»ö« | perl6.org/ | nopaste: paste.lisp.org/new/perl6 | 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 lichtkind on 5 March 2010.
00:00 cdarroch left 00:01 szabgabx joined 00:16 lest_away is now known as lestrrat 00:19 azert0x left 00:29 snarkyboojum left 00:38 Schwern joined 00:39 gbacon left 00:54 adhlssu07 joined
pugssvn r31259 | lwall++ | [STD] don't give the wrong advice on sub() 01:04
01:05 whiteknight left 01:08 Visitor99 joined 01:16 adhlssu07 left 01:17 alester joined 01:23 stepnem left 01:27 snarkyboojum joined 01:39 Visitor99 left
snarkyboojum what's the best url to point someone at when attempting to explain what Rakudo Star will/might mean? use.perl.org/~pmichaud/journal/39411? 01:48
pmichaud so far, yes.
snarkyboojum pmichaud: cheers :) 01:49
pmichaud question for the channel: I think our .perl may be a little too naive as specced (more)
consider:
my $r = 1..5; my @r = $r, 3; say @r.perl
what should be the output?
currently I'm voting in favor of "([1..5], 3)" 01:50
(although this is not at all what the spec says at the moment.) 01:51
diakopter what's the argument against [[1..5], 3]
pmichaud okay, I've changed my mind :-)
01:51 colomon joined
pmichaud now I'm going to argue in favor of ( (1..5).item, 3 ) :-) 01:51
one argument against [[1..5], 3] is that it doesn't contain a Range :) 01:52
snarkyboojum I'd find it confusing that it seems to say that @r contains an array, when 1..5 is a range 01:53
or am I misunderstanding what @r contains? :)
pmichaud @r is an array of two elements
the first element is a non-flattening range, the second element is a 3
snarkyboojum pmichaud: yep, but 1..5 is still a range isn't it?
pmichaud yes, but if you were to eval the string "[1..5, 3]" you souldn't end up with a 2-element array. 01:54
snarkyboojum ah ok
pmichaud and one of the claimed purposes of .perl is that you can evaluate the returned string back to the same object.
snarkyboojum right :) 01:55
diakopter rakudo: [1..3].item.WHAT.say
p6eval rakudo cfbeb5: OUTPUT«Array()␤»
01:55 Schwern left
diakopter item should return a range? 01:55
01:56 JeffreyKegler joined
pmichaud you put it inside of an array 01:56
[1..3] is an Array containing a lazily evaluated array
diakopter rakudo: (1..3).item.WHAT.say # oh
pmichaud [1..3].item is that same Array
p6eval rakudo cfbeb5: OUTPUT«Range()␤»
TimToady yes, the $r suppresses flattening in the list assignment
diakopter rakudo: ( (1..5).item, 3 ).say 01:57
p6eval rakudo cfbeb5: OUTPUT«123453␤»
01:57 Schwern joined
diakopter rakudo: ( (1..5).item, 3 ).perl 01:57
p6eval rakudo cfbeb5: ( no output )
diakopter rakudo: ( (1..5).item, 3 ).perl.say
p6eval rakudo cfbeb5: OUTPUT«(1..5, 3)␤»
pmichaud that's wrong in current master, probably fixed in my list branch
well, it's closer in my list branch 01:58
TimToady rakudo: (1..5).say
p6eval rakudo cfbeb5: OUTPUT«12345␤»
pmichaud > ( (1..5).item, 3).perl.say
(1..5, 3)
diakopter (that's what it says above)
pmichaud anyway, I'm thinking that .perl ought to be smart enough to distinguish items versus flattening objects and dtrt
(I've started doing this in the new branch, but wanted to ask the channel before doing anything much more radical with it) 01:59
01:59 Visitor99 joined
TimToady I'd argue that should say '1..5' 02:00
sorry 'bout the lag--my family is microwaving on 2.4Ghz :/
pmichaud that can be done also :-)
TimToady I mean (1..5).say should say that 02:01
pmichaud right
what about ~(1..5) ?
is that still "1 2 3 4 5"?
TimToady Stringy can be different from Str
.say is presumably forcing .Str 02:02
Visitor99 rakudo: my Rat $rat=1.2; my Int $int=5; my Str $str="string!"; my @arreglo = ($rat, $int, $str); for @arreglo { when ($_~~Rat) { say "$_ =rational"; } when ($_~~Str) { say "$_ =string"; } when ($_~~Int) { say "$_ =integer"; } }
p6eval rakudo cfbeb5: OUTPUT«string! =rational␤»
TimToady or some other-level stringifier that isn't ~
Visitor99 that line works ok when using "if" instead of "when"
i had the idea that "when" was similar to "if". 02:03
02:03 JimmyZ joined
pmichaud Visitor99: note that you don't need the $_ ~~ in the when clauses. 02:03
colomon rakudo: my Rat $rat=1.2; my Int $int=5; my Str $str="string!"; my @arreglo = ($rat, $int, $str); for @arreglo { when Rat { say "$_ =rational"; } when Str { say "$_ =string"; } when Int { say "$_ =integer"; } }
p6eval rakudo cfbeb5: OUTPUT«1.2 =rational␤5 =integer␤string! =string␤»
Visitor99 oh,,,, cool.. :) 02:04
TimToady otherwise there would be little point in given/when
pmichaud rakudo: for 1.2, 5, 'string!' { when Rat { say "$_=rational" }; when Str { say "$_=string"; }; when Int { say "$_=integer"; } }
p6eval rakudo cfbeb5: OUTPUT«1.2=rational␤5=integer␤string!=string␤»
02:04 JeffreyKegler left
diakopter heh != 02:04
snarkyboojum rakudo: my $r = 1..5; my @r = $r, 3; @r.map: { .WHAT } 02:05
p6eval rakudo cfbeb5: ( no output )
snarkyboojum rakudo: (eval '[[1..5], 3]').map: { say .WHAT } 02:06
p6eval rakudo cfbeb5: ( no output )
snarkyboojum hehe I give up :)
pmichaud Visitor99: anyway, "if" tests a boolean condition, while "when" does a smartmatch 02:07
also, a matching "when" skips to the end of the block, while an "if" will fall through to continue executing the rest of the block
02:10 [mark] joined
TimToady but it's not clear that ~(1..5) should do the same as ~(1...5), since .. has a valid scalar meaning of "interval" 02:10
pmichaud sure, I can buy that 02:11
at the moment I was more interested in the proper handling o f.perl
er, of .perl
TimToady I'm inclined to be conservative about eager listification in the absence of a list context 02:12
scalars should generally behave as scalars unless provoked
diakopter rogue scalars are scary 02:13
TimToady so ~(1..5) may well be different from ~(1..5,)
pmichaud well, as far as .perl goes, I think rakudo gets that one right-ish atm 02:14
rakudo: say (1..5).perl
p6eval rakudo cfbeb5: OUTPUT«1..5␤»
diakopter suddenly regrets spending a week straight fuzzing rakudo on series & range literals.
pmichaud rakudo: say (1..5,).perl
p6eval rakudo cfbeb5: OUTPUT«(1..5, )␤»
pmichaud it's the array case that I find a bit more interesting for .perl 02:15
TimToady rakudo: say (1..5,)
p6eval rakudo cfbeb5: OUTPUT«12345␤»
TimToady rakudo: say (1..5)
p6eval rakudo cfbeb5: OUTPUT«12345␤»
TimToady that one's wrongish
pmichaud oh?
the 1..5 doesn't interpolate in list context?
diakopter rakudo: say (1..5,,Nil) 02:16
p6eval rakudo cfbeb5: OUTPUT«12345␤»
TimToady oh, you're right
that is a list context
rakudo: (1..5).say
p6eval rakudo cfbeb5: OUTPUT«12345␤»
TimToady that one is wrongish
pmichaud I'm willing to say that should be different
I wouldn't have a problem with ranges stringifying to something else
and presumably rakudo: say (1..5).item # "1..5", not "12345" or "1 2 3 4 5" 02:17
TimToady correct
pmichaud so, when say stringifies the items of its list, it uses Stringy or Str ?
TimToady Str 02:18
pmichaud and same for interpolation in double-quotes?
TimToady ~ is high-level, Str is low
pmichaud my $r = 1..5; say "$r"; # ??
TimToady well, ~(1..5) may also be '1..5' 02:19
more interesting is junctions
~(1|2|3) is "1"|"2"|"3" but (1|2|3).Str is 'any(1,2,3)'
but I'd say string interpolation is Stringy 02:20
not Str
"$r" is sugar for ~$r
pmichaud okay, and infix:<~> is also using Stringy interpretations of things 02:21
TimToady nod
pmichaud I have a feeling that's going to confuse people :) 02:22
that say $r would be different from say "$r"
or, more precisely, that say $r, " hello" will end up being different from say "$r hello"
TimToady if $r is a range, those are the same output 02:23
assuming range is always "1..5"
pmichaud oh, so in the case of range, .Stringy and .Str end up being the same 02:24
TimToady but ... is likely to be different
that's what I'm thinking, since Ranges are intervals first
pmichaud okay
TimToady and only lists of integers secondarily
whereas a Series is always implicitly listy
pmichaud we'll need to get some tests in place for some of this :-)
TimToady we've been detangling .. and ... for a while now :) 02:25
pmichaud and strings, too.
anyway, I'm hoping the new list implementation will help us in our detanglings.
it's definitely resulting in much cleaner code overall.
TimToady \o/ 02:26
pmichaud and it's no longer a question of "what methods do we need" but rather "which classes should have which ones?"
colomon \o/
JimmyZ what does detanglings mean?
pmichaud And there's usually only one or two obvious places for each.
so I'm pretty happy with the overall API.
TimToady JimmyZ: do you know "tangle"?
JimmyZ yes 02:27
thanks.
TimToady detangling is removing the tangle :)
JimmyZ thanks again
TimToady np, you're English is much better than my Mandarin :) 02:28
*your
my English ain't so good either...
JimmyZ knows that TimToady's perl is much much much better than him 02:35
diakopter LOLz at indefinite 'him' antecedent 02:36
JimmyZ knows his english is poor too, wrong wording. 02:37
diakopter I wasn't laughing at you :) your english was clear
JimmyZ ^_^
diakopter properer english would be "his" instead of "him"
JimmyZ yeah 02:38
02:39 JimmyZ left 02:44 JimmyZ joined
Visitor99 the rakudo that is used when i invoke rakudo: in this irc channel, it's the same that is available via github? 02:49
colomon Visitor99: yes 02:51
unless it's gotten behind.
rakudo: say 1
p6eval rakudo cfbeb5: OUTPUT«1␤»
colomon okay, it's up to date at the moment. 02:52
02:54 masonkramer joined
TimToady std: sub() {} 02:54
p6eval std 31259: OUTPUT«===SORRY!===␤Word 'sub' interpreted as 'sub()' function call; please use whitespace around the parens at /tmp/tmmJNnExhn line 1:␤------> sub⏏() {}␤Unexpected block in infix position (two terms in a row) at /tmp/tmmJNnExhn line 1:␤------>
..su…
TimToady looks like std is also up-to-date
diakopter std: sub{}
p6eval std 31259: OUTPUT«ok 00:01 104m␤»
02:56 gbacon joined, diakopter sets mode: +oooo colomon ingy sorear Juerd 02:57 TimToady sets mode: +vv buubot dalek, TimToady sets mode: +vv ilogger2 IRSeekBot, TimToady sets mode: +vv p6eval phenny, TimToady sets mode: +v pugssvn 03:02 stepnem joined
snarkyboojum could someone give me some advice on what I'm doing wrong here? gist.github.com/438641 - I get 'Null PMC access in invoke()' 03:09
I'm sure it's a UTS error (User Too Stupid) :) 03:12
pugssvn r31260 | sorear++ | [dfa/Cursor] Refactor & implement quantifiers.
