␤»
00:28 cognominal joined
TimToady just as .Str isn't supposed to thread junctions 00:28
ingy is lost 00:29
TimToady .Str is a low-level "turn this into a single string no matter what"
00:29 Pravin joined 00:30 Pravin left
TimToady ~$x is the same as .Stringy, which says, 'do stringification', and can autothread over junctions 00:30
00:30 cosimo joined
sorear .Stringy turns an object into a Str, Buf, or Cat !?!? 00:30
TimToady say (1|2).Str
rakudo: say (1|2).Str
p6eval rakudo e58b16: OUTPUT«any(1, 2)␤»
TimToady rakudo: say ~(1|2)
p6eval rakudo e58b16: OUTPUT«any(1, 2)␤»
TimToady which is wrong 00:31
00:31 mtnviewmark joined
TimToady should say any("1","2") 00:31
rakudo: say ~(1|2).Stringy
p6eval rakudo e58b16: OUTPUT«Method 'Stringy' not found for invocant of class 'Int'␤ in main program body at line 1␤»
TimToady rakudo: say (1|2) ~ "" 00:32
p6eval rakudo e58b16: OUTPUT«any("1", "2")␤»
TimToady liek that, only unary :)
*ke
sorear std: my $x is smiling;
p6eval std 31610: OUTPUT«ok 00:01 111m␤»
sorear rakudo: my $x is smiling;
TimToady std doesn't check traits at all
p6eval rakudo e58b16: OUTPUT«No applicable candidates found to dispatch to for 'trait_mod:<is>'. Available candidates are:␤:(Mu $child, Role $r)␤:(Routine $r, Any :default($default)!)␤:(Code $block, Any $arg?, Any :export($export)!)␤:(Mu $child, Mu $parent)␤:(Mu $type where ({ ... }), Any :rw($rw)!)␤:(Mu
..$type…
00:33 mtnviewmark left
sorear wait, how did rakudo even get that far 00:33
I thought 'my $x is smiling' did a name lookup on ::smiling
before calling any multis
TimToady only to see if it's a known type
if not, it must be a trait that is handled by a compile-time dispatch 00:34
sorear what if it is a known type?
TimToady then it's an "isa"
sorear my $x is Foo => Foo is container type 00:35
my $x is cow => trait_mod:<is>($declarand, :cow)
?
TimToady or that, yes, on a non-class
rakudo: my $x is Scalar of Int = 42; say $x
ingy what's a more direct way to do subst besides ~~ s/// ?
p6eval rakudo e58b16: OUTPUT«No applicable candidates found to dispatch to for 'trait_mod:<is>'. Available candidates are:␤:(Mu $child, Role $r)␤:(Routine $r, Any :default($default)!)␤:(Code $block, Any $arg?, Any :export($export)!)␤:(Mu $child, Mu $parent)␤:(Mu $type where ({ ... }), Any :rw($rw)!)␤:(Mu 00:36
..$type…
ingy because the ~~ is screwed up
TimToady rakudo: my @x is Array of Int = ^42; say ~@x
p6eval rakudo e58b16: OUTPUT«Modifications to classes are not allowed after instantiation.␤ in 'trait_mod:<is>' at line 61:CORE.setting␤ in main program body at line 20:/tmp/lnm3f03urY␤»
TimToady um 00:37
lue don't quote me, but I think s/// is described in S05
TimToady rakudo: my $x = 'abc'; $x.=substr(/b/,'r'); say $x 00:38
rakudo: my $x = 'abc'; $x.=subst(/b/,'r'); say $x
p6eval rakudo e58b16: ( no output )
rakudo e58b16: OUTPUT«arc␤»
00:38 LionMadeOfLions joined
TimToady rakudo: my $x = 'abc'; say $x.subst(/b/,'r'); 00:38
p6eval rakudo e58b16: OUTPUT«arc␤»
TimToady that kind of more direct?
ingy rakudo: my $x = 'abc'; $x.subst(/b/,'r'); say $rx 00:40
rakudo: my $x = 'abc'; $x.subst(/b/,'r'); say $x
p6eval rakudo e58b16: OUTPUT«===SORRY!===␤Symbol '$rx' not predeclared in <anonymous> (/tmp/nGyLgWWCRw:20)␤»
rakudo e58b16: OUTPUT«abc␤»
lue
.oO(~~ is smart match, so I guess it should deal with regex matches too. A pun in P6 is just waiting.)
ingy rakudo: my $x = 'abc'; $x.=subst(/b/,'r'); say $x
p6eval rakudo e58b16: OUTPUT«arc␤»
ingy TimToady: yeah
Null PMC access in find_method('new') was caused by ~~s/// but .=subst() worked fine :( 00:42
lue rakudo: my $a = Regex.new(<alpha>) 00:44
p6eval rakudo e58b16: OUTPUT«too few positional arguments: 2 passed, 3 (or more) expected␤ in main program body at line 1␤»
lue rakudo: my $a = /<alpha>/; say $a.WHAT 00:47
p6eval rakudo e58b16: OUTPUT«Regex()␤»
lue rakudo: my $a = /<alpha>+/; my $b = "!! smart" ~~ $a; say "$\ match" 00:49
p6eval rakudo 692aa1: OUTPUT«===SORRY!===␤Unsupported use of $\ variable; in Perl 6 please use the filehandle's :ors attribute at line 20, near " match\""␤»
lue rakudo: my $a = /<alpha>+/; my $b = "!! smart" ~~ $a; say "$b match"
p6eval rakudo 692aa1: OUTPUT«smart match␤»
TimToady decommuting &
lue
.oO(the pun needs a little work, but at least it's out there.)
 00:51
00:54 jaldhar joined
ingy is 'this' a p6 keyword? 00:55
lue afk 00:58
00:58 jaldhar left
sorear ingy: yes, but it's spelled 'self' 01:00
01:18 meppl left
ash_ rakudo: class Foo { has $.a; method bar { self.a } }; say Foo.new(:a(10)).bar; 01:30
p6eval rakudo 692aa1: OUTPUT«10␤»
awwaiid rakudo: class Foo { has $.a; method bar { .a } }; say 02:09
p6eval rakudo 692aa1: OUTPUT«␤»
awwaiid er 02:10
rakudo: class Foo { has $.a; method bar { .a } }; say Foo.new(:a(10)).bar;
p6eval rakudo 692aa1: ( no output )
ash_ rakudo: class Foo { has $.a; method bar { $!a } }; say Foo.new(:a(10)).bar; 02:13
p6eval rakudo 692aa1: OUTPUT«10␤»
ash_ rakudo: class Foo { has $.a; method bar($this:) { $this.a } }; say Foo.new(:a(10)).bar; # if you want to use this instead of self... 02:15
p6eval rakudo 692aa1: OUTPUT«10␤»
02:15 gfx joined
lue I'm guessing string encoding standards (Unicode, Shift-JIS, etc.) comes from Parrot? 02:20
TimToady string encoding standards usually come from standards bodies 02:22
lue It's availability comes from Parrot, I meant. :) 02:24
s/'//
02:26 jaldhar joined
lue or rather: Are my options for encoding methods in Rakudo limited by what's been implemented in Parrot? 02:30
TimToady well...considering rakudo doesn't really do buffers yet, it's whatever you get as options on open 02:32
02:33 snarkyboojum left 02:35 stef__ joined
lue one of masak's commits seems to suggest we don't have control over available encoding methods directly. 02:36
[ I could be misinterpreting it though :) ]
02:37 snarkyboojum joined 02:38 stef_ left
sorear lue: encodings are just algorithms, you can write your own 02:42
sorear is thinking classes need to be a special kind of sub
TimToady why? 02:45
02:49 ash_ left 02:57 sftp left 02:59 xinming left 03:01 ash_ joined
lue is trying to find out how to write such an algorithm 03:07
03:10 xinming joined 03:13 xinming left
sorear TimToady: because they have a lexical scope and are cloned 03:16
03:23 ash_ left 03:29 xinming joined 03:57 xinming left 03:59 xinming joined 04:07 snarkyboojum left 04:18 hercynium left 04:35 ash_ joined, bjarneh left 04:38 justatheory left, azert0x joined, azert0x left 04:41 azert0x joined, azert0x left, ash_ left
sorear rakudo: my class A { say A.new; } 04:53
p6eval rakudo 692aa1: OUTPUT«too few positional arguments: 1 passed, 2 (or more) expected␤ in main program body at line 1␤»
sorear pugs my class A { say A.new; }
pugs: my class A { say A.new; }
p6eval pugs: OUTPUT«<obj:A>␤»
sorear elf: my class A { say A.new; }
p6eval elf 31610: OUTPUT«syntax error at (eval 124) line 4, near "{ package A"␤BEGIN not safe after errors--compilation aborted at (eval 124) line 5.␤ at ./elf_h line 5881␤» 04:54
sorear mildew: my class A { say A.new; }
p6eval mildew: OUTPUT«Can't open perl script "mildew": No such file or directory␤»
04:58 xinming left 05:01 tylerni7 left 05:03 envi^home joined 05:08 tylerni7 joined 05:09 tylerni7 left, tylerni7 joined
lue is going insane trying to find one webpage about implementing an encoding standard 05:12
05:15 tylerni8 joined 05:16 tylerni7 left 05:21 tylerni8 left, tylerni7 joined, tylerni7 left, tylerni7 joined 05:29 azert0x joined 05:30 xinming joined 05:31 ashleydev joined 05:44 tylercurtis left 05:47 PerlJam left 05:48 PerlJam joined
sorear lue: character encoding standards are completely insane 05:50
every country has ten of the stupid things
also, everybody hates encodings invented by foreigners
I will be amazed if Unicode catches on outside the US
here's one of the more awesome ones: www.unicode.org/reports/tr16/ 05:52
(implemented natively in the Perl 5 core!)
05:53 meppl joined
lue I can't find one document on how to implement an encoding standard, that's what's making me loopy right now :) 05:57
But I agree, way too many encoding standards O.o 05:58
sorear what, exactly, are you having trouble with? 06:00
what encoding standard are you trying to implement?
I would assume you're trying to find the spec for an encoding stnadard
06:01 cls_bsd joined
lue well, I'm looking for general tips on implementing encoding standards (probably ISO-8859-1 first, to let some tests pass). 06:02
sorear 1. find the definition of the encoding 06:03
2. implement it
it really is that easy
find the person who implemented :=; this is 1000x easier
lue found himself
sorear never can tell with those IRC people 06:04
they're often collective pseudonyms
lue The idea I got is a low-level one, where I get the number and then manually pull the letter from a correctly encoded font :D 06:05
If it's that easy, then I'll gladly give that idea up. 06:06
sorear ok 06:07
a character encoding is an algorithm for mapping strings to byte sequences
are you following me
lue yes. 06:08
sorear most encodings (Baudot, SCSU and the ISO-2022 family are the main exceptions here) encode each character independantly
a becomes 61, « becomes C2 AB, «a« becomes C2 AB 61 C2 AB 06:09
in one well-known encoding
still follow
lue yes. 06:10
[I think the idea I got is the kind of idea you implement at the OS level]
06:12 Guest38858 left, Guest38858 joined, Guest38858 is now known as pragma_
sorear just out of curiousity, how old are you 06:14
lue 15 06:15
I know how character encoding works, what happened is I got an overcomplicated idea which made my life miserable.
sorear hah. I knew it
06:16 constant left
lue Now that I know that's overcomplicating things, my life is good again. thank you sorear. 06:16
sorear back when I was 15 I, too, was obsessed with exactly this sort of thing
06:17 constant joined, constant left, constant joined
sorear (I think I still have a copy of that OS I wrote with a text editor, a shell, and a B compiler lying around somewhere. Idiot me didn't understand B+ trees back then so the whole thing is nigh-unusably slow...) 06:18
so, for most encodings, you can encode characters one at a time
a string is a lot like a list of characters
rakudo: say "«a«".comb.perl
p6eval rakudo 692aa1: OUTPUT«("«", "a", "«")␤»
sorear you can unpack it into a real list with .comb
how about a really, really simple encoding: UTF-32BE 06:19
UTF-32BE is defined using Unicode 06:20
Unicode assigns numbers to characters, so it's already most of an encoding
but the numbers Unicode assigns go up to ~1 million, and normal character encodings are only allowed to use 0 - 255
UTF-32 uses base 256 arithmetic to represent each character in 4 "digits" 06:21
rakudo: sub char2utf32($ch) { my $unicode_index = ord($ch); (($ch div 16777216) mod 256), (($ch div 65536) mod 256), (($ch div 256) mod 256), (($ch div 256)) }; say "«a«".comb.map(&char2utf32).perl.say 06:22
p6eval rakudo 692aa1: ( no output ) 06:23
sorear rakudo: sub char2utf32($ch) { my $unicode_index = ord($ch); (($ch div 16777216) mod 256), (($ch div 65536) mod 256), (($ch div 256) mod 256), (($ch div 256)) }; say "«a«".comb.map(&char2utf32).perl
p6eval rakudo 692aa1: ( no output )
sorear what gives, Rakudo
rakudo: sub char2utf32($ch) { my $unicode_index = ord($ch); return (($ch div 16777216) mod 256), (($ch div 65536) mod 256), (($ch div 256) mod 256), (($ch div 256)) }; say "«a«".comb.map(&char2utf32).perl
p6eval rakudo 692aa1: ( no output )
sorear ok, rakudo appears to be broken 06:26
> sub char2utf32($ch) { my $u = ord($ch); return (($u div 16777216) mod 256), (($u div 65536) mod 256), (($u div 256) mod 256), (($u mod 256)) }; say "«a«".comb.map(&char2utf32).flat.perl 06:27
(0, 0, 0, 194, 0, 0, 0, 171, 0, 0, 0, 97, 0, 0, 0, 194, 0, 0, 0, 171)
much better
p6eval--
lue: still with me? 06:28
lue yes.
sorear do you understand what that code is doing?
lue yes, it takes a character (I presume to utf8) and turns it into utf-32 06:29
[I'll read up on the specifics of UTF-32]
sorear I just explained UTF-32.