03:15 masonkramer left
Visitor99 rakudo: my $a; say $a.WHAT; $a = 1; say $a.WHAT; $a = Mu; say $a.WHAT; 03:21
p6eval rakudo cfbeb5: OUTPUT«Any()␤Int()␤Mu()␤»
03:21 plobsing joined
Visitor99 rakudo: my $a; say $a.WHAT; $a = 1; say $a.WHAT; $a = Mu; say $a.WHAT; $a = Any; say $a.WHAT; 03:25
p6eval rakudo cfbeb5: OUTPUT«Any()␤Int()␤Mu()␤Any()␤»
Visitor99 my $a = 123; say $a.reverse; 03:31
rakudo: my $a = 123; say $a.reverse;
p6eval rakudo cfbeb5: OUTPUT«123␤»
TimToady rakudo: my $a = 123; say $a.flip
p6eval rakudo cfbeb5: OUTPUT«321␤»
Visitor99 if there is no reverse for an int.. why such a method exists for ints? 03:32
pmichaud it's treated like a list of one element 03:33
in general, most non-list objects will act like a list of one element if you attempt to use a list method on them
TimToady such methods are more a part of the language than a method of any particular type
though any given type may override them, of course 03:34
snarkyboojum rakudo say Int.^methods(:local).join(' ');
rakudo: say Int.^methods(:local).join(' ');
p6eval rakudo cfbeb5: OUTPUT«log log10 sech truncate sin sqrt Int asin Bool cosh succ sign asinh acosech abs reals ceiling unpolar floor asech Rat round acosec acosh Num cosech cotan atan2 tan Real atan cos acos tanh perl Numeric Str acotanh ACCEPTS from-radians pred atanh Bridge exp cosec sinh to-radians
..WHIC…
03:35 gbacon left
TimToady it seems odd that Int would have all those Real methods as :local 03:35
pmichaud agreed 03:36
rakudo sez 03:37
augment class Int does Real {
so I'm guessing the Real methods are composed into the Int class
beyond that, I don't know the current state of things in rakudo -- I've been letting colomon++ et al. handle those while I'm focused elsewhere. 03:38
s/things/Real and numerics/
03:39 meppel joined
TimToady it's like you want Int is RealInt, and RealInt does Real, or something 03:41
but it's probably not important 03:42
03:43 meppl left
sorear is now on the most technically problematic point of dfa/Cursor, named character classes 03:45
03:45 snarkyboojum left, Visitor99 left
lue ponders sized ints and other number types, if applicable 03:48
rakudo: my int8 $a = 0xFF; say $a.WHAT 03:54
p6eval rakudo cfbeb5: OUTPUT«===SORRY!===␤Malformed my at line 11, near "int8 $a = "␤»
lue rakudo: my Int8 $a = 0xFF; say $a.WHAT 03:55
p6eval rakudo cfbeb5: OUTPUT«===SORRY!===␤In "my" declaration, typename Int8 must be predeclared (or marked as declarative with :: prefix) at line 11, near " $a = 0xFF"␤»
TimToady rakudo has no idea what you're talking about :)
lue it's interesting, every single number type with size is lowercase
TimToady yes, because those are the native types, which care 03:56
03:56 tedv left
lue is it a parrot-side problem or rakudo-side? [me checks ROADMAP] 03:56
pmichaud little of both.
it's also not high priority at the moment 03:57
(relative to other things that need doing)
lue Yeah, it's nice to have on ROADMAP, with footnote about Parrot devel needed.
03:58 snarkyboojum joined
sorear ROADMAP should be called TIMELINE & ROADMAP should document what each little piece of the compiler is /for/ 03:59
sorear has solved the named character classes issue
pmichaud sorear: then what would we do with "compiler overview.pod"? ;-) 04:00
arguably ROADMAP should be "development roadmap", yes.
lue
.oO(spaces are still a no-no in file names, IMO. It messes up my use of the command line. :) )
pmichaud just too lazy to type underscore at the moment. 04:01
lue or, it could be call NYI :)
"Hmm, what to do on rakudo today? Oh! I know, let's check NYI!" 04:02
pmichaud it's more like WWTFIWBI
("when will the feature I want be implemented?")
lue WINIR — What Is Not In Rakudo 04:03
rakudo: say ord("_") - ord(" ") 04:06
p6eval rakudo cfbeb5: OUTPUT«63␤»
sorear What are the valid forms of Unicode character property test in Perl 6? 04:07
pmichaud See S05.
sorear yes, I asked after reading that and being confused
pmichaud ah. 04:08
I'd have to re-read to refresh my memory... just a sec
TimToady S05:1848 04:09
pmichaud so, looks like things like isLu, islt, etc are the standard ones
then there are some shortcut ones, like <alpha> and <digit>
sorear Where is the whole list?
lue Is there anywhere I can find out what languages you can currently get come from?
TimToady the Unicode Consortium keeps changing the list :)
pmichaud I went by the tests in t/spec/S05 to determine a list :)
sorear My current wild goose chase currently has me at perluniprops(3), but it's not clear what the Perl 6 syntax for Age: 3.1 is
TimToady P5 appears to use \p{age=3.0} 04:12
so probably ends up with something smartmatchy, like <age(v3.0)> 04:13
this is lookin in lib/unicore/To/DAge.txt from the 5.12 dist 04:14
sorear perhaps in trying to implement all Unicode properties in the LTM engine at the same time I am getting slightly ahead of myself.
TimToady perhaps
but I'd certainly be tempted to hook in anything that P5 can do :)
even if you have to translate all the tables to P6
sorear I'll do that. Eventually. 04:15
TimToady someone should write a mktables that just spits out p6 instead of p5 :)
sorear I wonder what the minimum set I need for STD.pm6 is
<alpha>, definititely
TimToady we don't use very many directly, except those that have methods in Cursor 04:16
sorear [_[:alpha:]\pL] # what.
TimToady <[_]+alpha+isL> in p6ese
er, hmm 04:17
sorear \w \s \d we also use
TimToady I see the problem :)
sorear I'm mostly just suprised that [:alpha:]\pL isn't redundant.
TimToady that's because it's trying to get posix's def of alpha in there too, I think... 04:18
snarkyboojum just wrote a very very inefficient pure Perl6 MIME base64 encoding thing
TimToady or perhaps it was a microoptimization for TRE
snarkyboojum when is pack/unpack slated for implementation?
TimToady I don't remember why I had both of those in there, except I vaguely recall TRE flakiness 04:19
sorear What's the most efficient way to write $foo =~ /\pL/ in Perl5? 04:20
04:20 patspam left
lue What programming languages can I currently use code from? 04:21
TimToady I don't think you can have direct access to the data structures used by \pL, if that's what you're asking
snarkyboojum lue: are you asking about language interop on parrot? 04:22
TimToady but if you want to pull in the bitmap and lookup the ord yourself, you can
04:22 alester left
sorear How would I pull in the bitmap? 04:23
TimToady if you say do 'unicore/Name.pl' you get the output of Name.pl
sorear oh hey! this "unicore/" thing is shiny 04:24
sorear looks at the files in it
oh, *that*'s what mktables is
lue as in, importing Qt4 from C (for example)
sorear Why is mktables 14 ksloc? 04:25
snarkyboojum can anyone familiar with Q:PIR help with this breakage/misuse gist.github.com/438641? :)
lue: I don't understand what you mean by that, but that's probably just me 04:26
lue rakudo: use libc:from<C> 04:27
p6eval rakudo cfbeb5: OUTPUT«===SORRY!===␤"load_language" couldn't find a compiler module for the language 'C'␤»
lue like that. 04:30
TimToady mktables has grown considerably since I first wrote it...
but then, so has Unicode
snarkyboojum lue: ah :) now I can honestly say I don't know :)
lue It allows you to use libraries from anywhere. 04:31
snarkyboojum doesn't :from<perl5> work to some extent? :) 04:32
lue rakudo: use numpy:from<Python>
p6eval rakudo cfbeb5: OUTPUT«===SORRY!===␤"load_language" couldn't find a compiler module for the language 'Python'␤»
sorear TimToady: I just don't like the idea of porting 14,000 lines of code :(
TimToady don't, just run the resulting .pl files 04:33
snarkyboojum lue: I guess the ppl who know are too busy
sorear so apparently Perl parses unicore/lib/Gc/L.pl every run 04:34
snarkyboojum mumbles into his beard
TimToady I suspect it cheats if it sees only ascii 04:36
04:36 simcop2387 left
sorear *shakes fist* 04:49
utf8::SWASHGET has been rewritten in C and is no longer accessible from Perl-space except through the regex engine 04:50
04:50 simcop2387 joined
TimToady swashes are very utf-8-centric; I wouldn't steal them for p6 04:51
05:00 c1sung joined
sorear what do you mean? 05:01
oh I see
05:01 ash___ left
lue afk and goodnight 05:02
sorear user-defined unicode properties!?!? 05:09
05:13 kaare joined 05:14 kaare is now known as Guest50632 05:26 redicaps joined 05:35 Visitor99 joined 05:37 snarkyboojum left 05:39 orafu left 05:41 orafu joined 05:47 snarkyboojum joined
Visitor99 hi again... 05:49
is there a way i can write this line shorter? : when ($_ mod 2 == 0) { ... } 05:50
sorear when * !% 2
er
when !mod 2
* !mod 2
hejki when not $_ % 2 { .. } :) 05:51
diakopter std: when !%2 {} 05:52
p6eval std 31260: OUTPUT«ok 00:01 105m␤»
Visitor99 and this? when ($_ mod 2 == 3) { ... } when ($_ mod 2 == 5) { ... }
(the "mod 2 == 0" was not a very good example) 05:53
diakopter std: when %2==3 {}
p6eval std 31260: OUTPUT«ok 00:01 105m␤»
diakopter I guess
sorear when ($_ mod 2 == 3) can also be written when False 05:55
it's more efficient, too
diakopter explain 'when False'? 05:59
I mean, I'm giving you the benefit of the doubt that you're not being pedantic (read: distastefully unhelpful) about operator precedence 06:00
06:01 ggoebel left
pmichaud I'm pretty sure that %_ mod 2 can never == 3 :-) 06:01
so it's always false
or, if precedence is the other way, then % mod 0 is ... well.
diakopter er 06:02
that wasn't the question 06:03
pmichaud ah.
diakopter it was always $_
06:05 masak joined
masak oh hai, #perl6 06:05
phenny masak: 14 Jun 20:35Z <moritz_> ask masak if he thinks moritz.faui2k3.org/tmp/tryrakudo.svg looks like a sane structure for try.rakudo.org
diakopter grrrrrrrrr
pmichaud masak: o/
masak moritz_: yes, I think it looks sane.
ciphertext snarkyboojum: if you're still wondering, the problem with your Base64 program is that the namespace should be ['parrot'; 'MIME'; 'Base64'] instead of just ['parrot'; 'MIME'] 06:06
masak moritz_: would stage one be getting something like that to work, but entirely without stylesheets? then one could invite a weekly challengee to do the styiling. 06:07
snarkyboojum ciphertext: cheers - will try it
sorear diakopter: X mod 2 can never be 3, or 5 06:09
06:10 gfx joined
sorear has an efficient-ish algorithm for solving SAT/TAUT on Boolean expressions of Unicode properties 06:14
06:15 Su-Shee joined 06:17 uniejo joined, JimmyZ left 06:18 am0c joined
diakopter heh; I knew that... someday. 06:18
ago
snarkyboojum ciphertext: lovely :) thanks for picking that up. works a treat now 06:19
06:19 plobsing left 06:20 diakopter left
ciphertext snarkyboojum: it gave me a good excuse to dive into learning a little parrot, which i've been meaning to do 06:20
snarkyboojum ciphertext: :)
ciphertext++ even 06:21
Visitor99 hehe yes i knew but was in part to focus on the $ part... the code i was writing was about numbers divisible by 2 or 3 or 5 06:22
rakudo: for (4,5,15,20,35) {.say; when $_ mod any(2,3,5) {' div by 2, 3 or 5'.say} }; 06:24
p6eval rakudo cfbeb5: OUTPUT«4␤ div by 2, 3 or 5␤5␤15␤20␤35␤»
snarkyboojum ciphertext: github.com/snarkyboojum/Perl6-MIME-Base64 - there you go 06:25
Visitor99 rakudo: for (4,5,15,20,35) {.say; when $_ mod 2|3|5) {' div by 2, 3 or 5'.say} };
p6eval rakudo cfbeb5: OUTPUT«===SORRY!===␤Missing block at line 11, near ") {' div b"␤»
Visitor99 rakudo: for (4,5,15,20,35) {.say; when $_ mod 2|3|5 {' div by 2, 3 or 5'.say} };
p6eval rakudo cfbeb5: OUTPUT«4␤5␤ div by 2, 3 or 5␤15␤20␤35␤»
cono rakudo: for (4,5,15,20,35) { .say if $_ mod 2|3|5 } 06:28
p6eval rakudo cfbeb5: OUTPUT«4␤5␤15␤20␤35␤»
pmichaud \o/ list branch is now passing all of S02-builtin_data_types 06:29
cono yahoo
Visitor99 :D cool
sorear !!! 06:30
Visitor99 rakudo: for (4,5,15,20,35) { .say if $_ mod 2|3 } 06:31
p6eval rakudo cfbeb5: OUTPUT«4␤5␤15␤20␤35␤»
pmichaud (well, except for the tests we marked as todo/skip :-) 06:33
cono todo in *++ :D
Visitor99 rakudo: for (4,7,5,15,20,35) { if any($_ mod 2,$_ mod 3,$_ mod 5)==0 {$_.say} } 06:36
p6eval rakudo cfbeb5: OUTPUT«4␤5␤15␤20␤35␤»
masak someone has been to something called #yr2010, and seems pleased with Rakudo and Perl 6: twitter.com/mmcleric/status/16178658081
pmichaud perhaps "yapc russia"? 06:37
it was this past weekend
masak ah. yes.
this gentletwitterer, OTOH, was underwhelmed: twitter.com/alecthegeek/status/16199980776 06:38
06:39 justatheory left
pmichaud seems about right. 06:40
Visitor99 rakudo: for (4,7,5,15,20,35) { if $_ mod any(2,3,5)~~0 {$_.say} } 06:43
p6eval rakudo cfbeb5: OUTPUT«4␤5␤15␤20␤35␤»
masak snarkyboojum: (re pack/unpack) -- I've been wanting those two functions for over a year now. the good news is they're part of my GSoC schedule, so I'll eventually get to them :) we suspect they have more to do with Buf than with Str. 06:44
Visitor99 is there any way to know which of the numbers in the 'any' junction was chosen? i would like to print "$_ is divisible by (and then the corresponding divisor)" 06:45
06:45 redicaps left
pmichaud Visitor99: no 06:45
cono rakudo: {.say} for <1 2 3> 06:47
p6eval rakudo cfbeb5: ( no output )
cono rakudo: for <1 2 3> {.say}
p6eval rakudo cfbeb5: OUTPUT«1␤2␤3␤»
cono "for" lose it's reverse form? 06:48
pmichaud no, but you just told it to create a closure
it never executes the closure
rakudo: .say for <1 2 3>
p6eval rakudo cfbeb5: OUTPUT«1␤2␤3␤»
cono ah )
thanks 06:49
pmichaud with the curlies, it's not an immediate block, so it doesn't get executed.
moritz_ rakudo: for 4,7,5,15,20,35 -> $n { for 1, 2, 3 -> { if $n !% $_ {say "$n is divisible by $_" ; last } } }
p6eval rakudo cfbeb5: OUTPUT«Too many positional parameters passed; got 1 but expected 0␤ in main program body at line 11:/tmp/Kl44tU9BTS␤»
moritz_ rakudo: for 4,7,5,15,20,35 -> $n { for 1, 2, 3 { if $n !% $_ {say "$n is divisible by $_" ; last } } }
p6eval rakudo cfbeb5: OUTPUT«4 is divisible by 1␤7 is divisible by 1␤5 is divisible by 1␤15 is divisible by 1␤20 is divisible by 1␤35 is divisible by 1␤»
moritz_ rakudo: for 4,7,5,15,20,35 -> $n { for 2, 3, 5 { if $n !% $_ {say "$n is divisible by $_" ; last } } } 06:50
p6eval rakudo cfbeb5: OUTPUT«4 is divisible by 2␤5 is divisible by 5␤15 is divisible by 3␤20 is divisible by 2␤35 is divisible by 5␤»
cono !% it means div produce 0? :)
mod* 06:51
masak pmichaud: how come { .say } is an immediate block on its own, but not with a statement_modifier:<for> on it?
moritz_ a !% b is the same as !(a % b)
cono wow
pmichaud iiuc, immediate blocks don't get statement modifiers. It could be a rakudo bug, though.
Visitor99 rakudo: for (4,7,5,15,20,35) -> $n { given (2|3|5) { if $n mod $_ ==0 {say "$n is divisible by $_"} } } 06:52
p6eval rakudo cfbeb5: OUTPUT«4 is divisible by any(2, 3, 5)␤5 is divisible by any(2, 3, 5)␤15 is divisible by any(2, 3, 5)␤20 is divisible by any(2, 3, 5)␤35 is divisible by any(2, 3, 5)␤»
06:52 parduncia joined
masak pmichaud: no, I recall a test addressing this with a p6l link attached to it. 06:52
so it's probably right.
and I should go read the p6l thread in question. :)
huh. I'm wrong. :/ 06:54
06:54 Trashlord joined
masak we have Rakudo todo tests saying the closure should be executed. 06:54
pmichaud yeah, istr that also
cono rakudo: ([/] 1,2,3).perl.say
p6eval rakudo cfbeb5: OUTPUT«1/6␤»
masak makes a svn blame to see if he was at least right once :)
"r29799 | pmichaud | 2010-02-22 16:44:52 +0100 (Mon, 22 Feb 2010) | 6 lines [t/spec]: Update testing in for.t of statement_modifier:for on blocks. The original tests were based on a p6l comment posted by Larry in September 2006, but later changes to S04 (May 2007) indicate that statement modifiers should execute explicit closures. This commit updates the tests to be consistent with the synopses." 06:56
pmichaud ah.
masak so yes, rakudo-todo.
pmichaud late February through mid-May is pretty much a blur in my head :0|
masak TimToady++ pmichaud++
pmichaud: it would be. 06:57
pmichaud: it's very nice to have you back, and hope that things are well/improving on the home front. 06:58
cono Maybe for perl Configure.pl --gen-parrot need to make svnclobber in parrot? Coz I always file to make parrot :( I remove parrot folder every time when parrot depending revision changed
pmichaud masak: the do appear to be.
masak \o/
pmichaud *they
cono s/file/fail/
pmichaud we need to change the ordering of the "make realclean" step in the parrot dir. 06:59
need to do "make realclean" before doing the "svn co"
cono realclean are not help time to time
moritz_ cono: svnclobber is very impolite to those who have patch files in their parrot/ dir
cono dunno why
ic
pmichaud realclean doesn't work across svn commits in the parrot repo, because of changes to the makefile structures 07:00
sorear gives in to the urge to split _AUTOLEXgenDFA into several smaller functions despite the fact that Perl5 doesn't have a functional inliner
pmichaud so, if we realclean before doing the svn update, things should be better.
moritz_ on lists branch: 07:01
> say sprintf "%04f-%02d-%02d", 2010, 3, 5
printf also doesn't produce any output 07:02
pmichaud just fixed sprintf locally
moritz_ I seem to have a knack for just investigating the stuff where you're one step ahead of me :-)
pmichaud fixed in 753e45 07:03
(at least for me it was fixed)
well, I'm working from top-down in the spectests, so if you're doing the same, you're likely to be seeing the same things I am :)
moritz_ does the Date.t file pass with it?
pmichaud haven't tried.
moritz_ that's where I noticed the failure
(not top down!)
pmichaud anyway, let me know if it's still failing in 753e45 :) 07:04
moritz_ I suspected as much, and randomly picked something from the middle
masak that 'say sprintf' pattern is cute. I use it sometimes.
someone once suggested we have a 'sayf' shortcut. didn't go over well, I think. :P 07:05
pmichaud oh, part of the sprintf fix is uncommitted
one second. 07:06
Visitor99 rakudo: for (4,7,5,15,20,35) -> $n { given none(2,3,5) { if $n mod $_ ==0 {say "$n is not divisible by $_"} } }
p6eval rakudo cfbeb5: OUTPUT«7 is not divisible by none(2, 3, 5)␤»
pmichaud pushed 68354d6
07:07 Visitor99 left
masak Visitor99: cool, but confusing. maybe use all(2,3,5) and != 0 instead. 07:08
more from the underwhelmed gentletwitterer: "@dryfter demonstrated Perl6, plus some Go and Scala. We had a disappointing time with Perl6. Some Notes on Perl 6 came from it." perl.net.au/wiki/Melbourne_Perl_Mon...story_2010 07:10
aww, they didn't know about fakedbi! :( 07:11
masak tells them
07:11 am0c left
pmichaud for anyone looking to help with the 'list' branch, the S03 tests are probably good candidates now 07:16
masak today is Parrot release day, no? 07:17
pmichaud yes, unfortunately.
moritz_ yes
range.rakudo seems to loop
masak why 'unfortunately'?
pmichaud I'm not likely to have closure fixes in place before the Parrot release.
(we can work around it, but I'd rather have them "in Parrot" ) 07:18
masak pmichaud: I agree with your estimate that list semantics is more important than closures.
moritz_ same for regex fixes
cono What objects have cmp && eqv methods?
masak pmichaud: at least the closures stuff has a very simple workaround.
pmichaud and I think we will have closure done soonish
masak cono: all of them, in some sense. 07:19
pmichaud just need to get lists straightened around -- there's a lot to the refactor (as can be seen by the number of commits needed to fix it all up :-)
moritz_ spectest looks much better now 07:20
pmichaud it's also nice that I've been able to delete large sections of redundant code :)
cono noting found: grep "method cmp" src/core/*.pm
pmichaud I think it's an operator. 07:21
sub infix:<cmp>
cono TODO said: * tests for cmp() and eqv() methods of objects
pmichaud yeah, spectest does look a lot better 07:23
(now running a full spectest for the first time in... well, days)
yes, does appear to be an infloop in range.t 07:24
looking.
masak rakudo: say 1..4 cmp 3..5 07:25
p6eval rakudo cfbeb5: OUTPUT«-1012345␤»
masak rakudo: say (1..4) cmp (3..5)
p6eval rakudo cfbeb5: OUTPUT«-1␤»
masak I wonder how it figures that.
pmichaud I think that if cmp isn't defined for a type, it defaults to string comparison.
rakudo: say (3..5) cmp (11..14) 07:26
p6eval rakudo cfbeb5: OUTPUT«1␤»
07:27 JimmyZ joined
pmichaud ah, yes, the RangeIter doesn't know about string comparisons. That needs to be ported from master into the new RangeIter code. 07:28
I'll skip the test for now, but perhaps someone can work on it a bit. 07:29
pugssvn r31261 | pmichaud++ | [t/spec]: Fudge and update some tests for the rakudo 'list' branch. 07:31
pmichaud btw, for the #phasers meeting today, I've started putting my review comments and questions into docs/review-notes.txt in the 'list' branch 07:32
so if people want to prepare some answers or think about them a bit, that's a good place to start :)
cono can I create S03-operators/cmp.t ? 07:33
pmichaud cono: yes, please do! 07:34
sorear anyone with a pugsbit can add tests
pmichaud there might be some cmp tests already -- so might be worth checking for that. 07:35
if there are, they can be refactored.
sorear pugsbits are handed out freeely to anyone who can pass a turing test, which we secretly administer to all channel joiners
cono :) 07:36
pugssvn r31262 | sorear++ | [dfa/Cursor] Refactor _AUTOLEXgenDFA and design the NFA representation for named character classes 07:40
07:41 whiteknight joined 07:43 JimmyZ left
pmichaud ah, the S32 tests are likely candidates for low hanging fruit also 07:44
07:46 meppel left
whiteknight pmichaud? you're up early today 07:49
...or up late
masak rakudo: role A { method foo { say "A" } }; role B { method foo { say "B" } }; class C does A does B {}
p6eval rakudo cfbeb5: OUTPUT«===SORRY!===␤Method 'foo' collides and a resolution must be provided by the class␤»
masak would it be possible for the error message to specify which class?
also perhaps line number and file?
arnsholt When's takadonet usually around? 07:50
masak arnsholt: from 15-ish and onwards. 07:51
he's on New World time :)
rakudo: our &s = &say; role A { method foo { s "A" } }; role B { method foo { s "B" } }; class C does A does B { method foo { self.A::foo } }; C.new.foo
p6eval rakudo cfbeb5: OUTPUT«A␤»
masak \o/
pmichaud whiteknight: up late
about to head to sleep
been working on the rakudo list refactor -- want to try to get it in before this month's release 07:52
(may hold the release a day or two -- need to discuss that at #phasers today)
arnsholt masak: Excellent. I'll keep my eyes open around then, thanks 07:53
08:00 lestrrat is now known as lest_away 08:01 sdo_ joined 08:02 Irwin left 08:03 jferrero joined
pmichaud afk, sleep 08:03
08:05 whiteknight left
cono rakudo: ("привет").perl.say 08:07
p6eval rakudo cfbeb5: OUTPUT«"привет"␤»
08:07 rob joined, gfx left
masak rakudo: ("привет").uc.say 08:08
p6eval rakudo cfbeb5: OUTPUT«ПРИВЕТ␤»
cono masak: in my console something wrong with this :( 08:09
masak rakudo: "привет".subst('и', 'е').subst('т', 'д').uc.say
p6eval rakudo cfbeb5: OUTPUT«ПРЕВЕД␤» 08:10
masak cono: then you probably don't have uci installed.
cono: it tells you during Parrot config whether uci is found or not.
moritz_ icu, actually 08:11
cono hah
when I try perl6 -e, it say to me: no ICU lib loaded
but when I'm in interactive mode, it prints like: perl -le 'my $c ="привет"; utf8::encode($c); print $c' 08:12
sorear <3 unicode 08:13
cono :D
08:14 varun joined
moritz_ I do't think the REPL ever worked well with Unicode 08:15
cono repl? 08:16
moritz_ the interactive Read-Evaluate-Print-Loop 08:17
cono ic
pugssvn r31263 | sorear++ | [dfa/Cursor] Switch internal NFA processing to the class-aware form 08:21
sorear I predict dfa/Cursor will be mergable by this time tomorrow
moritz_ likes 08:25
08:25 synth left 08:28 szabgabx left
cono ICU installed, but in repl still have bug :( 08:30
perl6 -e works fine
08:30 dakkar joined
moritz_ 10:15 <@moritz_> I do't think the REPL ever worked well with Unicode 08:33
cono ah :) 08:34
my bEd :)
08:34 wallberg joined
sorear What Unicode properties are equivalent to \w? 08:34
moritz_ I think it's Letter, Digit or _ 08:36
masak ah! the 'w' in \w means 'word character'. that's how _ gets to be in there. 08:39
despite what has sometimes been said on the channel, _ is *not* alphanumeric :) 08:40
moritz_ rakudo: say '_' ~~ /<.alpha>/
p6eval rakudo cfbeb5: OUTPUT«_␤»
moritz_ in Perl 6 it is. 08:41
masak oh. and here I thought I had come upon a grain of sense. 08:42
08:47 dakkar left
mberends that one has bitten me a few times as well :-/ 08:55
08:57 rgrau` left, rgrau joined 09:00 dakkar joined
sorear How does case insensitive matching work in perl 6? 09:03
e.g. "ß" ~~ m:i/SS/
masak sorear: that's the syntax, so I assume you're asking about the semantics.