If you still need to "read up", then no, you're *not* following me and I need to explain something
What do I need to explain?
06:30 xinming left, cognominal left
lue Sorry, I didn't mean it like that. 06:31
06:32 stef__ left, xinming joined
lue [I actually don't know what I meant there; I incorrectly thought there was more magic to UTF-32] 06:32
sorear Oh. Ok. 06:33
(I'm just doing a list of numbers because Rakudo doesn't support byte strings well yet. masak++ for being in the middle of fixing this)
Another important family of encodings are the "SBCS" encodings, or Single-Byte Character Set 06:34
SBCS encodings, as their names imply, devote 1 and only 1 byte to each character
therefore, they can faithfully represent no more than 256 characters
lue (ASCII and ANSI immediately come to mind, although ASCII is technically 7-bit) 06:35
sorear all SBCS encodings predate Unicode, and SBCS encodings were the dominant family (outside of Asia, anyway) before recently
ANSI is not a protocol
er
ANSI is not a character set
06:36 cognominal joined
sorear ANSI is a byte-oriented interactive protocol for communicating with video terminals like the DEC VT-102 06:36
among other services, ANSI allows a character encoding to be negotiated, allowing text to be sent to the terminal 06:37
lue Alright. (I must've misread something.)
sorear ANSI also provides services for editing already sent text, changing font sizes, ejecting sheets of paper, making noises, reprogramming the terminal so that the space bar generates "rm -rf /\n", and many other things 06:38
to my knowledge, no complete implementation exists
note that ANSI is actually the name of a standards body
the standard most people are talking about has a much different name
as is common for standards, it is actually available from several different organizations
you can get a free copy at www.ecma-international.org/publicat...ma-048.pdf 06:39
this is the single most confusing document I have ever attempted to read
I've read disassembler output which made more sense 06:40
also, it's not just you
lots of people misunderstand what it is
end of tangent 06:41
ASCII is, yes, an example
it's the mother of most SBCSes, actually
most national SBCSes came about by importing US-ASCII computers and adding some other characters 06:42
most of them added characters in the 128-255 range that ASCII didn't use (because ASCII was designed for use as part of a 7-bit teleprinter line protocol)
not quite always, though 06:43
a lot of people replaced $ with their favorite currency symbol
pugssvn r31611 | lwall++ | [spec] random cleanup of fossils from before proto became a multi wrapper
06:43 PZt left
sorear the Finns replaced [ \ ] { | } with Ä Ö Ë ä ö ë 06:44
the Finnish code deserves special mention for two reasons.
One, all IRC servers nominally use it, and as such #moo[ and #moo{ are considered to be the same channel (because they're really #mooä and #mooÄ and it's case insensitive 06:45
Two, trigraphs were added to C to support them (and possibly other contries that did the same thing?) if( array ??( index ??) ) ??< code here ??> 06:46
lue That's interesting.
sorear The Japanese (back before computers were powerful enough to support a kanji font) used an 8-bit code with the katakana sylabery in part of 128-255 06:47
but they also switched out \ for the yen sign for some reason
anyways, towards the end of the SBCS era, every country that used non-US-English letters had their own, incompatible, SBCS 06:48
06:49 gfx left
lue
.oO(what joy international emails and the Internet in general must've been)
 06:49
sorear I should also mention EBCDIC, IBM's ASCII competitor, which was origenally intended to be a compatible extension of the old (pre-1900) punch card formats, but they managed to invent six versions of it that weren't even compatible with *each other*... luckily this is dead now
exactly
international e-mail is what about the end of this era 06:50
brought about
near the end, the ISO tried to improve the situation as much as they could while still using 8-bit codes 06:51
so they came up with a few encodings each of which served several countries, collectively known as ISO-8859 06:52
06:52 LionMade0fLions joined
sorear ISO-8859-1 was designed to serve England, France, Spain, the U.S, Germany 06:52
lue (about 13 or 14 of them, if I recall)
TimToady -1 became -15 when they added the euro
but people still just call it latin-1 06:53
sorear it has ASCII, «french quotes», áccentos, eñe, sharp s (ß), a few other things
TimToady and that became the first 256 chars of Unicode
.u ¬ 06:54
phenny U+00AC NOT SIGN (¬)
sorear Unicode is not a character encoding. It's a character *set*. Unicode, for the first time, made it possible to talk about characters independant of any one encoding
06:54 LionMadeOfLions left, breinbaas joined
TimToady but the codepoints have abstract numbers that just happen to match latin-1 06:55
sorear after Unicode was invented (early 90s), changing text between encodings became a well-defined proposition
they match a lot of existing encodings
TimToady as long as you believed in UCS-2
sorear I think the general rule was "as little creativity as possible, match status quo whenever you can"
stuff like smilies are in Unicode mostly because a few of the SBCSes had them 06:56
TimToady I know a bunch of CJK folks who would say a great deal too much creativity was used :)
Zapf Dingbats and such 06:57
sorear I know almost nothing about pre-Unicode CJK encodings, so I'm not talking about that aspect of the history
you're welcome to though
lue what a wonderful history you're giving by the way, sorear.
sorear this is the end of my tangent. 06:59
encodings used now in the West fall into two groups
the "legacy" SBCS encodings are now defined as a 1:1 map between Unicode character numbers and bytes
TimToady non-linear for anything not ISO-8859-1ish 07:00
and the UTFs 07:01
which presume to represent any valid Unicode codepoint
albeit by different schemes
lue I can't help but wonder why Unicode stopped at such a seemingly arbitrary number, 0x10FFFF 07:02
TimToady UTF-32 is a fixed-width scheme
sorear that's the other group of encodings used now
TimToady they figgered 16 planes were enough
sorear the Unicode-based encodings
the Unicode-based encodings are defined as a simple mathematical translation of the codepoint numbers 07:03
TimToady UTF-16 was original viewed as fixed width before they invented surrogates to represent chars above ffff
but it's really variable width
UTF-8 is anabashedly variable width
*un
sorear the simplest Unicode-based encoding is the one I showed you earlier - UTF32
moritz_ good morning
TimToady o/
lue good morning moritz_ o/
TimToady UTF-32 is more or less equivalent to UCS-4 07:04
and UCS-2 was equivalent to UTF-16 before they invented surrogates
lue can't wait to see what happens to Unicode when we discover sentient aliens as advanced as us... *muhahahaha*
TimToady UCS-1, if it existed, would more or less be Latin-1
sorear I think merging our character set standards will be the least of our worries when that happens 07:05
TimToady mostly the western world is moving almost exclusively to UTF-8 for interchange of data
in the east there tends to be a preference for UTF-16 because it is more compact for east asian characters 07:06
(where they use Unicode at all; many continue to use Big-5 or ISO-2022 or whatever their original non-unicode character set was) 07:07
ingy how do I subst "x/y/z" into "x/y"? 07:08
sorear the common east asian characters in Unicode are assigned around 4000-8000, the rare ones are assigned above 10000
TimToady see the perl5 encode module for much more on this subject
ingy sorear: the bareword 'this' seemed to cause silent errors... 07:09
sorear ingy: it's spelled self in perl6
TimToady: does an East Asian think of 一 as U+4E00 or as a codepoint in their native set?
(independant of how they encode it)
ingy sorear: I know that. But I was using a $this.foo and forgot the $ and it got no compile or runtime errors 07:10
TimToady they think of it as a character they learned in school, and don't much give a rip about the other :)
std: this.foo
ingy sorear: so I assumed 'this' meant something
p6eval std 31611: OUTPUT«===SORRY!===␤Undeclared routine:␤ 'this' used at line 1␤Check failed␤FAILED 00:01 110m␤»
TimToady rakudo: this.foo
p6eval rakudo 692aa1: OUTPUT«Could not find sub &this␤ in main program body at line 20:/tmp/Pdi9pnukUK␤»
ingy *shrug*
IT HAPPENED!
:) 07:11
TimToady were you trapping run-time errors?
rakudo doesn't catch undef routines at compile time like std does
ingy maybe
I have no ida what I'm trapping most of the time
:)
00:08 <@ingy> how do I subst "x/y/z" into "x/y"? 07:12
:]
TimToady same way you would in Perl 5, but quote the /
moritz_ rakudo: say 'foo x/y/z bla'.subst('x/y/z', 'x/y')
sorear I need a better way to type stuff like 一
p6eval rakudo 692aa1: OUTPUT«foo x/y bla␤»
sorear that's control-shift-4E00 :/ 07:13
lue well, my biggest concern now is how you can implement new encoding standards (like UTF-32) in P6, and use them for Buf and IO
sorear unfortunately I've yet to find an input method with documentation in English
moritz_ sorear: compose - - ?
lue an em dash? —
ingy rakudo: say "x/y/z".subst(/(.*)\/.*/, $0)
p6eval rakudo 692aa1: OUTPUT«x/y/z␤»
sorear moritz_: that's not a dash, that's the CJK character for one
TimToady I can just type CTRL-space ichi
sorear I can't read Japanese at a software manuals level 07:14
TimToady: what do I need to install to make that work?
moritz_ ingy: if you want to use any match variables on the RHS of a subst, you need to use a closure, which gets $/ as the first positional parameter
rakudo: say "x/y/z".subst(/(.*)\/.*/, -> $/ { $0 })
TimToady 一 is not software manual level; it's first grade. :)
p6eval rakudo 692aa1: OUTPUT«x/y/z␤» 07:15
moritz_ huh
sorear TimToady: the documentation for installing input methods is well above a first-grade level
moritz_ rakudo: say "x/y/z".subst(/(.*)\/.*/, -> $/ { 'WTF' })
p6eval rakudo 692aa1: OUTPUT«x/y/z␤»
moritz_ rakudo: say "x/y/z".subst(rx/(.*)\/.*/, -> $/ { 'WTF' })
p6eval rakudo 692aa1: OUTPUT«x/y/z␤»
sorear the only input method I've successfully installed is compose key handling for «, and that took months of shotgun debugging
TimToady I use anthy
moritz_ rakudo: say "x/y/z".subst(rx/(<-[/]>*)\/.*/, -> $/ { 'WTF' }) 07:16
p6eval rakudo 692aa1: OUTPUT«WTF␤»
lue I use IBus with anthy, then I type CTRL Space - Enter
moritz_ rakudo: say "x/y/z".subst(rx/(<-[/]>*)\/.*/, -> $/ { $0 })
p6eval rakudo 692aa1: OUTPUT«x␤»
moritz_ backtracking into captures
ingy GO moritz_!!!
:)
sorear TimToady: urxvt good enough or do I also need a better terminal?
TimToady nokloo 07:17
ingy I'm just trying to get the dir part of a pathname...
TimToady I just use gnome term
lue is there some sort of magic role or something to make your encoding usable with IO and such?