oh wait, that example is tricky :) 09:04
I would be surprised if Perl 6 is spec'd to handle that.
sorear yes. semantics, and dark corners thereof
yes, S05 says exactly nothing on the subject
moritz_ rakudo: say "ß" ~~ /:i SS/
p6eval rakudo cfbeb5: OUTPUT«␤»
sorear other than "ignores case distinctions"
moritz_ rakudo: say uc "ß"
p6eval rakudo cfbeb5: OUTPUT«SS␤» 09:05
masak o.O
German is strange.
moritz_ masak: want to submit a bug report about inconsistency?
it sure is.
masak submits rakudobug
moritz_ if Perl 6 isn't able to handle Unicode properly, which language is?
masak moritz_: you sound like someone with True Believer Syndrome :P
Trashlord how's it going 09:06
moritz_ I'm a True Believer
masak moritz_: :D
cono perl5 :)
perl5 awesome in unicode
moritz_ coughs
masak except when it isn't.
moritz_ there's much Unicode-stuff that works very well in perl 5 09:08
but there are myriads of dark corners too
arnsholt There are a couple of dark corners in Unicode as well
moritz_ though it has improved in the last two major releases
masak "major" being 5.10 and 5.12 ? :) 09:09
moritz_ yes
masak also known as "minor"? :)
09:09 lest_away is now known as lestrrat
moritz_ masak: they are major releases of Perl 5 09:09
masak: and minor releases of Perl
:-)
masak any sufficiently advanced rationalization is indistinguishable from sophistry... :P 09:10
moritz_ chuckles
pugssvn r31264 | cono++ | * Add testing of cmp operator 09:17
09:19 am0c joined
masak I'm thking of adding auth/ver support to pls core. 09:19
cono btw, is that a bug? (v5) perl -e'\sort{eval$;}%!' && echo "exit 0" 09:20
masak I think if we're to introduce auth/ver for realz, we have to change the structure of projects.list. 09:21
mberends go for it, sooner rather than later 09:22
masak let's say there are two modules A in the ecosystem. we ignore the distinction between project and module for now.
there's A:auth<foo> and A:auth<bar>. which one do we pick? 09:23
moritz_ the one the user has a preference for
masak mberends: did you ever introduce auth/ver support into proto? I haven't checked.
moritz_ if such a thing doesn't exist, BOOM
mberends yes, that was a simplification we decided on in Copenhagen. Where it's truly ambiguous, flag an error.
masak ok.
mberends masak: no, proto didn't understand auth, but did try to upgrade vers 09:24
moritz_ where "preference" can be "explicit config option" or maybe "used this module from one authority before"
or... just install both :-)
(j/k)
mberends that's making lots of extra work for the implementer to satisfy a rare corner case. keep it simple, rather. 09:26
masak as to "explicit config option", the syntax in deps.proto will surely withstand an extention to allow for auths.
moritz_ probably preferable
mberends one auth at a time would be great anyway
masak that said, probably better to call it deps.pls :)
moritz_ deps.installthemforme.pls 09:27
masak is sorely tempted... 09:32
so... (social failure mode time again) ...if I have an arch-enemy whose life I want to complicate, all I need to do is to create projects with identical names to that person's projects, and none of his projects will install without human intervention? 09:37
even better, all the upstram projects which used those projects will also fail to install until someone clearly specifies the auth in the deps.pls file. 09:39
very much havoc with relatively little effort.
sorear imho auth should be required.
masak on the project level, maybe it should. 09:40
sorear with an error message that tells you the correct auth, if it's unambiguous among installed modules
masak that's another error message that'd hit innocents...
sorear on a file level, we don't need auth; Perl 6 should have a concept of packaging, and all modules in an installed package should know what auth/ver the package depends on 09:41
masak interesting idea to prefer an installed module if it's unique. 09:42
buu perlbot: placeholders is 09:44
perlbot: placeholders is You want placeholders: www.geekuprising.com/create_better_...aceholders
er
sorear if you're familiar with .NET's assembly mechanism, that's what I'm talking about; somebody at Microsoft stole my idea
buubot, seen masak 09:45
buubot sorear: I last saw masak saying "my @a = (1, 2, 3); my $i = 4; for my $elem (@a) { print $elem, "\n"; push @a, ++$i if $elem % 2 }" at Fri Jun 11 13:13:25 2010 Z.
snarkyboojum masak: regarding pack/unpack - good stuff :)
sorear buu: What's wrong with seen?
buu sorear: It only records when people actually talk to the bot.
Due to a somewhat complicated design flaw, plugins like seen only get a chance to activate if people talk directly to the bot.
Hrm.
sorear aha. 09:46
masak buu: I'm talking to you now! 09:47
oops.
buubot: I'm talking to you now!
buubot masak: Couldn't match input.
buu Haha
masak keep mixing the two of you up :)
buutbot: Couldn't match input? 09:48
buubot: Couldn't match input?
buubot masak: Couldn't match input.
masak though so
buubot: Why did the chicken cross the road?
buubot masak: Couldn't match input.
masak lunch & 09:49
09:51 azert0x joined
cono rakudo: my @q = <1 2 3>; my @z = 1, 2, 3; (@q cmp @z).say 10:16
p6eval rakudo cfbeb5: OUTPUT«0␤»
cono is that a bug?
arnsholt Don't think so 10:17
cono rakudo: my @q = <1 2 3>; @q.map({.WHAT}).perl.say
p6eval rakudo cfbeb5: OUTPUT«(Str, Str, Str)␤»
cono rakudo: my @q = 1, 2, 3; @q.map({.WHAT}).perl.say
p6eval rakudo cfbeb5: OUTPUT«(Int, Int, Int)␤»
arnsholt rakudo: <1 2 3>.perl.say; (1, 2,3).perl.say
p6eval rakudo cfbeb5: OUTPUT«("1", "2", "3")␤(1, 2, 3)␤»
cono different types
arnsholt Yup
Also, I'm not sure how cmp is specced to work on arrays 10:18
cono rakudo: my @q = <01 2 3>; my @z = 01, 2, 3; (@q cmp @z).say
p6eval rakudo cfbeb5: OUTPUT«-1␤»
pugssvn r31265 | sorear++ | [dfa/Cursor] Implement translation from NFA to DFA in the presence of named character classes 10:19
r31266 | sorear++ | [dfa/Cursor] Extend the DFA execution engine to support DFAs with Unicode property decision trees 10:30
sorear OK, I think those two commits cover the last of the major technical issues
sorear sleep
10:39 jhuni left
colomon rakudo: my @q = <1 2 3>; my @z = 1, 2, 3, 4; (@q cmp @z).say 10:42
p6eval rakudo cfbeb5: OUTPUT«-1␤»
colomon oh, if there is no special handling for a type (like there is for numbers), cmp converts both its arguments to Str and compares those. I believe that's what's happening here for arrays. 10:44
10:45 Schwern left
colomon TimToady: Rakudo only defines basic arithmetic and a few conversion functions for Int, everything else you saw on that list was picked up from Real or maybe even Numeric. 10:46
10:49 mjk joined 10:50 tadzik joined 10:59 jhuni joined 11:02 lestrrat is now known as lest_away 11:04 gawker2 joined 11:08 paroxyzm joined, paroxyzm left, paroxyzm_ joined 11:20 azert0x left 11:22 azert0x joined
colomon pmichaud: gist.github.com/438996 -- Seems like Ranges in [ ] don't do the right thing yet. But what the heck is the series doing there at the end? 11:32
pmichaud: (trying to figure out the best way to implement .batch again, to get zip.t and presumably a number of other test files to run again) 11:33
ciphertext colomon: is that on the list branch? 11:35
colomon ciphertext: yes
ciphertext ok. fyi, the weird series behavior is the same on master 11:36
rakudo: (1...10)[0...4]
p6eval rakudo cfbeb5: ( no output )
ciphertext rakudo: say (1...10)[0...4]
p6eval rakudo cfbeb5: OUTPUT«01234␤»
colomon rakudo: say (1...10)[0,1,2,3,4] 11:37
p6eval rakudo cfbeb5: OUTPUT«12345␤»
colomon rakudo: say (1...10)[2...4] 11:38
p6eval rakudo cfbeb5: OUTPUT«Null PMC access in getprop()␤ in 'say' at line 1␤ in 'say' at line 4724:CORE.setting␤ in main program body at line 11:/tmp/2itAeUA6kN␤»
masak submits rakudobug
colomon ooooo, it's a closure bug, I'll bet.
rakudo: my @a = 1...10; say @a[0...4] 11:39
p6eval rakudo cfbeb5: OUTPUT«12345␤»
colomon closure bug, nothing to see here.
masak the closure bug usually doesn't result in a Null PMC access.
rakudo: my @a = 1...10; say @a[2...4]
p6eval rakudo cfbeb5: OUTPUT«345␤»
masak but yes, seems like it's parallel gather blocks interfering that's the underlying cause. 11:43
colomon I will be unspeakably happy when that bug is eliminated.
11:43 proller left 11:49 arnsholt left
masak std: module; 11:57
p6eval std 31266: OUTPUT«===SORRY!===␤Undeclared routine:␤ 'module' used at line 1␤Check failed␤FAILED 00:01 107m␤»
masak std: class;
p6eval std 31266: OUTPUT«===SORRY!===␤Undeclared routine:␤ 'class' used at line 1␤Check failed␤FAILED 00:01 104m␤»
masak S11 (strangely) claims those should work.
rakudo: say (v6..*).perl 11:58
p6eval rakudo cfbeb5: OUTPUT«Could not find sub &v6␤ in main program body at line 11:/tmp/Mqj5WHdqxU␤»
takadonet morning all 12:00
masak takadonet: \o
takadonet: arnsholt was looking for you earlier.
takadonet masak: he was? do you know what for?
masak no idea. 12:01
takadonet well I'm here :)
is anyone from perl 6 going to feed the troll today? use.perl.org/~ank/journal/40396 12:02
masak I've read it. I don't have anything to add.
I liked moritz_++'s comment: "I'm a True Believer" :)
12:02 ggoebel joined
masak (it wasn't in direct connection to that post, though.) 12:03
12:03 Mowah joined
baest I would like to see some links concerning this: "they simply cannot accept the fact that the whole concept is broken, even after it's been shown to them over and over." 12:03
masak ank certainly has claimed it over and over.
moritz_ baest: he'd probably link to his previous blog posts 12:04
baest :)
moritz_ baest: which aren't more worth reading than this
masak it was something to do with operators being akin to prepositions or something.
baest moritz_: no, I've read them (or at least some of them)
oh well, you win some and some is lost 12:05
are
mathw We don't feed trolls, I thought the policy was hugging? 12:06
12:07 hugme joined
masak hugme! \o/ 12:07
Su-Shee takadonet: ah, that troll. :)
12:07 skids joined
Su-Shee my theory is he's hoping that chromatic does a why and just vanishes. ;) 12:07
masak fat chance :P
Su-Shee but his children will starve because of perl 6! 12:08
takadonet Su-Shee: been a while since I was called a troll
cool 12:09
snarkyboojum I found his website www.ank.com.ar/
Su-Shee takadonet: I didn't call _you_ one but the author of the posting :)
takadonet ahh
12:10 foodoo joined
Su-Shee www.ank.com.ar/jade7/ this is funny though :) 12:10
masak except the parts that are more sad than funny. 12:11
Su-Shee ah well he isn't going to bring down perl 6 ;)
mathw Of course not 12:12
Su-Shee also according to hugme policy it's wise to embrace satire. ;)
mberends fwiw, I think ank writes in a deliberately offensive way, but there is also quite a lot of truth between the vitriolic parts. We should be careful not to overdose on our Kool-Aid.
masak aye. 12:13
I've also wondered about TheDamian's extreme lack of commenting. :)
moritz_ why should he comment?
snarkyboojum hehe.. too busy programming Perl6 no doubt :)
masak moritz_: hm, s/commenting/tests/, then. 12:14
mathw Well yes, Damian's code is pretty much incomprehensible
But that's not a point against the entire Perl 6 effort
masak of course not.
mathw You can write incredibly clever yet completely bizarre code in any language
masak and some people never proceed from that insight about Perl: "look, you can write *this*! danger, danger!" 12:15
mathw which is just unrealism
12:15 pjcj joined
Su-Shee ah well let's satisfy the people actually waiting in line for certainf eature to do their projects and let's see what happens then. also, on #perl the bitching about perl 6 subsided recently. 12:16
mathw I've seen Python code that was as hard to figure out as the worst Perl outside of golfing
moritz_ it's been a year since my last "how to contribute to Perl 6" post on perlmonks
tadzik The first example from Dive Into Python looks to me more cryptic than any Perl I ever wrote
mathw There was one particularly fun project which had four different classes all called Connection in different modules, passed them between modules freely and always called the parameter 'connection', yet you never knew which *kind* of Connection it was 12:17
moritz_ I'd like to keep up the tradition, and write another such node soonish
Su-Shee who cares what python does? it's our task to develop a good, clean, pretty perl 6 style and promote that.
mathw yup
moritz_ any suggestions of what needs to go into it?
mathw and for my part, I think it involves fairly liberal use of type specifiers
huf whoa, this ank guy is crazy :D 12:18
nice.
cognominal are Perl 6 critics are an epistemologic proof of existence? Probably not of the physical critic existence, nor of Perl 6 existence :)
mathw But that might be my C++ and Haskell bias
12:18 bluescreen joined
masak cognominal: no, it's more like Ghandi-con 2. :) "then they ridicule you". which counts as progress, I guess. 12:19
12:19 bluescreen is now known as Guest16581
Su-Shee I've read an interesting blogposting today why ruby folks are rather willing to adopt clojure and what this got to do with the writing style of the languages. 12:19
masak or maybe it's stage 3, "then they attack you".
Su-Shee "ghandi-con" *haha* :)
masak it's a Slashdot meme. :) 12:20
Su-Shee never really read it. ;)
masak it's occasionally informative, but mostly a source of extreme geek humour. 12:21
cognominal Masak. Well, Gandhi cited by esr always puts me ill at ease. There is like a contradictory logic in that utterance.
masak cognominal: :)
cognominal You don't advocate guns and cite Gandhi
masak evidence suggests otherwise :P
cognominal The world is illogic, absurd. So no logic whatsowever is a proof of existence. But may be we don't exist. 12:23
masak maybe not. but I've never seen the hypothesis that we don't lead to any important discoveries or insights. 12:25
...medical or engineering advances...
in fact, it seems to imply nothing at all, except an odd view of the universe. 12:27
12:28 am0c left 12:30 parduncia left 12:31 am0c joined
masak the version comparison semantics outlined in S02... is there a way to compare two strings programmatically according to the same rules? 12:33
because I think there should be. it would help pls to have access to such a way.
12:34 proller joined, parduncia joined, parduncia left
moritz_ who's doing our next release? 12:34
masak 12:35
moritz_ ++masak 12:36
colomon more important question is, when is our next release? 12:37
masak Thursday, I think.
moritz_ pmichaud wanted to discuss this today in #phaers, no?
*#phasers
colomon masak: I can't believe we're going to pull in lists and closures by then.\ 12:38
masak closures allegedly needed Parrot changes too.
so we might have to slip that one.
colomon it would be a hackathon of epic proportions.
masak s/slip/let slip/ 12:39
colomon I would vote against releasing R* with that bug.
moritz_ there are a lot of things I'd hate to see in R*
dalek kudo: 080700b | moritz++ | build/PARROT_REVISION:
bump PARROT_REVISION to 2.5.0 release
moritz_ like, lack of support for backslash escapes in regexes 12:40
mathw I'd hate to see the backtracking through captures bug
12:40 masonkramer joined
moritz_ there you go, we all have our "favorite" bugs 12:40
masak seems we do indeed have something to talk about during the #phasers meeting tonight.
mathw Well it's my favourite because I need it to fix Form :)
I'd like Form to be usable with * 12:41
but I don't see a workaround
takadonet come on lazy list! 12:43
12:44 ggoebel left
masak mathw: don't rely on backtracking :) 12:44
colomon gather / take is used ~30 times in core. With the closure bug and lazy evaluation, every one of those times is a landmine waiting to explode. And that's just one particular example of the closure bug.
mathw masak: I am not re-writing Form's grammar in non-backtracking mode 12:45
moritz_ does backtracking into non-capturing subrules with associated action methods work? 12:46
mathw I don't know
masak guesses "no" 12:49
pmichaud good morning 12:52
12:52 gawker2 left
masak morning, pm. 12:52
mathw oh hai pmichaud 12:53
moritz_: if my brain was working I'd bake up a small test case to find out
and then try to figure out how to make Form's grammar work without capturing, if it worked 12:54
colomon pmichaud: o/
moritz_ rakudo: grammar A { method TOP { ^<.a>* a $ }; token a { a } }; say A.parse('aa') 12:55
p6eval rakudo cfbeb5: OUTPUT«===SORRY!===␤Unable to parse blockoid, couldn't find final '}' at line 11␤»
moritz_ rakudo: grammar A { token TOP { ^<.a>* a $ }; token a { a } }; say A.parse('aa')
12:55 araujo left
p6eval rakudo cfbeb5: OUTPUT«␤» 12:55
moritz_ doesn't even work without action methods 12:56
mathw :(
moritz_ rakudo: grammar A { regex TOP { ^<.a>* a $ }; token a { a } }; say A.parse('aa')
p6eval rakudo cfbeb5: OUTPUT«aa␤»
moritz_ well
I took a token
which doesn't backtrac :/
mathw hah
pmichaud backtracking into rules and subpatterns is nyi
shouldn't be too hard to implement, just needs available time. :-| 12:57
12:57 hercynium left
mathw oh if only I could send you tuits 12:57
12:57 mikehh joined, jferrero left 12:58 hafos joined
pmichaud colomon: List doesn't (or shouldn't) need a .batch 12:59
colomon pmichaud: I understand that.
hafos I'm experiencing a problem where a Perl module won't install, and I get the following output: pastebin.com/vZary2vE - What does INSTALLDIRS=site mean? Doesn't tell me a whole lot and I'm not sure what it is
masak hafos: Perl 5? 13:00
moritz_ hafos: that seems to be a Perl 5 module
13:00 araujo joined
moritz_ hafos: we're discussing Perl 6 here 13:00
hafos You're right my bad
takadonet hafos: np
masak no worries.
colomon pmichaud: but there are dozens of tests written expecting the results of gather / take to have a .batch method.
pmichaud then those tests are no longer correct :-)
moritz_ hafos: but "OS unsupported" should be pretty clear anyway :-)
colomon i'd prefer to make progress on getting real bugs fixed than rewriting tests all day. 13:01
pmichaud so, todo or skip them
or even remove them
hafos moritz_: it's a normal debian installation, I thought it just meant the developers of perl don't offer official support or something ?
colomon yes, but then you don't get the benefit of seeing whether or not the tests actually work!
moritz_ hafos: you can't install Win32.pm on Debian. It needs the Windows API 13:02
hafos oh okay, thanks
pmichaud anyway, .batch (if it exists) really only makes sense on iterators, not on lists. since gather returns a list, we don't need to be testing .batch on it
what tests are these, anyway? 13:03
colomon every single test for infinite lists, for instance.
pmichaud file?
colomon zip.t is the one I am looking at.
but most of the series test files will also have it.
moritz_ S03-metaops/reduce.t
S03-metaops/zip.t 13:04
S32-list/pick.t
S05-modifier/continue.t
S03-opeartors/series*.t
pmichaud you could use .munch there instead
colomon until this weekend, batch was the accepted way to test infinite lists.
13:04 Schwern joined
pmichaud instead of .batch(5) you can use .munch(5) or you can use .[^5] 13:05
moritz_ what is .munch?
pmichaud it's basically the result of doing .shift $n times
(which is what .batch was doing)
moritz_ it's not spec either
.[^5] then 13:06
pmichaud well, it's not spec yet.
moritz_ before we run into the same trouble again
pmichaud almost everything with iterators is going to be changing
but yes, .[^5] is very safe.
colomon except that it doesn't work.
pmichaud it will return the first five elements of the list
well, then we fix it. :)
colomon that's why I implemented batch with [(^5).list]
pmichaud so, should Range ~~ Positional ? 13:07
I'm guessing yes.
moritz_ rakudo: say (1..5)[3] 13:08
p6eval rakudo cfbeb5: OUTPUT«Method 'postcircumfix:<[ ]>' not found for invocant of class 'Range'␤ in main program body at line 1␤»
colomon I would think no, but I guess it can.
moritz_ perl6: say say (1..5)[3
perl6: say say (1..5)[3]
p6eval pugs: OUTPUT«*** ␤ Unexpected end of input␤ expecting operator or "]"␤ at /tmp/QjquYMD86x line 2, column 1␤»
..elf 31266: OUTPUT«Parse error in: /tmp/wpUA6Foa65␤panic at line 1 column 0 (pos 0): Can't understand next input--giving up␤WHERE: say say (1..5)[3␤WHERE:/\<-- HERE␤ STD_red/prelude.rb:99:in `panic'␤ STD_red/std.rb:76:in `scan_unitstopper'␤ STD_red/std.rb:224:in `comp_unit'␤
..STD_red/std.rb:210:in…
..rakudo cfbeb5: OUTPUT«===SORRY!===␤Unable to parse postcircumfix:sym<[ ]>, couldn't find final ']' at line 11␤»
rakudo cfbeb5: OUTPUT«Method 'postcircumfix:<[ ]>' not found for invocant of class 'Range'␤ in main program body at line 1␤»
..pugs: OUTPUT«4␤1␤»
pmichaud rakudo: sub abc(@a) { say @a.elems; }; say abc(1..3); # do we expect this to work?
p6eval ..elf 31266: OUTPUT«Can't call method "postcircumfix__91_32_93" without a package or object reference at (eval 125) line 3.␤ at ./elf_h line 5881␤»
rakudo cfbeb5: OUTPUT«Nominal type check failed for parameter '@a'; expected Positional but got Range instead␤ in 'abc' at line 11:/tmp/aeRXKukCdd␤ in main program body at line 11:/tmp/aeRXKukCdd␤»
pmichaud if we expect to be able to pass 1..3 to a function taking an @-parameter, then 1..3 has to be Positional 13:09
I'm thinking yes.
moritz_ +1
13:09 varun left
pmichaud so, adding "does Positional" to Range should get .[^5] to work. 13:09
colomon I agree we expect to pass 1..3 to a function taking an @-parameter, but it seems to me this is another example of @-parameters being Positional being Not Quite Right 13:10
rakudo: say (1...10)[^5]
p6eval rakudo cfbeb5: OUTPUT«12345␤»
13:11 christine left
moritz_ colomon: that's also a good explanation 13:11
pmichaud I wonder if ^5 should really be 0...4 instead of 0..4 :-)
moritz_ instead of 0..^5
pmichaud oh, nm, that answers the question.
it needs to be 0..^5
colomon I dunno, that would significantly change the meaning. 13:12
pmichaud that's where the ^ comes from
colomon Not sure anyone uses it for anything other than the equivalent of 0...4
pmichaud anyway, > my @a = <a b c d e f g>; say @a[^3]
abc
moritz_ \o/
colomon moritz_: you want to change all the batches, or would you like me to? 13:13
pmichaud pushed.
colomon should be easy enough to write a Perl 5 script to do it automatically.
moritz_ colomon: I can do it, I don't feel productive anyway
it might break master though.
pmichaud rakudo: my @a = <a b c d e f g h>; say @a[^3] 13:14
p6eval rakudo cfbeb5: OUTPUT«abc␤»
pmichaud looks like master works already.
colomon moritz_: that's why I just checked it a second ago
13:14 araujo left
pmichaud unless it doesn't work from gather/take for some reason (flattening, possibly) 13:15
colomon moritz_: you should definitely test it on master first, a lot of those tests probably don't work for other reasons in the list branch yet.
pmichaud: it does, see my test about one screen back
rakudo: say (1...10)[^5]
p6eval rakudo cfbeb5: OUTPUT«12345␤»
colomon oh, wait
rakudo: say (1...*)[^5]
never mind, that's not a workable solution.
p6eval rakudo cfbeb5: ( no output )
pmichaud > say (1...*)[^5] 13:16
12345
13:16 araujo joined
pmichaud (from the branch) 13:16
colomon I dunno, I vote to leave the batch hack in list for the moment, and search and replace it to munch as soon as the branches merge.
pmichaud I'd prefer to s/r to the .[^$n] form 13:17
but yes, we can change the tests after merge
moritz_ waitwaitwait
13:17 christine joined
moritz_ those are not rakudo tests 13:17
they are spectests
if .batch isn't spec (and I kinda thought it was), it goes.
pmichaud it was spec, yes.
it may not be
moritz_ I couldn't find it in the spec
colomon moritz_: spec is shifting here.
pmichaud even if it remains in the spec, it's something you do to an Iterator
and gather/take (and the various operators) return lists, not iterators 13:18
this is partially why I've been saying to be careful about building too much on master's broken implementation, because it was deeply and fundamentally broken at a design level
moritz_ so I see two options 13:19
1) fudge them for master
2) temporarily disable them in master's t/spectest.data
pmichaud 3) remove the incorrect usages of .batch from t/spec
moritz_ 3) is precursor to 1) xor 2) 13:20
for which I have patch locally
pmichaud I'm fine with temporarily disable
that's easier than fudging
moritz_ ok
I'll see which ones need disabling 13:21
colomon why for heaven's sake would we turn off tests giving us useful information in the name of conforming to a spec that is changing this week?!
pmichaud colomon: because they don't conform to the spec
they dont conform to the existing spec
this is what I have been saying all along.
moritz_ colomon: because they neither confrorm to existing nor to the new spec
colomon I give up. 13:22
13:22 colomon left, rgrau_ joined, rgrau left
moritz_ hugme: hug colomon 13:23
hugme hugs colomon
dalek kudo: 99b6dd0 | pmichaud++ | build/PARROT_REVISION:
Eliminate hyphen from PARROT_REVISION.