TimToady mostly you have to make sure LANG=en_US.UTF-8 07:18
then most things work right
ingy rakudo: say "x/y/z".subst(/(.*)/.*/, -> $/ { $0 }) 07:19
p6eval rakudo 692aa1: OUTPUT«===SORRY!===␤Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 20␤»
ingy rakudo: say "x/y/z".subst(/(.*)[/].*/, -> $/ { $0 })
p6eval rakudo 692aa1: OUTPUT«===SORRY!===␤Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 20␤»
TimToady [] isn't character class in p6
use '/'
ingy rakudo: say "x/y/z".subst(/(.*)<[/]>.*/, -> $/ { $0 })
p6eval rakudo 692aa1: OUTPUT«x/y/z␤»
ingy rakudo: say "x/y/z".subst(/(.*)<before />.*/, -> $/ { $0 }) 07:20
p6eval rakudo 692aa1: OUTPUT«===SORRY!===␤regex assertion not terminated by angle bracket at line 20, near " />.*/, ->"␤»
moritz_ it's backtracking into captures that's NYI and causes trouble
ingy rakudo: say "x/y/z".subst(/(.*)<before </>>.*/, -> $/ { $0 })
p6eval rakudo 692aa1: OUTPUT«===SORRY!===␤regex assertion not terminated by angle bracket at line 20, near " </>>.*/, "␤»
07:21 xinming left
ingy rakudo: say "x/y/z".subst(/(.*)<before [/]>.*/, -> $/ { $0 }) 07:21
p6eval rakudo 692aa1: OUTPUT«===SORRY!===␤regex assertion not terminated by angle bracket at line 20, near " [/]>.*/, "␤»
TimToady pmichaud was gonna fix backtracking into things pretty soon
lue
.oO(the spec doesn't seem to say anything about creating new encodings...)
TimToady the spec wants to believe that every can live happily with unicode :)
*everyone 07:22
sorear lue: create a new encoding and thousands of people will swear to kill you.
ingy so how can I get the dirpath from a unix filepath?
sorear don't do it.
ingy :)
tmtowtdi, I'm sure
TimToady rindex :) 07:23
07:23 xinming joined
TimToady rakudo: say "x/y/z".subst(/(.*)'/'.*/, { $0 }) 07:24
p6eval rakudo 692aa1: OUTPUT«x/y/z␤»
TimToady rakudo: say "x/y/z".subst(/(.*)'/'.*/, -> $/ { $0 })
p6eval rakudo 692aa1: OUTPUT«x/y/z␤»
lue TimToady: weeeell, one test relies on ISO-8859-1 to work right.
(t/spec/S16-filehandles/io.t) 07:25
ingy rakudo: say "x/y/z".subst(/(.*)'/'.*/, -> $/ { 'WTF' })
p6eval rakudo 692aa1: OUTPUT«x/y/z␤»
TimToady rakudo: say "x/y/z".subst(/(.*)'/'.*/, 'foo')
p6eval rakudo 692aa1: OUTPUT«x/y/z␤»
lue goodnight o/
TimToady \o 07:26
rcsheets` what is that last character inside p6eval's angle quotes? it renders as a thin box for me.
TimToady rakudo: say "x/y/z".subst(/^(.*)'/'.*$/, 'foo')
p6eval rakudo 692aa1: OUTPUT«x/y/z␤»
TimToady newline 07:27
.u ␤
phenny U+2424 SYMBOL FOR NEWLINE (␤)
rcsheets` ah, ok. thanks.
ingy rakudo: $f = "x/y/file"; say $f.substr(0, $f.rindex('/'))
p6eval rakudo 692aa1: OUTPUT«===SORRY!===␤Symbol '$f' not predeclared in <anonymous> (/tmp/qNbvoUISVZ:20)␤»
TimToady most terminals it's an N over L
ingy rakudo: my $f = "x/y/file"; say $f.substr(0, $f.rindex('/'))
p6eval rakudo 692aa1: OUTPUT«x/y␤»
ingy :)
rcsheets` i assume it's just the font i'm using 07:28
TimToady rakudo: say "x/y/z".match(/^.* <?before '/'>/) 07:29
p6eval rakudo 692aa1: OUTPUT«x/y␤»
TimToady rcsheets`: could well be a font issue
sorear installing anthy and uim-anthy and restarting X did nothing.
I need docs for this, and not ones written in Japanese 07:30
ctrl-space ichi -> "ichi"
TimToady what kind of machine?
sorear PC, debian sid 07:31
sorear tries installing gnome-terminal
07:32 tedv left
TimToady you might need to tell it that ctrl-space is what turns it on somehow 07:32
it might be bound to a different seq by default 07:33
or you might need to speak to whatever manages your input methods and enable anthy in it
or it could be your term
historically some systems have been loathe to believe that anyone in US_en would want east asian keybindings 07:34
but that has improved in the last couple years
er, en_US 07:35
*loath
07:35 skids left
sorear I think I need to install something to manage my input methods 07:35
I don't currently have anything fitting that description
TimToady scim or ibus, ibus being the newest
ibus is current default on ubuntu 07:36
ingy rakudo: $*PROGRAM_NAME
p6eval rakudo 692aa1: ( no output )
ingy rakudo: say $*PROGRAM_NAME
p6eval rakudo 692aa1: OUTPUT«/tmp/827m47pFcy␤»
ingy rakudo: say $*PROGRAM_NAME.rindex('/')
p6eval rakudo 692aa1: ( no output ) 07:37
TimToady rakudo: say $*PROGRAM_NAME.WHAT
p6eval rakudo 692aa1: OUTPUT«Str()␤»
sorear installs ibus, ibus-gtk, ibus-anthy
ingy I get unimplemented unicode in 'Cool::rindex' at line 2220:CORE.setting 07:38
sorear apparently I now need to upgrade libc6, gcc, and python
ingy rakudo: say $*PROGRAM_NAME.rindex('/')
p6eval rakudo 692aa1: ( no output )
07:39 skids joined
TimToady rakudo: say rindex('/',$*PROGRAM_NAME).perl 07:39
ingy rakudo: my $p = $*PROGRAM_NAME; say $p.rindex('/')
p6eval rakudo 692aa1: ( no output ) 07:40
ingy rakudo: my $p = $*PROGRAM_NAME.Str; say $p.rindex('/')
TimToady timing out maybe
p6eval rakudo 692aa1: ( no output )
TimToady rakudo: say rindex('/',$*PROGRAM_NAME).WHAT
ingy rakudo: say "OK"
p6eval rakudo 692aa1: ( no output )
rakudo 692aa1: OUTPUT«OK␤»
TimToady Cross-charset index not supported in 'Cool::rindex' at line 2212:CORE.setting in main program body at line 1:foo 07:41
sorear wonders if he should uninstall uim and xim 07:42
TimToady so some kind of bug that p6eval is deigning not to report 07:43
sorear p6eval doesn't report errors recently
07:43 ashleydev left
TimToady I don't think you should need to, as long as you aren't giving old env vars 07:43
sorear I think it's part of moritz_ 's information suppression campaign
TimToady I've got: XMODIFIERS=@im=ibus 07:44
ingy TimToady: is $*PROGRAM_NAME not a string?
TimToady rakudo: say '/'.PARROT
p6eval rakudo 692aa1: OUTPUT«Str␤»
TimToady rakudo: say $*PROGRAM_NAME.PARROT
p6eval rakudo 692aa1: OUTPUT«String␤»
ingy Str vs String?
rakudo: say $*PROGRAM_NAME.Str.PARROT 07:45
p6eval rakudo 692aa1: OUTPUT«String␤»
TimToady rakudo: my $s = $*PROGRAM_NAME ~ ''; say $s.rindex('/')
p6eval rakudo 692aa1: ( no output )
TimToady looks like a leaky parrot 07:46
ingy rakudo: my $s = ~$*PROGRAM_NAME; say $s.rindex('/')
p6eval rakudo 692aa1: ( no output )
07:47 xinming left
sorear perl: warning: Setting locale failed. 07:48
TimToady rakudo: say ($*PROGRAM_NAME ~ "").rindex('/')
sorear D:
p6eval rakudo 692aa1: ( no output )
TimToady do you have the utf8 locales installed?
locale -a should tell you 07:49
sorear I have en_US.utf8 07:50
TimToady and LANG=en_US.UTF-8 is what I use
though, oddly, local -a lists it as .utf8 07:51
probably normalizes
sorear $LANG is .utf8 here
TimToady possibly Perl wants the UTF-8 form
sorear perl -e '' doesn't carp after apt-get returned
TimToady I dunno
sorear do I need to poke perl harder to get it to locale carp?
ingy rakudo: say "x/y/z".subst(/^.* <?before '/'>/, '')
p6eval rakudo 692aa1: OUTPUT«/z␤» 07:52
ingy rakudo: say "x/y/z".subst(/^(.*) <?before '/'> .*/, -> $/ { $0 })
p6eval rakudo 692aa1: OUTPUT«x/y/z␤»
ingy :'( 07:53
TimToady subst is supposed to set $/ in the outer scope so that the closure sees it without passing it as a param
but I don't think rakudo does that right yet
rakudo: my $p = "x/y/z"; $p ~~ s/(.*) '/'/$0/; say $p 07:54
p6eval rakudo 692aa1: OUTPUT«x/y/z␤»
07:55 skangas left
TimToady rakudo: my $p = "x/y/z"; $p ~~ s/(.*) '/' .*/$0/; say $p 07:55
p6eval rakudo 692aa1: OUTPUT«x/y/z␤»
TimToady rakudo: my $p = "x/y/z"; $p ~~ s/.* <( '/' .* )>//; say $p 07:56
p6eval rakudo 692aa1: OUTPUT«===SORRY!===␤Confused at line 20, near "$p ~~ s/.*"␤»
TimToady rakudo: my $p = "x/y/z"; $p ~~ s/.* '/'//; say $p 07:57
p6eval rakudo 692aa1: OUTPUT«z␤»
TimToady rakudo: my $p = "x/y/z"; $p ~~ s/$<foo>=[.*] '/' .*/$<foo>/; say $p 07:58
p6eval rakudo 692aa1: OUTPUT«Any()␤»
sorear according to pstree, ibus-daemon is now running 07:59
ingy rakudo: my $f = "x/y/file"; while $f ~~ s/<! '/'>$// {}; say $f
TimToady rakudo: my $p = "x/y/z"; my $foo; $p ~~ s/$foo=[.*] '/' .*/$foo/; say $p
p6eval rakudo 692aa1: OUTPUT«===SORRY!===␤Confused at line 20, near "while $f ~"␤»
rakudo 692aa1: OUTPUT«===SORRY!===␤Confused at line 20, near "$p ~~ s/$f"␤»
TimToady maybe you should just use split :)
sorear TimToady: pastie.org/1038636 - do you see anything obviously wrong here? 08:00
ingy rakudo: my $f = "x/y/file"; while ($f ~~ s/<! '/'>$//) {}; say $f
p6eval rakudo 692aa1: OUTPUT«===SORRY!===␤Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 20␤»
sorear it's still not working after despite ibus running
hmm
maybe I need to configure ibus to use anthy somehow.
TimToady I seem to recall someone (ingy?) having trouble with have all the other LC_* set 08:01
that cleared up after setting only LANG
I get a little ibus gidget in the top bar 08:02
ingy rakudo: my $f = "x/y/file"; export LANG="en_US.UTF-8"
export LANG="en_US.UTF-8"
p6eval rakudo 692aa1: OUTPUT«Could not find sub &LANG␤ in main program body at line 20:/tmp/OSx9lm_7yu␤»
TimToady with a little picture of a world in front of a keyboard 08:03
that lets you configure ibus
ingy rakudo: my $f = "x/y/file"; my $a = $f.split('/').pop; say $a.perl
p6eval rakudo 692aa1: OUTPUT«"file"␤»
sorear my computer isn't powerful enough for a top bar
I had to disable it
ingy rakudo: my $f = "x/y/file"; my $a = $f.split('/'); say $a.perl
p6eval rakudo 692aa1: OUTPUT«("x", "y", "file")␤»
ingy rakudo: my $f = "x/y/file"; my $a = $f.split('/'); $a.pop; say $a.perl
p6eval rakudo 692aa1: OUTPUT«("x", "y")␤» 08:04
ingy rakudo: my $f = "x/y/file"; my $a = $f.split('/'); $a.pop; say $a.join('/')
TimToady rakudo: my $f = "x/y/file"; say $f.flip.subst(/.*?'/'/, '').flip
p6eval rakudo 692aa1: OUTPUT«x/y␤»
sorear hmm. this 'ibus-setup' thing looks promisng
一! 08:05
sorear dances
ingy rakudo: say "abc".flip
p6eval rakudo 692aa1: OUTPUT«cba␤»
TimToady すごい！
ingy TimToady++ 08:06
TimToady in p6 we differentiate .reverse from .flip
.reverse is only for lists
sorear hey, ibus-setup even lets me disable control-space 08:10
sorear switches IM activation to a key combo he hits less often by accident
TimToady yeah, I have that problem too...
sorear TimToady++
ingy I _finally_ have TestML really working now. Starting to pass the TestML test suite... 08:16
p6/rakudo is a huge PITA.
but you all are great to work with! 08:17
sorear How, exactly, do you test a test harness
drbean sorear: 恭喜! 恭喜!
moritz_ sorear: like everything else: you put in example input, and monitor the response 08:18
ingy sorear: TestML is not a harness
08:18 TiMBuS joined
ingy it's a programming language neutral way to write unit tests, that works under any existing harness 08:18
sorear drbean: that doesn't look like any of the 15 hanzi I know 08:19
TimToady 恭喜發財 恭喜发财 [gong1 xi3 fa1 cai2] /congratulations for getting rich (traditional Chinese New Year's future use/ 08:20
sorear what did that come from?
08:20 azert0x left
TimToady originally, cedict 08:21
sorear originally?
TimToady well, I've had it since 2004 or so, so who knows what it is now... 08:22
my mouth keeps flopping open, so I'd better flop over soon on purpose before I flop over accidentally 08:26
zzz &
08:28 Ross joined
sorear perl6: say class { &?BLOCK.leave(2); } # Opinions? 08:43
p6eval rakudo 692aa1: OUTPUT«Could not find sub &?BLOCK␤ in main program body at line 20:/tmp/lna2SdHXlV␤»
..pugs: OUTPUT«*** No such subroutine: "&class"␤ at /tmp/vAdP1UyJjg line 1, column 5 - line 2, column 0␤»
sorear pugs: say class :: { &?BLOCK.leave(2); } 08:44
p6eval pugs: OUTPUT«*** ␤ Unexpected " {"␤ expecting "(", twigil or word character␤ at /tmp/dzaSDMqFDH line 1, column 13␤»
sorear pugs: say anon class { &?BLOCK.leave(2); }
p6eval pugs: OUTPUT«*** No such subroutine: "&anon"␤ at /tmp/O5Yf7aEIG1 line 1, column 5 - line 2, column 0␤»
dalek ecza: 9ddeaf8 | sorear++ | (3 files):
A few steps towards class declarations
 08:54
09:02 gfldex joined 09:10 pmurias joined, Ross left 09:17 Ross joined 09:39 PZt joined 09:40 quietdev joined
sorear results of today's walk: 09:54
- niecza is going to implement modules using a blatant rip of the .NET assemblies mechanism, which I will then convince TimToady to write into the spec
- precompilation of individual modules only really needs to be supported for the cases where whole program static compilation is possible 09:55
- STD.pm6 tangles static analysis too much; the module system, setting loader, trait system should be factored out to an $*ACTIONS-like contextually bound pair 09:56
- $STD::ALL could usefully be refactored, perhaps by only storing the roots of compilation units which are referenced by use, or something like that. OUTER is problematic, doubly so because of P5's circular reference stupidity 09:58
sorear out. 10:02
10:38 eiro joined, eiro left 10:58 redicaps1 joined 11:00 redicaps1 left 11:16 sftp joined 11:25 M_o_C joined 11:28 Kejohm joined 11:32 Kejohm left 11:35 s1n left
cognominal rakudo: say ?1..2 11:39
p6eval rakudo 692aa1:
..OUTPUT«111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111…
moritz_ wtf? 11:44
cognominal masaked
11:45 Ross left, Ross joined
cognominal #76442 11:45
moritz_ rakudo: say Bool::True..2 11:46
jnthn rakudo: say ?(1..2)
p6eval rakudo 692aa1:
..OUTPUT«111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111…
rakudo 692aa1: OUTPUT«1␤»
jnthn oh
Bool.succ on true is identity.
11:46 eiro joined, eiro left
moritz_ right 11:46
jnthn I bet that's what causes it. 11:47
cognominal I think Range should be a parametric role so as to force the two arguments to be of the same type.
not sure of the details though
moritz_ or at least if one endpoint is numeric, coerce the other ot numeric 11:48
cognominal it makes no sense to have extremty of a range to have different types. 11:49
I think my example should fail. 11:50
jnthn moritz_: That makes sense, yes 11:52
11:57 pragma_ left 11:58 M_o_C left 12:01 pmurias left 12:05 pragma_ joined 12:06 pragma_ is now known as Guest38594 12:09 kid51 joined 12:25 ilogger2 joined 12:30 SmokeMachine joined 12:31 rv2733 joined, SmokeMachine left 12:32 SmokeMachine joined 12:37 takadonet joined
takadonet morning all 12:38
13:04 SmokeMachine left 13:10 SmokeMachine joined 13:13 takadonet left 13:16 SmokeMachine left 13:40 tylercurtis joined 13:41 TiMBuS joined 13:44 masak joined
masak #perl6! \o/ 13:44
jnthn #masak! \o/ 13:45
masak rakudo: say (my $a) = 1,2,3; say $a # discovered by TimToady++
p6eval rakudo 692aa1: OUTPUT«Any()␤»
masak did it ever get reported?
jnthn masak: Now i'm awake-ish.. 13:46
rakudo: say ((my $a) = 1,2,3);
p6eval rakudo 692aa1: OUTPUT«1 2 3␤»
jnthn rakudo: (say (my $a)) = 1,2,3;
p6eval rakudo 692aa1: OUTPUT«Any()␤» 13:47
masak ah.
jnthn It's only a bug if it's not meant to have parsed as the second of those.
masak asks viv
13:53 mulander joined
masak viv puts the 'say' outermost. gist.github.com/470719 13:54
masak submits rakudobug
jnthn masak++ 14:04
moritz_ masak++ # ufo 14:11
masak moritz_++ # p6eval yapsi target
moritz_ masak: I haven't cron-jobbed the rebuild yet 14:12
masak no worries. I'm not hacking on it right now.
might be later today, though. :) eager to set my teeth in the new runtime.
moritz_ just made Tufte compile 14:13
14:13 azert0x joined, azert0x left, pmurias joined
moritz_ and I realized I don't understand enough ruby to continue porting scruffy 14:14
14:23 azert0x joined, azert0x left
masak Austrian Airlines and I seem to have just agreed on some nice, cheap YAPC::EU flight tickets for me. :) 14:27
moritz_ you use an Australian airline to fly from Sweden to Italy? sweet :-)
masak er, no :) 14:28
has anyone else investigated trains from Bologna to Pisa? 14:29
jnthn Austria is the one without Kangaroos 14:30
masak: Yes
masak: I'll be doing that.
masak me too. on the Monday.
jnthn When? 14:31
masak I'll be arriving in Bologna 19:15 on 2010-08-02.
jnthn Ah, OK
Waaaay later than me.
masak ok.
jnthn (Like, 7 hours later.) 14:32
Anyway, probably it's a case of connecting in Florence.
Though there may be some direct trains too.
14:33 timbunce joined
masak jnthn: does one pre-book trains, or just buy them JIT? 14:33
14:33 azert0x joined, azert0x left
jnthn masak: I plan to buy them on the day. 14:33
masak: I did that kinda thing when in Italy before now. 14:34
Worked out.
masak you plan to buy them afk, and not thourhg the intertubes?
*through
jnthn Probably
I'm not sure what options for buying them over the Intertubes are for Italy
I think I looked into it when doing vacation in Italy previously, and ended up not doing so. 14:35
masak ok.
14:35 azert0x joined 14:36 azert0x left, azert0x joined
masak "No subsets of Ada are allowed. Babbage is just the opposite. None of Babbage is defined except its extensibility - each user must define his own version. To end the debate of large languages versus small, Babbage allows each user to make the language any size he wants. Babbage is the ideal language for the "me" generation." -- www.tlc-systems.com/babbage.htm 14:37
that sounds a bit like Perl 6, actually :) 14:38
moritz_ it sounds a bit more extremist than Perl 6
(which is a large language by default)
masak I guess it is.
or it's a small language with a default large language slapped on.
*lol* -- the calling convention is "call by telephone" :P 14:39
"This is especially effective for long-distance parameter passing."
moritz_ masak: I'm hacking a bit on SVG.pm... currently SVG.serialize takes a single positional argument 14:44
masak: which makes it a bit awkward to pass pairs to it 14:45
SVG.serialize((:x[]))
masak moritz_: hm. yes.
moritz_ masak: mind if I magically accept a single, named param instead?
masak no.
please go ahead.
moritz_ I'm also writing tests for it :-) 14:46
masak moritz_++
moritz_ with the background of refactoring later
masak sounds good.
14:47 patspam joined
masak sorear, lue: when I was 15, I was also toying with the idea of writing an OS. :-P 14:48
tylercurtis masak: You should. I cannot wait to see the release announcements for that. :) 14:53
TimToady ranges are not always the same type on both sides. see 1..* 14:54
After Christmas it should be our goal to make Perl 6 fast enough that we can write an OS in it...and then *not* write an OS in it. :) 14:57
14:57 skids joined 15:00 oha joined
moritz_ sounds boring :-) 15:00
rakudo: say .Int.perl for True, False 15:01
p6eval rakudo 692aa1: OUTPUT«Bool::True␤Bool::False␤»
moritz_ uhm.
masak: wanna submit? :-)
TimToady rakudo: say .Numeric.perl for True, False 15:02
p6eval rakudo 692aa1: OUTPUT«1␤0␤»
TimToady but I suspec .Int should do that too
rakudo: say True ~~ Int
masak ubsmits
p6eval rakudo 692aa1: OUTPUT«0␤»
moritz_ rakudo: sub f($t?) { say defined($t).Int }; f("a" => 3)
masak er, *submits 15:03
p6eval rakudo 692aa1: ( no output )
moritz_ gives "Method 'Int' not found for invocant of class 'Pair'" locally
which is wrong too, no?
is defined() parsed specially?
ingy morning 15:04
masak moritz_: no.
moritz_ so another bug?
masak let me confirm locally.
ingy just discovered that .WHICH is not really unique.
TimToady rakudo: say defined("a" => 3).WHAT
p6eval rakudo 692aa1: OUTPUT«0␤»
masak moritz_: I get the same.
masak submits rakudobug
moritz_ rakudo: say 1 xor 1 15:05
p6eval rakudo 692aa1: OUTPUT«1␤»
moritz_ huh.
ingy rakudo: my $h = {}; my $n = $h.WHICH; say $h.WHICH; say $n.WHICH;
p6eval rakudo 692aa1: OUTPUT«35409216␤35409216␤»
TimToady rakudo say (1 xor 1)
rakudo: say (1 xor 1)
p6eval rakudo 692aa1: OUTPUT«␤»
moritz_ ingy: that'a known problem, .WHICH should include type information 15:06
masak that's right, right?
ingy I think it would be better to add a '.' or something to the beginning of scalars' WHICH
masak the 'say 1 xor 1' thing.
moritz_ masak: yes, looser precedence than I thought 15:07
ingy moritz_: no, I think WHICH should just be unique
masak moritz_: makes sense that it's the same as 'or', though.