 13:24
moritz_ sorry, I was sloppy
pmichaud np :) 13:25
moritz_ oh 13:27
.batch does appear in S03
on a list
4050: (@dwimmyside xx *).batch(@otherside.elems)
13:29 arthur-- left
pmichaud well, even in that case it would be wrongish, as it would need to be .batcharg instead of .batch 13:31
(.batch flattens)
moritz_ in S03:4050 it's meant to flatten, no? 13:32
pmichaud I'm still trying to figure that out 13:33
moritz_ it would be pretty fatal if not
for example if you <<+<<
and the RHS doesn't flatten
13:33 Zapelius joined
pmichaud yes, okay, flattening 13:33
moritz_ it passes an array as the second operand to + at some point
and works on the number of elements. D'oh. 13:34
13:34 JimmyZ joined
pmichaud I'd rewrite that now as (@dwimmyside xx *).flat.[^@otherside.elems] 13:35
moritz_ no need for the .elems actually 13:36
pmichaud that too :-)
moritz_ unless +@a and @a.elems are allowed to disagree
Zapelius what's the replacement for p5 hex() ? 13:37
pmichaud afk for a bit
moritz_ Zapelius: :16() 13:38
rakudo: say :16('a0')
p6eval rakudo cfbeb5: OUTPUT«160␤»
Zapelius k, thanks :)
moritz_ works with other bases too :-)
13:42 meppl joined 13:43 arthur-- joined
szabgab literal, any time to update App::Grok? 13:43
literal you have something in mind? 13:44
13:44 Zapelius left
szabgab I don't know, azawawi told me to ping you :) 13:45
he is on #padre
literal ok
szabgab as I'd like to make Padre ready for the release of Rakudo *
so we have a superb Perl 6 support
13:46 patspam joined 13:48 Guest50632 left
moritz_ szabgab++ 13:49
szabgab I am just talking 13:50
moritz_ talking can be powerful
especially if you talk in a programming language, and write it down :-) 13:51
13:52 plobsing joined, uniejo left
PerlJam imagines a future when Padre is written in Perl 6. 13:52
szabgab rakudo: say (1,2 X 4,7).perl; my @x = (1,2 X 4,7); say @x.perl; 13:54
p6eval rakudo cfbeb5: OUTPUT«(1, 4, 1, 7, 2, 4, 2, 7)␤[(1, 4), (1, 7), (2, 4), (2, 7)]␤»
szabgab why the difference?
moritz_ seems wrong to me 13:55
PerlJam indeed.
szabgab rakudo: say (1,2 X 4,7).elems
p6eval rakudo cfbeb5: OUTPUT«4␤»
szabgab rakudo: say (1,2 X 4,7).perl
p6eval rakudo cfbeb5: OUTPUT«(1, 4, 1, 7, 2, 4, 2, 7)␤»
PerlJam that's really wrong.
moritz_ actually I'd expected it to be just the other way round, if I didn't knew that list assignment flattening is broken in rakudo
szabgab so my slides are useful for something after all :)
pmichaud @x flattens 13:56
yes, it looks broken.
pmichaud tries it in list branch
PerlJam pmichaud: how goes the list work? 13:57
pmichaud PerlJam: we're making good progress, but it's definitely taking longer than I'd like.
szabgab rakudo: for (1, 2, 3, 4, 5) -> $x, $y? { say "$x $y"; }
p6eval rakudo cfbeb5: OUTPUT«1 Any()␤2 Any()␤3 Any()␤4 Any()␤5 Any()␤»
pmichaud > say (1,2 X 4,7).perl
((1, 4), (1, 7), (2, 4), (2, 7))
> my @x = (1,2 X 4,7); say @x.perl;
[1, 4, 1, 7, 2, 4, 2, 7]
>
that's what I would expect, anyway (and it's what the list branch gives) 13:58
13:58 wallberg left
moritz_ pmichaud++ 13:59
szabgab pmichaud, good, as that's what my slides say, as it was working some time ago
so I just wait till it gets merged
what about the for loop with the optional value?
rakudo: for (1, 2, 3, 4, 5) -> $x, $y? { say "$x $y"; }
p6eval rakudo cfbeb5: OUTPUT«1 Any()␤2 Any()␤3 Any()␤4 Any()␤5 Any()␤»
13:59 mjk left
szabgab that looks broken too 14:00
pmichaud still broken in the branch
I'm guessing .count is giving wrong results
no, that's not it 14:01
hrm.
oh!
the problem is that our 'for' statement still isn't using 'map'
14:01 ash__ joined
pmichaud (1,2,3,4,5).map(-> $x, $y? { say "$x $y"; }) 14:01
rakudo: (1,2,3,4,5).map(-> $x, $y? { say "$x $y"; }) 14:02
p6eval rakudo cfbeb5: ( no output )
pmichaud rakudo: (1,2,3,4,5).map(-> $x, $y? { say "$x $y"; }).eager
p6eval rakudo cfbeb5: OUTPUT«1 2␤3 4␤5 Any()␤»
szabgab it should give 3 iterations, 1 2 3 4 and 5 Mu ?
pmichaud ...and that we don't have sink context yet :)
but yes, 3 iterations, as shown by the .map above 14:03
szabgab What would be the value of the $y in the last iteration?
pmichaud Any
(parameters default to Any)
14:03 snarkyboojum left
szabgab ok, I'll read about it 14:03
pmichaud (unless declared with a different constraint)
14:05 proller left
szabgab ty 14:06
is there any difference between [1, 2] and (1, 2) ?
14:06 proller joined
JimmyZ array and list? 14:06
szabgab rakudo: my @words = <moo foo bar moose bu>; say @words.sort.perl 14:07
p6eval rakudo cfbeb5: OUTPUT«("bar", "bu", "foo", "moo", "moose")␤»
szabgab rakudo: my @words = <moo foo bar moose bu>; say @words.sort: { $_.bytes } 14:08
p6eval rakudo cfbeb5: OUTPUT«bumoofoobarmoose␤»
pmichaud szabgab: (1, 2) is a parcel, [1, 2] is a scalar array
szabgab rakudo: my @words = <moo foo bar moose bu>; my @x = @words.sort: { $_.bytes }; say @x.perl
p6eval rakudo cfbeb5: OUTPUT«["bu", "moo", "foo", "bar", "moose"]␤»
szabgab rakudo: my @words = <moo foo bar moose bu>; my @x = @words.sort.perl; say @x.perl
p6eval rakudo cfbeb5: OUTPUT«["(\"bar\", \"bu\", \"foo\", \"moo\", \"moose\")"]␤»
szabgab rakudo: my @words = <moo foo bar moose bu>; my @x = @words.sort; say @x.perl 14:09
p6eval rakudo cfbeb5: OUTPUT«["bar", "bu", "foo", "moo", "moose"]␤»
szabgab so if I put .perl on the result of sort I get () - a parcel and if put .perl on the @x then I get [] scalar array 14:10
pmichaud yes
szabgab now I only need to read about parcel and scalar array :)
pmichaud that's the current spec. I was discussing that we might want to change it (about 12 hours ago :)
I think that both should return ( ) 14:11
masak Alias++ # use.perl.org/~Alias/journal/40398
tadzik Is .perl something more than something like Data::Dumper? Can it be read by Perl6, evaled or something?
pmichaud Indeed, Alias++ 14:12
14:12 am0c left
pmichaud tadzik: .perl is supposed to produce a string that when evaled will produce the value that was .perl'd 14:12
tadzik So it's useful for either debugging or serialization? 14:13
pmichaud tadzik: but our current definition (in the spec) almost never achieves that. Yesterday I decided to invoke hubris and start changing Rakudo so that .perl would work more the way I think it should... but that started deviating waaaay too far from the spec so I abandoned it
masak rakudo: class A { has $.foo }; say A.new(:foo(42).perl 14:14
p6eval rakudo cfbeb5: OUTPUT«===SORRY!===␤Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 11␤»
masak rakudo: class A { has $.foo }; say A.new(:foo(42)).perl
p6eval rakudo cfbeb5: OUTPUT«A.new()␤»
pmichaud and the discussion that TimToady++ and I had on the subject tended more towards strings and less about .perl (and I have other priorities right now) so I skipped it for now.
masak: yes, that's an example. But I'm more concerned with something like
14:14 Sanitoeter joined
masak rakudo: my $code = { say "OH HAI" }; say $code.perl 14:15
p6eval rakudo cfbeb5: OUTPUT«{ ... }␤»
pmichaud rakudo: my @a = 1,2; say (1, @a, 3).perl;
p6eval rakudo cfbeb5: OUTPUT«(1, [1, 2], 3)␤»
masak o.O
shouldn't that flatten?
pmichaud right
but the spec says that arrays and lists are produced using brackets
(ignoring their flattening nature) 14:16
I think that "(1, 1, 2, 3)" is equally wrong.
mathw oh yes, Alias++
masak pmichaud: it's what I would expect.
pmichaud I think the result should be "(1, (1, 2), 3)"
.perl shouldn't automatically flatten a parcel, either.
masak oh, I think I see what you're saying. 14:17
pmichaud if I do .perl on a Parcel, it should give me back a string that evals to a Parcel
(to an equivalent Parcel)
masak aye.
mathw not just some random Parcel :) 14:18
pmichaud same with (1,2) X (3,4) --- I expect .perl to not flatten the result 14:19
nor to itemize the intermediate lists into [...] arrays 14:20
dimid rakudo: (any <1 2 3>).perl
p6eval rakudo cfbeb5: ( no output )
JimmyZ It's useful when you want to use it.
oh, wrong
tadzik where do I find some comprehensive docs about Perl 6 types?
dimid rakudo: say (any <1 2 3>).perl
p6eval rakudo cfbeb5: OUTPUT«any("1", "2", "3")␤»
dimid rakudo: say (any<1 2 3>).perl
p6eval rakudo cfbeb5: OUTPUT«any()␤»
masak tadzik: spec.pugscode.org/
tadzik: specifically, S02 and S32.
dimid why any <1 2 3> != any<1 2 3> ? 14:21
tadzik thanks masak 14:22
pmichaud dimid: the lack of the space changes the circumfix angles into postcircumfix angles
masak dimid: because the latter is hash indexing.
pmichaud masak++ says it better :)
masak dimid: in the case of distinguising between infix and postfix operators, whitespace is significant in Perl 6. 14:23
dimid i see. thanks
masak solution: use parens, or whitespace. preferably consistently. :) 14:24
dimid =)
JimmyZ It's very different from Perl 5
14:24 Trashlord left
tadzik rakudo: my $a = (tim => 'toady', foo => [1, 2, 3]).perl; my $b = eval $a; say $b.WHAT; say (tim => 'toady', foo => [1, 2, 3]).WHAT 14:26
p6eval rakudo cfbeb5: OUTPUT«Seq()␤Parcel()␤»
masak JimmyZ: yes. the price you pay for the increased flexibility in Perl 5 is that sometimes you won't know whether something is a term or an operator.
tadzik why the difference?
masak tadzik: because of the assignment to $b, I'd think. 14:27
pmichaud tadzik: assigning a parcel to a scalar converts it into a Seq
tadzik I see. Seq is a scalar?
masak rakudo: my $b = (tim => 'toady', foo => [1, 2, 3]); say $b.WHAT; say (tim => 'toady', foo => [1, 2, 3]).WHAT
p6eval rakudo cfbeb5: OUTPUT«Seq()␤Parcel()␤»
pmichaud Seq is a list
masak .oO( "a Seq in the hand is worth a Parcel in the bush" ) 14:28
tadzik hmm 14:31
rakudo: my %a = foo => 'bar', asd => 5; my $b = %a; say $b.WHAT
p6eval rakudo cfbeb5: OUTPUT«Hash()␤»
tadzik what happens here, am I magically holding a hash in %-prefixed variable?
erm, $-prefixed 14:32
pmichaud sure, scalars can hold almost anything
tadzik ah, okay
pmichaud it's kind of like a reference in p5
tadzik so prefixes are just for users in this case?
pmichaud but you don't have to do anything to dereference (or to worry about the reference being there)
tadzik yeah, I'm really happy about that
pmichaud the prefixes (they're called "sigils", btw) do control flattening
tadzik what do you mean? 14:33
pmichaud rakudo: my @a = 1,2,3,4; my $b = @a; say (@a, 5, 6).elems; say ($b, 5, 6).elems;
p6eval rakudo cfbeb5: OUTPUT«6␤3␤»
pmichaud array sigils will flatten in list context, while scalar sigils won't
14:34 EvanCarroll joined
tadzik mhm 14:34
EvanCarroll zomfg, regexp::grammars is awesome
mathw I keep forgetting what a Parcel actually is 14:35
I really need to re-read that part of the spec
pmichaud In general, a Parcel is the thing that is constructed by &infix:<,>
that's the way I tend to remember it
so, (3, 4) is a Parcel 14:36
14:36 gbacon joined, plobsing left
mathw and Parcels do the expected sort of thing if you treat them like lists 14:36
pmichaud yes
mathw Is the argument list for a subroutine call also a Parcel? 14:37
pmichaud mathw: yes
in fact, that's really why Parcels exist
so that they can be interpreted in different ways depending on context
mathw that's quite nice
so anything with (..., ...) is actually the same thing
pmichaud conceptually, yes
mathw and then you do something with it and it becomes what you're asking it to be 14:38
pmichaud in that case, the Parcel gets converted to a Capture which is then used for signature binding
mathw and Capture provides the stuff that a subroutine needs to make its parameters visible to its block, or something like that 14:39
pmichaud yes, as well as knowing how to deal with slurpy arguments, and named arguments
a Capture is able to take a Parcel like 1, a=>4, :b(5) 14:40
and know how to map that onto a variety of different signatures
mathw Excellent
I have much more clue now, thanks :)
14:42 Guest23195 joined
tadzik moritz_: your grammars example from perl5-to-6 does not compile for me 14:49
wklej.org/id/351219/ 14:50
14:50 slavik left 14:51 slavik joined, [mark] left
tadzik and there's a typo, URL in grammar declaration, URI in URI.parse 14:51
14:52 pmurias joined
tadzik I believe - needs to be escaped 14:53
ash__ yes it does need to be escaped now 14:54
baest also inside a char class?
ash__ alpha: say /<[-]>/ ~~ '-' 14:55
p6eval alpha 30e0ed: OUTPUT«perl6regex parse error: Unescaped '-' in charlist (use '..' or '\-') at offset 267, found '-'␤in Main (file <unknown>, line <unknown>)␤»
ash__ i like that error message better than master rakudo's error message
rakudo: say /<[-]>/ ~~ '-'
p6eval rakudo 99b6dd: OUTPUT«===SORRY!===␤Obsolete use of hyphen in enumerated character class;in Perl 6 please use .. instead at line 11, near "]>/ ~~ '-'"␤»
pmichaud rakudo master's message comes from STD, though. 14:56
std: say /<[-]>/ ~~ '='
p6eval std 31266: OUTPUT«ok 00:01 109m␤»
pmichaud huh
ash__ :P does it?
pmichaud did STD get rid of that particular warning?