TimToady S02:583
moritz_ ingy: which it would be achieve with type info, yes
TimToady rakudo is not up to spec there
ingy adding a '.' would also, moritz_
and simpler
:)
masak rakudo: sub f($t?) { say defined($t).WHAT }; f("a" => 3) 15:08
moritz_ ingy: the old spec said that for value objects (Str, Int, Num etc.) .WHICH == $_
p6eval rakudo 692aa1: OUTPUT«0␤»
masak o.O
ingy moritz_: and I agree with that
moritz_ ingy: then a dot doesn't help
ingy moritz_: oh?
moritz_: example?
moritz_ unless you do it only for non-value objects
TimToady S02:583 !!! 15:09
masak hugme: hug TimToady
hugme hugs TimToady
moritz_ TimToady: you may be the language designer, but we're still free to ignore you :-)
ingy TimToady: to whom do you speak?
masak rakudo: sub f($t?) { say defined($t).WHAT }; f()
p6eval rakudo 692aa1: OUTPUT«0␤»
diakopter winks awake
masak rakudo: sub f($t?) { say defined($t).Int }; f() 15:10
p6eval rakudo 692aa1: ( no output )
TimToady ingy: when I cite the spec in the middle of a conversation, it means the spec already discusses this, and it might have some bearing on the conversation, unless you're moritz_++ :) 15:11
masak I like the exception clause to that. 15:12
ingy moritz_: ''.WHICH -> '.'; '.'.WHICH' -> '..'; $h.WHICH.WHICH -> '.35409216'
TimToady: except you were in the middle of 3 conversations
:)
diakopter std: S02:583 !!!
p6eval std 31611: OUTPUT«===SORRY!===␤Confused at /tmp/v1zB_w_eZb line 1:␤------> S02:⏏583 !!!␤ expecting any of:␤ coloncircumfix␤ signature␤Undeclared name:␤ 'S02' used at line 1␤Parse failed␤FAILED 00:01 108m␤»
ingy TimToady: and my reponse to your blurt would then be, "url?". 15:13
ingy just woke up on the grumpy side of the bed 15:14
moritz_ irc.pugscode.org
masak ingy: just go to the IRC logs, and it'll be a link.
moritz_ which turns S02:583 into an URL
ingy o neat! 15:15
15:15 oha left
moritz_ ingy: we've spent a long time optimizing those autolinking features :-) 15:15
masak moritz_++ # again 15:16
diakopter rakudo: say ?|0?|0
p6eval rakudo 692aa1: OUTPUT«1␤»
diakopter rakudo: say ?|0!?|0 15:17
p6eval rakudo 692aa1: OUTPUT«0␤»
ingy moritz_++ 15:18
masak afk & 15:19
15:19 masak left
ingy TimToady: so... I'm glad that S02:583 mentions YAML, because I was just starting to write a YAML dumper for p6 when I came across the WHICH issue. For YAML's purposes, adding a '.' to the start of scalars (and using the current numbers for objects) would probably work just great. 15:21
TimToady: that's akin to a scheme I use for p5...
then again, I'm not too awfully concerned about the collisions for now :) 15:22
bbi2-3h... 15:23
TimToady thinks "fractional addresses..."
15:28 ashleydev joined
moritz_ phenny: tell masak that I've pushed some fairly substantial changes to SVG, making it use methods internally instead of subs. My plan would be to rename it to XML::Writer or so, and then start a new SVG.pm, which automatically promotes some attributes into :style<...>, and optionally adds an SVG header 15:31
phenny moritz_: I'll pass that on when masak is around.
15:33 asig2 joined 15:40 asig2 left 15:48 ashleydev left
timbunce What's the best way to represent exportable constants in a role? 16:05
TimToady std: my constant pi is export = 3; # won't work in rakudo though 16:13
p6eval std 31611: OUTPUT«ok 00:01 112m␤»
TimToady or possibly
std: has constant pi is export = 3;
p6eval std 31611: OUTPUT«ok 00:01 109m␤» 16:14
TimToady constant is supposed to be 'our' by default
rakudo: my constant pi = 3;
p6eval rakudo 692aa1: OUTPUT«===SORRY!===␤Malformed my at line 20, near "constant p"␤»
diakopter std: constant pi is export = 3; 16:15
TimToady but rakudo is still parsing it as a scope declarator rather than a sub declarator
p6eval std 31611: OUTPUT«ok 00:01 109m␤»
timbunce TimToady: thanks. What's a workable approach for rakudo meanwhile?
moritz_ wonders if enums are exported
TimToady rakudo: role Foo { constant pi = 3 }; say Foo::pi
p6eval rakudo 692aa1: OUTPUT«Can not find sub Foo::pi␤ in main program body at line 1␤» 16:16
moritz_ rakudo: role Foo { our $pi = 3 }; say $Foo::pi
p6eval rakudo 692aa1: OUTPUT«Null PMC access in type()␤ in main program body at line 20:/tmp/M57BNs8HR2␤»
moritz_ rakudo: module Foo { our $pi = 3 }; say $Foo::pi
p6eval rakudo 692aa1: OUTPUT«3␤»
moritz_ feijl
moritz_ submits rakudobug
timbunce moritz_: yeah, that'll do. thanks
16:17 TiMBuS left
jnthn moritz_: fwiw, I'm not sure how to make that one work. 16:17
our scoped stuff in roles doesn't make too much sense or at least needs some more thought 16:18
TimToady rakudo: role Foo { our $pi = 3 }; class F does Foo {}; say $F.pi
p6eval rakudo 692aa1: OUTPUT«===SORRY!===␤Symbol '$F' not predeclared in <anonymous> (/tmp/6oAiHlFE7w:20)␤»
TimToady rakudo: role Foo { our $pi = 3 }; class F does Foo {}; say $F::pi
p6eval rakudo 692aa1: OUTPUT«Null PMC access in type()␤ in main program body at line 20:/tmp/WzM6KtH3Vy␤»
jnthn TimToady: heh, that's not going to work for sure
moritz_ jnthn: maybe by punning the role into a class of the same name? 16:19
TimToady rakudo: role Foo { our $pi = 3 }; Foo.new; say $Foo::pi
p6eval rakudo 692aa1: ( no output )
TimToady kaboom?
jnthn TimToady: er 16:20
timbunce thinks I'll generate methods for now and wait for the dust to settle...
jnthn TimToady: Why would the meta-object for a role know about about an our-scoped variable in the role body?
16:20 justatheory joined
jnthn *know anything about 16:20
diakopter moritz_: we could make p6eval output the "killed..." message I think
TimToady std: has constant $.pi = 3;
moritz_ diakopter: +1 if you know how to do it
p6eval std 31611: OUTPUT«ok 00:01 110m␤»
diakopter heh
jnthn TimToady: Anyway, for it to work at all we probably need to say that creating a concrete role creates a package along side it, then we'd need to get the concrete form of the role and .WHO it to get the package and lookup in that. 16:21
16:21 skangas joined
jnthn Since it could easily have been role Foo[$x] { our $y = $x; } 16:21
TimToady twoo 16:22
jnthn In general, anything our-scoped inside a role is horribly awkward.
TimToady well, perhaps the generic role has a canonical longname by which its package is named 16:23
moritz_ I'm trying to port some Ruby code to Perl 6, and... URKS. @points.extend Scruffy::Helpers::PointContainer unless @points.kind_of? Scruffy::Helpers::PointContainer
jnthn TimToady: That doesn't address it so much though
TimToady actually, the canonical name of Foo[$x] is just Foo, but Foo[Int] is the longname, I guess :) 16:24
jnthn TimToady: Because it's about instantiations of the role rather than the "role" itself (where a role as it appears in the program is really specifying a role factory that may produce many roles).
TimToady so Foo:: is the package of the role factory, but ::('Foo[Int]'):: or some such is the package of the instant 16:25
*instance
lue y oh hai o/
TimToady except, of course, Foo:: doesn't work terribly well for a multi-role 16:26
jnthn All roles are multi-roles in a sense 16:28
TimToady rakudo: role Foo[Int] { method x { say 1 } }; role Foo[Num] { method x { say 1.414 } }; class X does Foo[Int] {}; X.new.x
p6eval rakudo 692aa1: ( no output )
TimToady shouldn't that work?
heh, Could not find sub &say 16:29
jnthn > role Foo[Int] { method x { 1 } }; role Foo[Num] { method x { 1.414 } }; class
X does Foo[Int] {}; say X.new.x
1
Yes, it's a lexical chain bug with roles
pmurias TimToady: how should i go about adding the Perl6:: prefix to packages in STD? 16:30
jnthn I hoped pmichaud++'s fixes would help, but sadly not. :-(
TimToady rakudo: role Foo[Int] { method x { 1.say } }; role Foo[Num] { method x { 1.414.say } }; class X does Foo[Int] {}; X.new.x
p6eval rakudo 692aa1: OUTPUT«1␤»
TimToady rakudo: role Foo[Int] { method x { 1.say } }; role Foo[Num] { method x { 1.414.say } }; class X does Foo[Num] {}; X.new.x
p6eval rakudo 692aa1: OUTPUT«1.414␤»
jnthn Anyway, role Foo { } is really taken to be role Foo[] { } 16:32
That is, a role that takes no type parameters.
16:35 silug joined
TimToady pmurias: I thought you said you weren't changing STD 16:39
pmurias did I? 16:41
TimToady or maybe sorear++ indicated that 16:42
just packaging it in Perl6:: for CPAN
pmurias I created an installable src/STD-dist 16:43
TimToady which, up till now has been a namespace for flakey emulations of Perl 6
which STD arguably is pretty flakey :)
pmurias it's has some minor changes needed to specify search locations for some things and make one bug workaround work
moritz_ pmurias: if you give me a link to a document that describes how to install and run mildew, I can try to fix up the mildew: evalbot target
pmurias moritz_: that would be great 16:44
TimToady anyway, my intent is that someday when Perl 6 is real, STD be a real top-level package for any language derived from STD 16:45
pmurias TimToady: i want to add an Perl6:: prefix to stuff so that the dist can be place on CPAN
TimToady yes, well, that's kinda CPAN's problem
pmurias * placed
moritz_ pmurias: you can also prevent it from being indexed by appropriate entries in the META.yaml file 16:46
pmurias moritz_: it's intended that mildew will be installed with cpanm Mildew 16:47
TimToady if Perl 6 folks can't create new top-level package names without getting the agreement of CPANfolk, then I've got a problem with that.
pmurias i don't think CPANfolk would complain if dist built from STD-dist was uploaded on cpan 16:48
but in it's current form it would create a lot of top-level package names 16:49
and i'm not sure we want to use things like DEBUG and Stash
TimToady maybe we need real namespaces (and I don't mean packages in the parrot sense) 16:50
pmurias moritz_: dzill --install-command 'cpanm .' in src/STD-dist,v6/Mildew,v6/SMOP,v6/Mildew-Setting should install mildew
moritz_ pmurias: why not just no-index all files in the STD distribution, add an (indexed) .pod file explaining what it's all about, and be happy? 16:51
TimToady and I'd rather prefix all those types with STD:: than Perl6::, I think
it would be nice to have a namespace prefix mechanism that would make GLOBAL:: really mean GLOBAL::STD:: or some such, maybe 16:52
needs more thinking...but I hate ::-itis
pmurias moritz_: if you don't want to use Dist::Zilla i could supply a bunch of tarball 16:53
* tarballs
moritz_ pmurias: I'm installing dist::zilla now... shouldn't be a problem 16:54
TimToady and STD is called STD because it's not just for STD::P6, but also for STD::P5, STD::Ruby, STD::Python, STD::APL, etc 16:56
pmurias STD::P5 would be very handy for p5 users 16:57
TimToady it's half written already
ran out of tuits
moritz_ p5 users have PPI, no?
TimToady PPI, at least historically, tended to have no understanding of what the code actually means
moritz_ like me :-) 16:58
TimToady p6 needs a p5 parser so it can at least find the end of the { use v5; ... } block 16:59
pmurias and PPI is not extensible
TimToady and perhaps also run the block some day
pmurias TimToady: wouldn't it be possible to harness the perl5 parser for that? 17:00
TimToady std: rx:P5/[a-z]/ 17:01
p6eval std 31611: OUTPUT«ok 00:01 125m␤»
17:01 patspam left, patspam1 joined, patspam1 is now known as patspam
TimToady it uses that subset of it already 17:01
std: rx/[a-z]/ 17:03
p6eval std 31611: OUTPUT«===SORRY!===␤Invalid regex metacharacter (must be quoted to match literally) at /tmp/KcnLZdfvEt line 1:␤------> rx/[a-⏏z]/␤Potential difficulties:␤ [a-z] appears to be an old-school character class; please use <[a..z]> if you mean a character
..c…
pmurias moritz_: if you encounter any problems with installing Mildew mention them to me 17:04
TimToady pmurias: not easily, unless you teach the perl5 parser to treat an unmatched } as an EOF
moritz_ pmurias: just a question... can I easily install it into a location that's not the system perl, so that I can wipe it again easily?
pmurias local::lib 17:05
moritz_ I feared that answer
pmurias that's what i use
moritz_ I use it too
pmurias moritz_: what's the problem with that?
moritz_ which is the problem
I want to have a location for my modules
and a separate for mildew modules
so that I can wipe mildew stuff without wiping mine
pmurias i see
cpanm has and -l option for specifing which local::lib directory to install to 17:06
moritz_: do you think that will work? 17:08
moritz_ I'm trying it... 17:09
Unrecognized command: --install-command.
pmurias dzill install --install-command? 17:10
moritz_ right, just arrived at the same conclusion
pmurias * dzil
moritz_ huh. 17:12
# Failed test 'use STD;'
# at t/use_ok.t line 4.
# Tried to use 'STD'.
# Error: Can't locate File/ShareDir.pm in @INC
cpan[1]> install File::ShareDir 17:13
Going to read '/root/.cpan-5.12.1/Metadata' Database was generated on Fri, 09 Jul 2010 23:27:09 GMT
File::ShareDir is up to date (1.02).
pmurias: I have to leave now to buy some food before the stores close, will try again later or tomorrow
pmurias moritz_: ok 17:14
moritz_: fixed 17:17
pugssvn r31612 | pmurias++ | [STD-dist] add a missing prereq noticed by mortiz++
17:20 ashleydev joined 17:21 tedv joined
pugssvn r31613 | lwall++ | [specs] add C<also> declarator to prefix any trait to be applied to the outer declaration 17:21
r31613 | this lets us say "also is Int" without the 'is' conflicting with Test's is function
r31613 | and it's extensible as new trait_mods are added
pmurias moritz_: you don't have File::ShareDir?
moritz_ pmurias: I have 17:22
pmurias and the tests don't find it? 17:23
moritz_ pmurias: it's just... I have two perls, and currently no idea if dzil and cpanm pick up the right one
anyway, right now it's happily installing a huge dependency tree
pmurias looking at the @INC should clear some things up
moritz_ which works for me as long as no action on my part is required :-) 17:24
pmurias that's the intended installation experience ;) 17:25
moritz_ it's a piece of irony that I need DateTime::TimeZone even though none of the code works with time zones
but Params-Validate requires it, it seems 17:26
pmurias you are still at STD-dist?
moritz_ no, mildew
STD-dist was easy, because STD ran on the evalbot server previously
pmurias mildew has a lot of dependencies 17:27
moritz_: so you decided to postpone shopping?
moritz_ pmurias: my girlfriend decided for me :/ 17:28
still 32min left :-)
17:34 Mowah joined
pmurias moritz_: how's the installation going? 17:39
pugssvn r31614 | lwall++ | [basic.t] s/does/also does/ in the one spot that uses it 17:41
17:52 Mowah left 17:55 sftp joined 18:00 paroxyzm joined
lue rakudo: my Buf @a = 1,2,3; @a[3] = 4; say @a; say @a.WHAT 18:00
p6eval rakudo 692aa1: ( no output )
lue > my Buf @a = 1,2,3; @a[3] = 4; say @a; say @a.WHAT 18:04
maximum recursion depth exceeded
TimToady that's an array *of* Buf 18:05
not an array of Int
lue oh. I was under the impression that Buf was a type of array. 18:06
arnsholt Then you want my Buf $a
TimToady putting the type in front gives the type of each individual element
which works for $a because there's only one 18:07
paroxyzm Hi! Im interested in perl6 regexes.. i found that article: perlgeek.de/en/article/5-to-6#post_19 and it says : 'aaaa' ~~ m/ a | aaa | aa /;
lue rakudo: my Buf $a = 1,2,3; say $a
p6eval rakudo 692aa1: ( no output ) 18:08
paroxyzm On my pc the return value is 'a' not 'aaa'
lue rakudo: say I_AM_A_FAILURE; # seems like p6eval isn't returning errors 18:09
p6eval rakudo 692aa1: OUTPUT«Could not find sub &I_AM_A_FAILURE␤ in main program body at line 20:/tmp/fVRVwEgqHV␤»
lue rakudo: say 'aaaa' ~~ m/ a | aaa | aa / 18:12
p6eval rakudo 692aa1: OUTPUT«a␤»
tylercurtis Does Rakudo have LTM?
lue yes (IIRC), but it's quite clearly broken there :)
moritz_ only for constant prefixes of proto regexes 18:13
paroxyzm What was the meaning of that output?
moritz_ pmurias: ! Bailing out the installation for Mildew-0.01. Retry with --prompt or --force.
[DZ] [DZ] error running {{["cpanm -L /home/p6eval/mildew-install ."]}} at /usr/local/lib/perl/5.10.0/Moose/Meta/Method/Delegation.pm line 108
lue paroxyzm: that post is from 2008. I get the feeling things have changed since then. [as they frequently do in P6] 18:15
moritz_ paroxyzm: it's supposed to output 'aaa' 18:16
just not yet implemented yet
paroxyzm lue: Thanks :)
moritz_ lue: I try to keep those posts up-to-date
if you find something where it differs from current spec, please let me know
TimToady LTM is implemented in STD but not in rakudo 18:17
moritz_ and as I pointed out at the top of the page, I described spec status, not "is" status
jnthn TimToady: A while back you mentioned stuff about native types and part of it was Num.REPR. I've been working out how to slot this into my mental model of Perl 6...and slowly succeeding. But is the implication here that given a type, we can know its representation?
lue paroxyzm: I forgot to mention that sometimes it's the implementation's fault. :)
jnthn TimToady: That is, what happens if I go and Int.CREATE(:repr('P6Hash'))? 18:18
TimToady: But the real reason I'm asking this in a sense is that I've been struggling with fast attribute access in combination with not knowing the underlying representation.
paroxyzm lue: So what is the best implementation for regexes? 18:19
lue: anyway p6 will be big if those regexes are to become truth ;)
TimToady jnthn: if it's P6opaque, the repr can be anything you like, since it's opaque
jnthn TimToady: That is, knowing that the type is Dog, if we can Dog.CREATE(:repr<anything>), actually tells us nothing we can rely on at runtime.
And that means we hit real problems when trying to emit something more optimal than two names. 18:20
(the class and the attribute name)
lue paroxyzm: I can't say for sure, I'm a 'newb' myself :)
jnthn Or so far as I can see anyway.
I guess what I'm asking for is the REPR to be choosable by type. 18:21
paroxyzm lue: oh
18:21 justatheory left
jnthn So you commit to "this type will have this REPR" at class composition time. 18:21
18:21 rcsheets` left
jnthn But not by instance. 18:21
18:21 rcsheets` joined 18:23 paroxyzm left
TimToady you probably need to get the opinion of one of the other folks who have worked with Responder Interfaces, like ruoso++ or nothingmuch++ 18:23
jnthn Aye
TimToady possibly Dog with an alternative RI is just a different type somehow that doesn't pessimize what Dog is by default 18:24
18:24 rcsheets` left
TimToady using a different RI might just be a matter of mixing in a different .HOW 18:24
but I'm not the expert on anything hiding behind .HOW 18:25
TimToady has no knowhow
jnthn :-)
My problem isn't the existence of different representations. 18:26
It's more that not being able to know by type what representation I have. 18:27
The end result being that either everything is pessimal (too slow) or everything but P6opaque and a bunch of RIs for e.g. int, num etc are pessimal. 18:28
Which is OK for now but sucks for the guy who implements another REPR some day and then finds that the mechanism that is used to make P6opaque fast is useless. 18:29
(e.g. useless for his representation)
TimToady well, people who invent reprs had better know how to make them fast them own selves 18:30
jnthn Part of it is a code-gen thing though
TimToady I don't see how you can borrow much from something called "opaque"
it's opaque so we can cheat 18:31
if someone looks inside, they'll discover we're cheating :)
but there might be some value in declaring a default repr for a class 18:32
and we can even say "you can't override it" for now
jnthn Imagine we're code-generating an attribute lookup. We're in a single inheritance situation, so in theory we can do all attribute lookups by an offset into the object - eg. fast fast fast. We'd really like at compile time to go and say to the REPR "OK, I'm emitting a look for $!foo in class Bar, please give me a 'hint' that I can also emit in the code, if possible, and I'll give that back to you when we try and retreive the attribute at runtime" 18:33
Now, in an MI situation or we get augmenting going on and so on, it'll come back and say "sorry dave, I can't do that"
But in situations where it's possible, the REPR could give back something interesting. 18:34
If we can't make some assumption about the underlying REPR from the type object, then we either (a) can't do these things or (b) have to tag every hint with "oh and I got the hint from this REPR and if that ain't you, you'd best ignore it and just use the names" 18:35
Taht is, we can lazily pessimize, but it seems like it'll cost us a little to have the ability to do so. 18:36
TimToady such a type is, in essence, a native type in the sense that the repr is closed, though the class could add methods
and non-final, in that you can add more offsets under SI 18:37
18:37 timbunce left
TimToady rather than lazily pessimizing, I'd rather take the sum of what we know at CHECK time to know which reprs we can lock down 18:37
jnthn My point is that
sub foo(Dog $x) { ... } 18:38
The type constraint Dog isn't a constraint on the REPR that could be passed in
Unless we associate a single REPR with a type.
TimToady I think we should drive the repr choice to be just sufficiently declarative to know whether we can optimize at CHECK time 18:39
jnthn OK
TimToady much like most of our other optimizations that you can choose to pessimize if you declare you want to
class Dog is repr(*) or some such
otherwise default is repr(P6opaque) 18:40
would that be good enough?
tylercurtis rakudo: subset Odd of Int where * % 2; my Odd $i = 6; say 'Shouldn't happen.'
p6eval rakudo 692aa1: OUTPUT«===SORRY!===␤Confused at line 20, near "say 'Shoul"␤»
jnthn TimToady: Yes :-)
tylercurtis rakudo: subset Odd of Int where * % 2; my Odd $i = 6; say 'Shouldn\'t happen.'
18:40 timbunce joined
jnthn TimToady: Thanks. :-) 18:40
p6eval rakudo 692aa1: ( no output )
TimToady and .bless(:repr) is illegal unless declared is repr(*) 18:41
jnthn :-)
Excellent, that helps a lot.
TimToady we aim to blaze
jnthn Yes, I've been spending a lot of time recently thinking about that.
jnthn decides to switch focus on enjoying the world cup match for a bit :-) 18:42
TimToady "we aim to breeze" would rhyme better
lue rakudo: role Thing { ... }; role Stuff does Thing { method a { say 'a' } }; role Thing does Stuff { method b { say 'b' } }; Thing.a 18:45
p6eval rakudo 692aa1: ( no output )
TimToady .new maybe? 18:46
lue well, the error is actually maximum recursion depth exceeded :) 18:47
rakudo: role Thing { ... }; role Stuff does Thing { method a { say 'a' } }; role Thing does Stuff { method b { say 'b' } }; Thing.new.a
p6eval rakudo 692aa1: ( no output )
lue I'm surprised it doesn't error when A does B and B does A. It's only when I start trying to use it. 18:49
TimToady it wouldn't compose until .new forces it to compose a class, I imagine
lue It could be a wonderful collage of evil if it worked, though :) 18:50
TimToady and roles ignore up doeses
s/up/dup
a 'does' on a role merely says, "Be sure you invite my friends along too." 18:51
tylercurtis rakudo: role Thing { ... }; role Stuff does Thing { method a { say 'a' } }; role Thing does Stuff { method b { say 'b' } }; class ActualThing does Stuff { }; ActualThing.new.a 18:52
p6eval rakudo 692aa1: OUTPUT«===SORRY!===␤maximum recursion depth exceeded␤»
TimToady well, perhaps that's not how rakudo does it :) 18:53
jnthn: ^^
lue that's the first time an error came up in p6eval today! \o/ 18:54
pmichaud good morning, #perl6 18:55
lue (maybe 'does' should be smart enough not to invite itself)
guten tag, pmichaud o/
pmichaud picasaweb.google.com/lh/photo/XQe71...directlink # view from my balcony :-)
18:57 hercynium joined
lue ooh! aah! 18:57
[ I like the part Picasa put under the picture: "Sign in to like this photo" :) ]
the first P6 pun ever: m/smart/ # well, the core of the pun, anyway 19:00
TimToady is that Maui I see?
pmichaud It is Maui, yes. :-) 19:01
TimToady every island smells a little different 19:02
pmichaud my son says it smells like popcorn here 19:04
TimToady the big island usually smells of sulfur :)
19:05 cono joined 19:08 aubreyja_ joined
sorear good * #perl6 19:09
pmurias hi
moritz_: could you nopaste ~/.cpanm/build.log? 19:10
19:14 aubreyja_ left 19:17 ashleydev left
sorear jnthn: talk to me about reprs 19:19
jnthn: 'Dog' looks like a class but it's really treated as a role 19:24
so what I'm planning to do for the general case is to pass a Dog-vtable 19:25
with functions for calling all Dog methods, and indexes for all Dog slots
or rather, accessors
19:26 aubreyja joined
sorear also, I'm going to extend the Dog|U notation 19:26
Dog|U - protoobject repr 19:27
Dog|D - standard object repr
Dog|<name> - named repr
the autospecializer will, among other things, generate |D forms of methods
19:32 mtk joined
lue afk 19:46
19:48 kensanata joined, eternaleye joined
cognominal what is the way to ask a parametric rôle that the parameter C<does> some other role? 20:01
role MyRole[T does U] # something like that
I think that Range should be something like role Range[T does TotalOrder] { ... } ; role TotalOrder[U] { method succ() --> U { ... } ... } 20:05
20:10 masak joined
masak ahoy! 20:10
phenny masak: 15:31Z <moritz_> tell masak that I've pushed some fairly substantial changes to SVG, making it use methods internally instead of subs. My plan would be to rename it to XML::Writer or so, and then start a new SVG.pm, which automatically promotes some attributes into :style<...>, and optionally adds an SVG header
masak renames it now, lest he forget 20:11
tylercurtis masak: Bennu is about to have an actual runtime that makes the build process even more hellish. :) 20:12
yapsi: my $a; say $a;
p6eval yapsi: OUTPUT«Any()␤» 20:13
tylercurtis On the plus side, that will soon work. :)
masak I feel I lack context. what/who's Bennu?
and which hellish build process are we talking about? 20:14
20:14 Trashlord joined
masak moritz_: ping 20:15
tylercurtis masak: My very early-stage Perl 6 to LLVM compiler that uses SIC. The build process of compiling code using it. 20:17
masak oh, right!
sorry, I forgot its name.
tylercurtis: you should add it to the poc-projects.list file in the pls branch of proto.