JimmyZ pmichaud: master sometimes gives me wrong messages 14:57
jnthn std: say /<[a-z]>/ ~~ '='
p6eval std 31266: OUTPUT«===SORRY!===␤Unsupported use of - as character range; in Perl 6 please use .. at /tmp/TDhv24tkYR line 1:␤------> say /<[a-z⏏]>/ ~~ '='␤Parse failed␤FAILED 00:01 105m␤»
jnthn I think it just got smarter.
14:57 hafos left
JimmyZ pmichaud: I'm sure not only this one 14:57
pmichaud std: say /<[-z]>/ ~~ '.'
p6eval std 31266: OUTPUT«ok 00:01 106m␤»
ash__ - by itself seems to be fine, but infix it complains
pmichaud okay, so STD only complains if - appears where one might expect a character range 14:58
14:58 rgrau_ left
pmichaud I can fix that (but not in the current parrot release) 14:58
mathw because you can always reorder it if you really meant -
ash__ -abc vs a-bc etc. (where its not supposed to be a range)
mathw yup 14:59
&
ash__ std: say /<[a\-z]/ ~~ 'a'; 15:00
p6eval std 31266: OUTPUT«===SORRY!===␤Unable to parse metachar at /tmp/bA10yw9ZYW line 1:␤------> say /⏏<[a\-z]/ ~~ 'a';␤Couldn't find final '>'; gave up at /tmp/bA10yw9ZYW line 1:␤------> say /<[a\-z]⏏/ ~~ 'a';␤ expecting any of:␤ character
..class …
ash__ std: say /<[a\-z]>/ ~~ 'a';
p6eval std 31266: OUTPUT«ok 00:01 106m␤»
ash__ i wonder what that means
is that an escaped - or is that a \ and a - separate? 15:01
pmichaud it's an escaped -
ash__ got ya, you always need \\ to mean \, right? (or commonly?) 15:02
pmichaud generally, yes.
dalek kudo: aff0402 | jonathan++ | src/builtins/Mu.pir:
Remove a fossil.
 15:03
kudo: f78f232 | jonathan++ | (3 files):
Switch Multi to inherit from Routine, not Code. Implement candidates on Routine

to an unrelated issue to do with augment and mapped Parrot types, .cando isn't found on multis yet; that needs a seperate fix. Patch courtesy of (Maxim Yemelyanov)++.
kudo: 207c528 | jonathan++ | src/metamodel/Attribute.nqp:
Start to stub in attribute traits a bit.
jnthn yaypatches :-) 15:04
masak yayjnthn!
15:05 bphillips joined
jnthn yaymasak! 15:05
masak++ # saw your Bug post
er, Buf
masak keeps typoing it as "Bug" too :)
jnthn ...wow, I could claim that as legitimate Freudian or legitimate typo. :-) 15:06
masak :)
pugssvn r31267 | moritz++ | [t/spec] replace .batch($x) with .[^$x]
pmichaud the more I work with the new list/iterator implementation, the more I'm convinced it's Right. 15:07
masak pmichaud++ 15:08
pmichaud: sounds like you're suffering from True Believer Syndrome :P
pmichaud maybe. but so many things that were convoluted in the previous (four or five) systems are just straightforward in this one.
masak that's a really good sign. 15:09
15:09 Trashlord joined 15:12 Mowah left
jnthn lolibloggedforonce # use.perl.org/~JonathanWorthington/journal/40399 15:12
pmichaud: Current spectest status for list? :-) 15:13
pmichaud about 500 failing tests 15:14
but little fixes are still recovering lots of tests
and we get lots of todo passes
masak ship it!
jnthn pmichaud: 500 fails is less than I thought we might be at by this point. Excellent!
pmichaud++
moritz_ jnthn: I have to say that I'm really impressed by those russian hackathoniers
dalek kudo: 410db0f | moritz++ | t/spectest.data:
disable some tests that now loop, due to .batch being replaced by .[^$num]
pmichaud yes, I figure with a little luck I should be able to merge today
moritz_ jnthn: can't we get a bunch of those as regular contributors? :-) 15:15
jnthn moritz_: Yes, me too. I think at least one was also Ukrainian. :-)
moritz_: A couple said they may :-)
moritz_ that would be great
jnthn Yes, I'd like it.
moritz_ any nationality is fine, really
did we already have a Kiev release? 15:16
jnthn No.
But I'd like to propose we do.
And I know they'd like that here. :-)
15:16 macdaddy joined
pmichaud I'd say go ahead and tentatively name July "Kiev.pm" 15:17
masak +1
moritz_ I'll put it in the release guide
pmichaud moritz_++
hmmm 15:18
15:18 [mark] joined
pmichaud ( <a b>, 'c' ).join(',') # what should this produce? 15:18
moritz_ I'd expect 'a, b, c' 15:19
pmichaud me too
so what's responsible for flattening the parcel, there?
moritz_ because that's what we get from join(', ', <a b>, 'c')
pmichaud does .join automatically flatten the list?
cono Kiev++
jnthn o/ cono :-)
masak rakudo: my $d = ''; class Foo { method some_meth_1 {$d = $d ~ self;}}; for '0' .. '5' -> $a { sub; };
p6eval rakudo 99b6dd: OUTPUT«===SORRY!===␤Symbol '$ss_SS_S_S__S_S_s' not predeclared in <anonymous> (/tmp/RF3RHI6UOR:1)␤»
cono jnthn: are still in Kiev? 15:20
jnthn cono: Yes
cono u*
jnthn cono: Leave tomorrow.
moritz_ masak: what is sub; supposed to do?
cono jnthn: how is the boat trip?
pmichaud I see several possibilities:
dalek kudo: 2721911 | moritz++ | docs/release_guide.pod:
propose Kiev as name of next release, with link to jnthn++'s blog post
jnthn Boat trip was nice. :-)
Relaxing. :-)
masak moritz_: I have *no* idea. this is JimmyZ privmsging me strange bug reports. :)
jnthn And another way to see Kiev.
pmichaud (1) Parcel.join flattens the parcel (meaning that various methods act differently on parcels)
cono jnthn: it's a pitty that's a work day :( 15:21
pmichaud (2) .join always flattens its invocant
jnthn Aye
pmichaud (3) some other magic I'm not seeing
jnthn If it had been weeknd, maybe more people could have come.
cono sure 15:22
15:22 bjarneh joined
ash__ .join seems like another way of writing: <a b> ~ ',' ~ 'c' to me 15:22
jnthn pmichaud: My guess is that join does it, or that join perhaps only exists in Cool and thus always puts what it has in flattening list context.
pmichaud well, I'm sure that .join is at least in Any 15:23
moritz_ all list methods are in Any
jnthn Oh.
Well, same but s/Cool/Any/. :-)
ash__ unless its supposed to call <a b>.join(',') too
pmichaud I'm going to assume .join flattens for now.
ash__ i'd expect to have to write that as (|<a b>, 'c').join(',') (but maybe thats just me) 15:24
pmichaud ash: what if I had said instead 15:25
my @a = <a b>; (@a, 'c').join(',')
masak rakudo: my $d; class A {method x { $d }}; for () { sub }
p6eval rakudo 99b6dd: OUTPUT«===SORRY!===␤Symbol '$ss_SS_S_S__S_S_s' not predeclared in <anonymous> (/tmp/I5MPO4gCTY:1)␤»
masak locally, "Symbol '$d' not predeclared in <anonymous>"
masak , honsetly confused, submits rakudobug 15:26
ash__ i'd stll expect @a ~ ',' ~ 'c' to have the same result as .join(','), but again, i might be in the wrong mindset here
pmichaud ash__: okay, you're consistent at least :)
moritz_ well
masak std: my $d; class A {method x { $d }}; for () { sub } 15:27
p6eval std 31266: OUTPUT«===SORRY!===␤Malformed block at /tmp/irG3_MPjN_ line 1:␤------> class A {method x { $d }}; for () { sub ⏏}␤ expecting any of:␤ name␤ routine_def␤ trait␤Parse failed␤FAILED 00:01 108m␤»
pmichaud one could argue that the other interpretation should be (@a, 'c').flat.join(',')
moritz_ given that ~(1..3, 4) now produces "1..3 4", maybe not joining the <a b> might not be too bad
ash__ i guess i kinda think of join as a list prefix [~], so, [~] $a, 2, 'bob'; but with the ability to say what goes in between the things your joining, 15:28
rakudo: say [~] 'hello', ' ', 'world', ' and ', ('hello', 'world').join(' ') 15:29
p6eval rakudo 99b6dd: OUTPUT«hello world and hello world␤»
JimmyZ hopes there will be web entry for submitting rakudo bugs
moritz_ if it means screwing up like parrot's trac, please not. 15:30
afk
pmichaud when is #phasers, again? I always forget the time. (utc preferred) 15:31
masak pmichaud: three and a half hours from now. 15:32
pmichaud so, 19:00 utc
I can remember the utc :-)
masak JimmyZ: I understand why you think that. I don't think it's a good idea. I'm extremely happy with the way things are set up now.
yes, 19:00 UTC. :)
pmichaud from t/spec/S32-list/sort.t, line 20 15:35
my @s = sort(:values(@a));
that looks like a bogus line to me.
I don't think one can use a named argument to set slurpy positionals 15:36
jnthn oh, #phasers
So for me it's...10pm. That's reasonable. :-)
pugssvn r31268 | pmichaud++ | [t/spec]: I don't think one can pass slurpy positionals by name. 15:38
jnthn pmichaud: I'm not too convinced we can either, though there's a trac ticket on that matter too...
er
RT
pmichaud also, in sort.t: 15:40
my @a = (1.1,2,NaN,-3.05,0.1,Inf,42,-1e-07,-Inf).sort;
my @e = (NaN,-Inf,-3.05,-1e-07,0.1,1.1,2,42,Inf);
my @s = sort @a;
is(@s, @e, 'array of mixed numbers including Inf/NaN');
... is there a reason for sorting @a twice? 15:41
masak no.
must be a thinko.
pmichaud I'm guessing we don't need the .sort
jnthn Looks sort of weird to me. 15:42
pmichaud
.oO( arggggh )
tadzik hmm 15:44
rakudo: fsck you
p6eval rakudo 99b6dd: OUTPUT«Could not find sub &you␤ in main program body at line 11:/tmp/gwA74bXPgM␤»
tadzik some sort of indirect notation? how does it work?
pmichaud tadzik: undeclared barewords are assumed to be subroutine invocations 15:45
so, since there's nothing called "you" declared, it's assumed to be a subroutine call to &you
tadzik yeah, I see. But why does it complain about you, not fsck? 15:46
pmichaud because that happens first
tadzik mhm
pmichaud we call &you, and pass the results of that to &fsck
so the call to &you fails first
tadzik I see
ash__ rakudo: &foo(); sub foo { say 'hello world' }; 15:47
[particle] &you could define &fsck
p6eval rakudo 99b6dd: OUTPUT«hello world␤»
[particle] so you can't fail on &fsck too
jnthn pmichaud: Just spotted review-notes.txt :-) 15:49
tadzik interesting
masak rakudo: our sub you { our sub fsck($) { say "OH HAI" } }; fsck you 15:50
p6eval rakudo 99b6dd: OUTPUT«OH HAI␤»
15:50 tedv joined
ash__ you fsck would fail in masak's example, because of the way the sub's are defined 15:50
jnthn I don't think it would.