20:18 mtk left, jferrero joined 20:19 mtk joined
tylercurtis Now, not only do you need yapsi, old rakudo, recent rakudo, and LLVM, you also need GNU libtool at /usr/local/bin/libtool(because I was too lazy to deal with the problem of OS X's libtool at /usr/bin(which is earlier in my path) being weird). :) 20:20
masak: Will do, once I finish pushing these changes.
masak \o/ 20:21
20:21 mtk is now known as mtk_
masak exciting times. I never dreamed of SIC compiling to LLVM. 20:21
20:21 mtk_ is now known as mtk
moritz_ masak: pong 20:22
masak moritz_: since I pinged, I concluded that the reason ufo wasn't working was because I broke it. nevermind. :) 20:23
it's fixed now.
moritz_ speaking of exciting times
Germany 3 -- 2 Urugay
masak congratulations.
football, right? :P 20:24
moritz_ right :-)
masak .oO( nailed it )
20:25 kfo joined, jferrero left
jnthn cognominal: role MyRole[::T where U] { ... } probably works 20:27
moritz_ pmurias: ah, found the problem: CHI won't install 20:28
Couldn't load class (CHI::Driver) because: Invalid version format (dotted-decimal versions require at least three parts) at /usr/lib/perl/5.10/version.pm line 194.
jnthn TimToady: Well, thing is that in role R1 { method m { } }; role R2 { method m { } }; role R3 does R1 does R2 { method m { } }, iiuc we'd like that to eliminate the methods from R1 and R2. 20:29
So it's not just as simple as a completely flat "and add these too" if we wnat this to work. 20:30
If that can fail and we get a conflict when we compose R3 into a class, well, fine, but I suspect that's not what's wanted.
pmichaud: Nice photo!
20:31 aubreyja left
masak moritz_: I don't like BEGIN { @*INC.push: 'lib', '../lib' }; 20:31
moritz_: and with ufo, it's not needed.
moritz_ masak: feel free to remove
masak removes
20:32 ashleydev joined
moritz_ CHI versions 0.34 and 0.35 fail with the same error message 20:32
tylercurtis masak: would you prefer a pull request or a patch? 20:33
masak tylercurtis: you're saying you don't have commit access?
:)
moritz_ this feels like it could be easy to fix... 20:34
masak tylercurtis: what's your name on github?
tylercurtis masak: ekiru
masak tylercurtis: welcome aboard.
tylercurtis: just push to poc-projects.list on the pls branch. 20:35
tylercurtis masak: done. 20:36
masak tylercurtis: ...and you got the comma on the line before. :) tylercurtis++
json--
moritz_: rename complete. 20:40
tylercurtis masak: Does JSON support an extra comma? 20:41
moritz_ no :(
masak hence the --
moritz_: could you also add yourself to AUTHORS in xml-writer?
moritz_ javascript does, but a brain-dead browser does not
masak: yes... and rename it in hugme 20:42
wait, the other way round :-)
masak ...and change the url in .git/config :)
moritz_ currently I don't have commit access
masak oh?
moritz_ stupid
ignore me
masak does so 20:43
moritz_ lib/XML/Writer.pir | 2123 +++++++++++++++++++++++++++++++++++++++++
doen't look like you need to ship it
lue ohai o/ 20:44
moritz_ oh, svg wasn't on hugme
masak kills off the .pir file 20:45
moritz_ faster
masak :)
indeed.
lue rakudo: my $a = Buf.new('N', 'E', 'S', 0x1A); say $a 20:46
p6eval rakudo 692aa1: ( no output )
masak lue: Buf.new takes a list of integers.
actually, right now it takes an array of integers. 20:47
that's likely to change.
tylercurtis rakudo: my $a = Buf.new([1,2,3,4,5]); say $a
p6eval rakudo 692aa1: ( no output )
moritz_ rakudo: $*OUT.write: Buf.new([:16<6d>, :16<c3>, :16<70>]) 20:48
masak tylercurtis: locally, that says "Buf()<0x10c9161e0>"
p6eval rakudo 692aa1: ( no output )
jnthn masak: Oh?
masak: Relying on slurpy laziness in Buf?
masak jnthn: yes. pmichaud suggested it should be like it was from the beginning. 20:49
jnthn: hold on, I'll dig up the blog post.
jnthn: use.perl.org/~masak/journal/40422
last paragraph.
moritz_ masak: I think allowing named arguments to XML::Writer.serialize was a bad idea 20:50
jnthn oh
masak moritz_: ok.
jnthn Well consistency is a good point, but my point was also good. :-)
moritz_ masak: because now I want to subclass it, and find myself duplicating the arugment extraction logic
jnthn masak: Maybe you want an alternative constructor.
masak jnthn: pmichaud had a counter-point against yours. do read the logs. 20:51
20:51 patspam left
masak moritz_: sometimes I feel the whole way named arguments and pairs rub against each other is "wrong". but it's been a long time now since I wrote code that made me feel that. 20:52
jnthn masak: OK, iirc, in encode you get a byte-buffer?
masak yes.
jnthn oh wtf 20:53
you're not doing at all what I thought
masak get a fresh bytebuffer, assign the string to it, and iterate over its bytes.
jnthn OK
...yes, that's exactly what I wanted us to avoid.
masak ok.
I haven't realized that till now. 20:54
jnthn ByteBuffer is capable of being index into positionally and keeps a reference to the original string.
Meaning that .encode doesn't allocate a load of new storage
If the Buf just references a ByteBuffer
Yes that's not the _only_ case.
masak hm, that is indeed nicer. 20:55
jnthn But I'd have thought it was a common enough one to make sure we avoid copying.
masak but it'll cease to work as soon as the Buf is treated as an array.
yes, I agree that it's a common case, and worth optimizing for.
jnthn Oh
has T @.contents; 20:56
Hmm
Ok, leave it as it is for now I guess
This is going to be hard to get right until we have compact arrays.
Or hard to get right and efficient.
And working trumps fast and wrong.
But just a heads up that we should re-visit this in the future to avoid so much copying. 20:57
(Looking at it more, it's not quite trivial.)
masak aye. 20:58
jnthn our multi sub infix:<eqv>(Buf $a, Buf $b) { return $a.contents ~~ $b.contents;
}
Naybe use eqv in there? 20:59
masak does it matter?
moritz_ seems more robust with eqv
masak oki, changing.
moritz_ if you have a Buf[Any], and one of the entries is a *
20:59 kensanata left
jnthn moritz_: Right. 20:59
masak good point.
21:01 dju joined 21:02 timbunce left
lue rakudo: my $a = Buf.new([0x4E, 0x45, 0x53]); say $a.decode 21:02
p6eval rakudo 692aa1: ( no output ) 21:03
21:03 dju left, dju joined
lue > my $a = Buf.new([0x4E, 0x45, 0x53]); say $a.decode 21:03
NES
masak \o/ 21:06
that's about what the tests do, too.
jnthn: did you find pmichaud's objection that speed might not be an issue after all for slurpies? 21:07
21:08 timbunce joined
moritz_ github.com/moritz/svg 21:11
dalek kudo: 7579f7a | masak++ | src/core/Buf.pm:
[Buf] infix:<eqv> for Bufs now uses eqv internally
jnthn masak: In the long run maybe it won't be 21:14
The trouble is that | is not always too smart at the moment.
But if you avoid that you're fine.
masak I see two use cases. 21:15
Buf.new(byte, byte, byte)
and Str.encode($encoding)
21:16 whiteknight joined
masak the latter might have to use | if we only allow the slurpy constructor. 21:16
21:16 tadzik joined
moritz_ why not allow both constructrs, as multis? 21:16
masak moritz_: that's a possibility.
moritz_: it falls, I believe, under pmichaud's "API based on Rakudo's current characteristics" 21:17
which may or may not be considered a disadvantage.
jnthn masak: But if you have a slurpy constructor and you pass an array, it'll get flattened into the slurpy
masak oh.
even if you provide both, as moritz_ suggests? 21:18
jnthn rakudo: sub foo(*@x) { say @x.elems }; my @a = 1,2,3; foo(@a);
p6eval rakudo 692aa1: OUTPUT«3␤»
moritz_ rakudo: multi f(@a) { }; multi f(*@a) { }; f(my @a = (1, 2, 3))
p6eval rakudo 692aa1: OUTPUT«===SORRY!===␤Redeclaration of symbol @a at line 20, near " = (1, 2, "␤»
jnthn huh
moritz_ doesn't look like a redeclaration to me
rakudo: multi f(@a) { }; multi f(*@a) { }; f(my @x = (1, 2, 3))
p6eval rakudo 692aa1: OUTPUT«===SORRY!===␤Redeclaration of symbol @x at line 20, near " = (1, 2, "␤» 21:19
moritz_ rakudo: multi f(@a) { }; multi f(*@a) { }; my @x = (1, 2, 3); f(@x)
p6eval rakudo 692aa1: ( no output )
jnthn moritz_: You could do that
er
ingy greetings
jnthn masak ^^
masak: But just the slurpy one and don't | the thing going in should be efficient today
o/ ingy 21:20
ingy \o jnthn
masak moritz_: rakudobug?
moritz_ masak: yes 21:21
masak jnthn: then it sounds to me I should change back to the just the slurpy.
masak submits rakudobug
jnthn masak: Aye 21:22
masak: For consistency too
masak: We can optimize further later. 21:23
jnthn afk for a bit
21:23 araujo joined 21:24 araujo left, araujo joined
sorear When do the executable statements in the body of a class { } run? 21:24
moritz_ is pretty sure that's in S12 somewhere
lue any reason why Buf -> Str is decode(), and Str -> Buf is encode()? That seems backwards to me [ not that it matters at all, just wondering ]
masak rakudo: multi f(@a) { }; multi f(*@a) { }; my @x; f(@x = (1, 2, 3)) 21:25
p6eval rakudo 692aa1: OUTPUT«===SORRY!===␤Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 20␤»
masak o.O
moritz_ rakudo: multi f(@a) { }; my @x; f(@x = (1, 2, 3))
p6eval rakudo 692aa1: OUTPUT«===SORRY!===␤Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 20␤» 21:26
moritz_ std: multi f(@a) { }; my @x; f(@x = (1, 2, 3))
p6eval std 31614: OUTPUT«ok 00:01 117m␤»
masak rakudo: multi f(*@a) { }; f(my @x = (1, 2, 3))
p6eval rakudo 692aa1: OUTPUT«===SORRY!===␤Redeclaration of symbol @x at line 20, near " = (1, 2, "␤»
masak rakudo: multi f(@a) { }; f(my @x = (1, 2, 3))
p6eval rakudo 692aa1: OUTPUT«===SORRY!===␤Redeclaration of symbol @x at line 20, near " = (1, 2, "␤»
21:26 whiteknight left 21:27 takadonet joined
takadonet hey everyone 21:27
moritz_ o/
rakudo: f(my @x = (1, 2, 3))
p6eval rakudo 692aa1: OUTPUT«===SORRY!===␤Redeclaration of symbol @x at line 20, near " = (1, 2, "␤»
moritz_ rakudo: f(my @x)
p6eval rakudo 692aa1: OUTPUT«Could not find sub &f␤ in main program body at line 20:/tmp/Z7mIzm_qJb␤»
masak moritz_++ 21:29
so it has nothing to do with multis, only with the assignment in the argument list.
rakudo: f(my $x = (1, 2, 3)) 21:30
p6eval rakudo 692aa1: OUTPUT«Could not find sub &f␤ in main program body at line 20:/tmp/mFoG7yiTUW␤»
moritz_ list assignment :-)
masak rakudo: f(my %x = (1, 2, 3))
p6eval rakudo 692aa1: OUTPUT«===SORRY!===␤Redeclaration of symbol %x at line 20, near " = (1, 2, "␤»
takadonet for loop are not lazy yet correct?
in conjunction with a gather/take 21:31
moritz_ they are, to some extend
rakudo: for 1..* {.say; last }
takadonet i.e: my @matches = gather for $/<fasta> -> $m {take $m.ast;}
p6eval rakudo 692aa1: OUTPUT«1␤»
21:32 silug left
takadonet could be something else then that is making it eager 21:32
moritz_ takadonet: that' easier written as my @matches = $<fasta>.map: { *.ast }
takadonet: and do use a current development version of rakudo - in the last release, array assignment was still eager
takadonet moritz_: I'm always using the newest development version! :) 21:34
moritz_ takadonet: great :-)
takadonet moritz_: code fails if I changed to using a map :( 21:35
moritz_ weird 21:36
takadonet Really could be the way I written it 21:37
21:41 whiteknight joined
masak goes home to sleep 21:42
thought I would get some Yapsi runtime hacking done today, but there wasn't enough consequtive thinking time.
ah well, maybe tomorrow.
'night.
lue goodnight masak o/
21:42 masak left
pmurias moritz_: the CHI dependency is not esential 21:46
lue rakudo: role B { has $.var = 3 }; class A does B { method z { say $.var;} }; A.z
21:46 kloeri left
p6eval rakudo 692aa1: ( no output ) 21:46
sorear lue: a byte sequence for text is useless except to store and transmit, so we call it encoded 21:47
decoding the bytes produces characters, which can be processed in many more ways
21:47 kloeri joined
pmurias moritz_: any other solutions forthe CHI problem other than making Mildew stop using it? 21:48
lue ah, ok.
sorear CHI?
pmurias sorear: i use it for caching the parse 21:49
sorear: it's a CPAN module 21:50
moritz_: if you comment the dependency out in dist.ini mildew will work just fine (with the default frontend) 21:51
21:52 mtk left, kloeri left
lue for class A does B why can I not access variables in B through A? 21:54
(or, more accurately, attributes)
21:56 takadonet left
tylercurtis rakudo: role B { has $.foo is rw; }; class A does B {}; my $a = A.new; A.foo = 'foo'; A.foo.say; 21:56
p6eval rakudo 692aa1: OUTPUT«Type objects are abstract and have no attributes, but you tried to access $!foo␤ in main program body at line 70␤»
tylercurtis rakudo: role B { has $.foo is rw; }; class A does B {}; my $a = A.new; $a.WHAT.say; $a.defined.say; 21:57
p6eval rakudo 692aa1: OUTPUT«A()␤1␤»
tylercurtis Hmm... I don't know. 21:58
lue rakudo: role B { has $.foo is rw = 3; }; class A does B { method z { say $.foo } }; A.z
p6eval rakudo 692aa1: ( no output )
21:59 kloeri joined 22:03 hsb joined
lue rakudo: role B { has $.foo is rw = 3; }; class A does B { method z { say B.foo } }; A.z 22:04
p6eval rakudo 692aa1: ( no output )
lue can Someone Who Knows Why™ explain why I'm not allowed to access B's attributes in A? 22:09
nothingmuch Q/sb end
jnthn lue: you can if you use the right syntax ;-) 22:10
lue I just found it. 22:11
jnthn lue: When you compose a role into a class, the attrs are just as if you had declared them in the class
So you can see $.foo in the method in the class too :-)
s/see/use/
lue rakudo: role B { has $.foo is rw = 3; }; class A does B { has B $.foo; method z { say $.foo } }; A.z
p6eval rakudo 692aa1: OUTPUT«===SORRY!===␤Attribute '$!foo' already exists in the class, but a role also wishes to compose it␤»
tylercurtis rakudo: role B { has $.foo is rw; }; class A does B {}; my $a = A.new; A.foo = 'foo'; A$a.foo.say;
p6eval rakudo 692aa1: OUTPUT«===SORRY!===␤Confused at line 20, near "A$a.foo.sa"␤»
tylercurtis rakudo: role B { has $.foo is rw; }; class A does B {}; my $a = A.new; $a.foo = 'foo'; $a.foo.say; 22:12
lue ...or not
p6eval rakudo 692aa1: OUTPUT«foo␤»
jnthn lue: You re-declared it
See what tylercurtis++ did :-)
lue ah. What I want to do is write methods in class A that mess with attributes in role B, but I can't seem to do that. 22:13
jnthn Sure you can
lue: The thing is that by composing a role the attributes are in your class now 22:14
lue: Thing of role composition like copying stuff from the role into your class.
*Think
tylercurtis rakudo: role B {has $.foo is rw; }; class A does B { method double_foo { $.foo *= 2; } }; my $a = A.new(:foo(5)); $a.foo.say; $a.double_foo; $a.foo.say;
p6eval rakudo 692aa1: OUTPUT«5␤10␤»
lue rakudo: role B { has $.foo = 3 }; class A does B { method zi { say $.foo } }; A.zi 22:15
p6eval rakudo 692aa1: ( no output )
jnthn A.new.zi 22:16
Otherwise the class has no instance attributes :-)
rakudo: role B { has $.foo = 3 }; class A does B { method zi { say $.foo } }; A.new.zi
p6eval rakudo 692aa1: OUTPUT«3␤»
lue OK. What was messing me up was bad coding then :)
22:18 kloeri left
lue
.oO(probably the closest thing to a jumptable in P6 is given/when (unless there's magic I'm not aware of)...)
 22:18
22:19 mtk joined 22:21 Sanitoeter joined 22:25 tadzik left 22:30 kloeri_ joined
TimToady given/when is designed to be optimizable to a jumptable if you have well-behaved values in the when bits 22:37
22:37 rv2733 left
TimToady that was the big reason for changing ~~ from symmetrical to asymmetrical 22:37
lue I'll only need 256 when's, from 0 to 0xFF :) 22:38
TimToady otoh, you can also use an array of functions 22:39
lue rakudo: sub A { say 'a' }; sub B { say 'b' }; my @a = A(), B(); @a[1]
p6eval rakudo 692aa1: OUTPUT«a␤b␤»
TimToady I doubt anyone is going to do the jumptable optimization any time soon, since (afaik) it was *never* done in Perl 5 (unlike Perl 4)
but maybe it'll be easier in 6 22:40
lue I gave up Python whilst coding my NES emulator because I got to the opcodes and found Python doesn't come with switch statements.
TimToady since it's just a matter of running down the statements and seeing if they're all integer whens, or string whens
lue
.oO(255 elseif's... *shudder*)
 22:41
TimToady in other words, Python's only way to do it didn't line up with your only way to do it. :)
presumably P6 can also do it with multi opcode (42) {...} 22:42
22:45 mtk left, timbunce left
lue I grappled between P5 and 6 for an afternoon, looking for something better, and I went with 6 because the -> (as opposed to .) confused me greatly. The End :) 22:46
sorear moritz_: do you know who changed p6eval to suppress error messages and why? 22:47
lue I should do multi opcode. It's extermely lazy! (after coding 256 opcode() methods, of course) \o/ 22:48
jnthn lue: Nah, pre-deref'd interpreter ftw. :-)
(you take the array of instructions and build an array of sub references, then just go through that array).
lue rakudo: multi method op(1) { say 'do' }; multi method op(2) { say 're' }; my $a = 2; op($a) 22:49
p6eval rakudo 692aa1: OUTPUT«Useless declaration of has-scoped Method in a module; add our or my to install it in the lexpad or namespace␤Useless declaration of has-scoped Method in a module; add our or my to install it in the lexpad or namespace␤Could not find sub &op␤ in main program body at line
..20:/tmp/85…
lue rakudo: multi sub op(1) { say 'do' }; multi sub op(2) { say 're' }; my $a = 2; op($a)
p6eval rakudo 7579f7: OUTPUT«re␤» 22:50
jnthn :-)
lue: also multi op(1) { say 'do' }
lue: multi name is just like multi sub name
but shorter
:-)
lue I'm glad I dumped Python! \o\ /o/
rakudo $satisfiedcustomer++ 22:51
jnthn Yay, I has @nice-hotel booked for $vacation :-)
Which ain't for a month and a half. Eh well. 22:52
lue how great is @_ ?
jnthn @_ ? 22:54
22:54 azert0x left
tylercurtis rakudo: say TimeMachine.new.travel(Date.today + 45).find(:person<jnthn>).WHERE 22:54
p6eval rakudo 7579f7: OUTPUT«Could not find sub &TimeMachine␤ in main program body at line 20:/tmp/v3NsTvtXeF␤» 22:55
lue rakudo: enum Vacancy <Vacant Booked>; my @nice-hotel; @nice-hotel[208] = Booked; # hotel management made esay by P6!
p6eval rakudo 7579f7: ( no output )
lue s/esay/easy/
rakudo: use TARDIS :ver<40TT>; # please oh please oh please... 22:56
p6eval rakudo 7579f7: OUTPUT«===SORRY!===␤Unable to find module 'TARDIS' in the @*INC directories.␤(@*INC contains:␤ lib␤ /home/p6eval/.perl6/lib␤ /home/p6eval//p2/lib/parrot/2.5.0-devel/languages/perl6/lib␤ .)␤»
tylercurtis lue: github.com/masak/tardis someday. 22:57
jnthn tylercurtis: <Germany Austria Slovenia>.any :-) 22:58
arnsholt std: given 5 -> $x { ... }
p6eval std 31614: OUTPUT«ok 00:01 113m␤»
arnsholt rereads S04 22:59
tylercurtis rakudo: given 5-> $x { say $x }
sorear lue: yer syntax is wrong anyway
p6eval rakudo 7579f7: OUTPUT«===SORRY!===␤Unsupported use of -> as postfix; in Perl 6 please use either . to call a method, or whitespace to delimit a pointy block at line 20, near " $x { say "␤»
tylercurtis rakudo: given 5 -> $x { say $x } 23:00
p6eval rakudo 7579f7: OUTPUT«5␤»
arnsholt rakudo: given 5 -> $x { when 5 { say "da" } }
lue tylercurtis: that's a time-travelling debugger.
p6eval rakudo 7579f7: OUTPUT«Use of uninitialized value in numeric context␤»
lue (That reminds me, I need to finish my Time Traveller's Tense Handbook, so tardis can speak properly)
arnsholt Wha'?
jnthn arnsholt: Well, you bound it to $x rather than the defualt $_, and when goes looking at $_. :P 23:01
arnsholt Oh, probably due to $_ being undef
Yeah, what you said ^^
tylercurtis lue: Do you have a tardis of your own you're working on?
jnthn I *think* Rakudo isn't meant to go binding it to both.
lue not yet, I was referring to masak's. 23:02
arnsholt As far as I can make from S04 an pblock with given doesn't make a lot of sense, but STD appears to disagree
lue [you're welcome to give me a fully-equipped genetics labratory though. Then I can start on my TARDIS :)] 23:03
23:08 mtk0 joined
sorear sighs. 23:10
23:14 rokoteko joined 23:17 mtk0 left 23:18 mtk0 joined, mtk0 left, pmurias left 23:19 mtk joined 23:25 kloeri_ is now known as kloeri
arnsholt I'm thinking of having a whack at giving NQP given/when 23:32
But before I get too deep into this, does anyone in the know know if this likely to be a desired feature?
tylercurtis would use it. 23:33
arnsholt Good that I'm not the only one, at least =)
sorear probably it would be best to implement it semi-syntactically 23:37
arnsholt What do you mean?
sorear pmichaud wouldn't like adding ACCEPTS to Integer, so when 42 { should desugar to $_ == 42, not 42.ACCEPTS($_)
arnsholt Ah, right 23:38
NQP has ~~ though, so as a first approximation I think it should work
sorear I can't find anything in S12 or the S12 tests on when class Foo { say "hi" } prints 23:52
23:53 masonkramer joined
sorear BEGIN, ENTER, and DO are the main contenders in my mind 23:53
23:55 Psyche^ joined 23:58 envi^home joined 23:59 Psyche^ is now known as Patterner

Powered by Cro and the Raku® Programming Language.

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

Thank you!