ash__ rakudo: our sub you { our sub fsck($) { say "OH HAI" } }; you fsck 15:51
p6eval rakudo 99b6dd: OUTPUT«Not enough positional parameters passed; got 0 but expected 1␤ in 'fsck' at line 11:/tmp/4vVZMhax4Y␤ in main program body at line 11:/tmp/4vVZMhax4Y␤»
jnthn Oh
ash__ ah, it fails for another reason ><
[particle] :)
jnthn well, OK...yes
fsck
:-)
masak hm. 'you fsck' sounds more offensive than 'fsck you' :)
tadzik maybe rakudo couldn'd stand it 15:52
ash__ i pronounce fsck as 'f-s check'
sounds like a command
jnthn masak: Well, it depends if it was a verb or an nouny-adjective in the first one. ;-)
ash__ s/you/sudo/ now its really a command
masak ash__: I somehow manage to pronounce it as a monosyllable.
the 'badges' list of stackoverflow is inspiring. stackoverflow.com/badges 15:56
15:58 Ross joined
masak nom & 15:59
15:59 masak left 16:00 ash__ left 16:02 cdarroch joined, cdarroch left, cdarroch joined, tedv left, gbacon left, JimmyZ left, sukria left, dimid left, eiro left 16:03 sukria joined 16:04 dimid joined, eiro joined, gbacon joined, JimmyZ joined 16:05 tedv joined
moritz_ hopes he didn't offend colomon++ earlier when talking about the test changes (.batch) 16:11
pmichaud moritz_: yeah, I'm not quite sure what happened there. Maybe it was me. 16:12
I can understand while colomon++ would be frustrated -- he's been wanting to do iterators and lists for quite a while
s/while/why/
16:17 paroxyzm_ left 16:18 justatheory joined 16:19 ash__ joined 16:21 pmurias left
EvanCarroll stackoverflow.com/questions/3047155...xpgrammars 16:34
16:38 envi^home left 16:39 am0c joined, [mark] left 16:41 JimmyZ left
[particle] i may see colomon tonight 16:51
if i do, i'm sure he'll assault me about it
TimToady blessed are the peacemakers, for they shall be assaulted. 16:52
pmichaud afk, lunch 16:53
[particle]: any chance you could make it to #phasers today at 1900 utc?
(2h07m from now)
one of the discussion items will be timing of the R* release (and what it will be based on) -- your contributions/opinions would be welcomed. 16:54
afk, lunch
[particle] pmichaud: i believe so, i'm sitting in that channel now
17:00 rob left 17:12 dakkar left
ash__ is there a non-unicode alternative to: $¢ ? 17:15
moritz_ I don't think we can use any non-unicode characters in the chat her 17:16
here
[particle] someone needs to write English.pm6 17:17
ash__ i mean, is there a way to tyepe that with ascii only?
type*
kinda like how you can use << or «
[particle] perl 6 assumes unicode support
moritz_ ash__: don't think so 17:18
17:18 arnsholt joined
[particle] i don't know of an ascii-equivalent for $¢ 17:18
should be easy to search STD.pm for it
ash__ got ya, so far, most unicode key parts have had some way of doing it without
arnsholt takadonet: *prod?*
moritz_ ash__: that's right, but $¢ isn't something that most users will need 17:19
ash__ makes sense, but what if you make parrot without icu? will it still work? (just wondering)
[particle] then parrot will be latin-1 only 17:20
and still have that char iirc
but upcase/downcase/etc will fail for unicode 17:21
moritz_ it's not like parrot can't store and compare UTF-8 strings without icu
[particle] take that with a grain of salt, i'm not remembering the details precisely
moritz_ and variable lookup is just lookup in hash tables 17:22
ash__ i just wasn't sure if it could parse $¢ without icu
[particle] it'll parse
moritz_ I'm pretty sure it can
17:22 meppel joined
arnsholt But if the source is UTF-8 $¢ will be interpreted differently, no? 17:22
17:23 ggoebel joined
[particle] ¢ is dec 162 iirc 17:24
moritz_ arnsholt: yes, but parrot can still decode UTF-8 to codepoints 17:25
[particle] ascii is 7-bit, so there's room for another 128 characters in a one-byte encoding
TimToady latin-1 is just Unicode restricted to <256
[particle] unicode puts ascii and latin1 in the first byte
17:26 Mowah joined
[particle] since ¢ is latin1, it's unicode and ascii (in the usa) 17:26
17:26 meppl left
[particle] other locales can use a different charset past ascii in one-byte encodings, but then they're not unicode-compliant 17:27
moritz_ well 17:28
the just don't have the byte value = codepoint identity
[particle] which means they have to be transcoded 17:29
TimToady which must be declared explicitly
17:29 meppel is now known as meppl
TimToady S02:52 requires the default to be Unicode, alwys. 17:30
and we're not gonna touch the locale mess 17:32
(without an explicit declaration to do so)
17:32 xabbu42 joined
moritz_ +1 17:32
[particle] so we're always safe in assuming <256 is ascii+latin1.
unless there's an explicit declaration otherwise 17:33
TimToady <256 *what* is the question :)
[particle] :)
TimToady you are not safe if you mix up UCS-1 with UTF-8
moritz_ well, there are no byte values >= 256 17:34
so I kinda hope [particle] is talking about codepoints
TimToady ah, a european admitting that "octets" is redundant...what is this world coming to? :)
TimToady ignores the fact that there used to be other byte sizes back in the day... 17:37
[particle] \n\r
ebcdic
...there's a lot to ignore... 17:38
nobody has proposed 40-bit floats
TimToady 12-bit words are pretty awesome
[particle] like a clarinet 17:39
TimToady I don't like licorice
arnsholt Mmmm. Salmiak =D 17:40
17:41 xabbu42_ joined, xabbu42 left, xabbu42_ is now known as xabbu42
Su-Shee salty ones! 17:43
17:43 rgrau joined
pmichaud back again 17:44
Su-Shee arnsholt: meet my favorite shop around the corner: www.kado.de ;) 17:45
arnsholt =D 17:46
Om nom nom...
Su-Shee _great_ varieties from all countries. 17:47
TimToady <sorear> TimToady: Why is @::ORIG kept as numbers instead of characters?
well, two reasons
an array of integers takes less memory
but the main reason is to move into the mindset where characters are integers, to make it easy to look up character properties by numeric index 17:48
and backtranslating from the integer with chr() is considered a temporary hack till we get the Unicode tables into Perl 6 directly 17:50
I agree that it would be very nice if p5 would let you look up char properties by integer
but it doesn't as far as I know 17:51
17:51 diakopter joined, rgrau is now known as kidd 17:52 synth joined
TimToady <sorear> TimToady: what is the purpose of %lexer_cache? Surely anything in %lexer_cache would also be in %::LEXERS 17:56
yes, I thought it was redundant when it was added (I forget by whom), but there was some contradictory NYTProf data at the time, so I left it in as relatively harmless 17:57
diakopter 'twasn't 18:01
TimToady <sorear> TimToady: what is the purpose of CursorBase.pmc 1940-1973?
18:01 clintongormley joined
TimToady not sure what you're asking here, so I'll assume you figgered it out unless you ask something more specific... :) 18:01
diakopter I think there were some commits after that q 18:02
18:02 bjarneh left
diakopter (that may have offset the line numbers) 18:03
18:08 broquaint left
pmichaud #phasers in 43 18:17
I just pasted a longish report and some discussion items 18:18
18:22 xabbu42_ joined, xabbu42 left, xabbu42_ is now known as xabbu42
TimToady <sorear> TimToady: Can you explain how $::PREFIX is used in RE_ast? 18:26
$::PREFIX is supposed to track the fates that have already been determined for the current alternative 18:28
there is likely a cleaner way to do this with parameters... 18:29
18:32 xabbu42_ joined, xabbu42 left, xabbu42_ is now known as xabbu42 18:34 xabbu42_ joined, xabbu42 left, xabbu42_ is now known as xabbu42 18:37 am0c left, pyrimidine joined 18:38 Wolfman2000 joined 18:39 solarion joined 18:51 Schwern left 18:53 masak joined
Tene hallo, masak. :) 18:55
masak hullo, Tene. :)
sorear What's an inversion list? 18:56
TimToady see google
but short answer, run length encoding that flips binary state
18:56 xabbu42 left, xabbu42_ joined
lue ohai 18:59
18:59 xabbu42_ left, xabbu42 joined
TimToady but I don't think it buys much over a list of range transition values where even/odd indexes tell you whether it was a 0 or 1 range 19:00
pmichaud that's what I'm actualy using 19:01
masak #phasers in 0. :)
TimToady on a low level, inversion list can do subtract and compare with 0
pmichaud basically, (start,end) pairs
19:01 perlygatekeeper left
TimToady which may be faster than comparing two nums 19:01
EvanCarroll is damian on freenode? 19:04
PerlJam EvanCarroll: no
19:04 xabbu42_ joined, xabbu42 left, xabbu42_ is now known as xabbu42, cono left
TimToady EvanCarroll: email is your best bet 19:04
EvanCarroll yea, I emailed 19:05
=(
19:06 cono joined
EvanCarroll how close is regexp::grammars to perl6 ? 19:07
19:08 xabbu42_ joined, xabbu42 left, xabbu42_ is now known as xabbu42
moritz_ it's inspired by Perl 6 19:08
tadzik oh, by the way, Perl6 inspirations. Anything like Moose native traits and 'handles' in Perl6? 19:09
TimToady but it gets no lovin from P5, so it's different in many ways
tadzik: those notions were in P6 first :)
EvanCarroll I think p6 has real type traits 19:10
tadzik I thought so at the first thought, but didn't fint it in docs
I wasn't reading specs though
EvanCarroll I've seen `0 with true` in a doc somewhere
or something like that
19:10 PZt left
tadzik everything that's in Moose is in Perl 6 also? 19:11
TimToady no, there are some divergences
tadzik is there anything worthy missing?
TimToady I don't think there's anything major that can't be done some other equivalent way, but we'll find out more when Moosers start coming to P6 after R* is out 19:12
tadzik ah, R* is for Rakudo Star 19:13
EvanCarroll it comes before Rakudo--
tadzik at first I thought you're avoiding the name "Rakudo" not to bring bad luck or something :)
Rakudo--?
masak attributes have metaclasses (or meta-somethings) in Moose. they're slightly less magical in Perl 6.
sorear literal: one thing that could use updating is your synopsis mirrors 19:14
EvanCarroll <rule: pair> <delim=(\'|\"|\$\$)> <char=(.)> <delim2=(?{ $MATCH{delim} })>
why doesn't that work =(
moritz_ EvanCarroll: doesn't look very P6ish to me 19:15
ash__ is that perl6?
moritz_ EvanCarroll: if you translate it to Perl 6 grammars first, I'm sure somebody will try to help you here
EvanCarroll it is Regexp::Grammars
=(
moritz_ and thus mostly off-topic 19:16
unless you make it on-topic by translating it to Perl 6.
EvanCarroll Perl 6 will run on parrot, which will run perl 5.
19:16 xabbu42 left
EvanCarroll Thus, on topic. 19:16
19:16 xabbu42 joined
lue walks around the back for a second entrance :) 19:16
TimToady what's this, guilt by association? :)
ash__ Parrot doesn't run on perl 5... parrots something else 19:17
EvanCarroll parrot is vm that has/will-have an implimentaiton of perl5.
Tene That's slightly more on-topic for #parrot on irc.perl.org, and much more on-topic for #perl. :) More seriously, looks like Regexp::Grammars is sufficiently different from Perl 6 grammars that nobody here thinks they can help you.
EvanCarroll ponie or something
tadzik Isn't that Perl 1? (: 19:18
EvanCarroll perl1 = punie.
tadzik Punie — An implementation of Perl 1.0
ah
EvanCarroll moreover, I'm klined from irc.perl.org, and banned from #perl.
tadzik :D
what have you done?
EvanCarroll search.cpan.org/~ecarroll/MooseX-Ty...ntained.pm 19:19
Tene EvanCarroll: ponie is long dead, and I don't know of anybody working on it in recent history. The current work is embedding perl5, which is working to some degree now.
moritz_ EvanCarroll: being banned from irc.perl.org doesn't make perl 5 questions here more on-topic
EvanCarroll: please do try to stay on-topic here. 19:20
Tene I don't know that we're really all that grumpy about on-topic. We've had plenty of off-topic discussions in here, and it's nto really otherwise busy ATM. 19:21
TimToady well, there's OT, and then there's OT that requires studying up on something OT
moritz_ Tene: it's about signal-noise ratio
Tene: if somebody contributes much signal, quite a bit noise is also OK
only noise is not appreciated 19:22
TimToady nobody here wants to study up on Regex::Grammars
lue (There's OnT, acceptable OfT, and OfT.)
EvanCarroll I would have assumed I wasn't the only person using regex::grammars
=(
guess I'm wrong
Tene If anyone here *already knew* about Regexp::Grammars, I wouldn't have a problem with them answering a question.
TimToady me either 19:23
Tene I didn't see a request that people go do research, either.
literal sorear: yeah, I think I'm gonna make grok download them to ~/.grok/ and allow the user to do something like grok --update 19:24
TimToady anyway, you might need to be patient with email to TheDamian--remember he's down under, and electrons flow differently therr
*there 19:25
moritz_ right, opposite electron spin and all
Tene EvanCarroll: good luck. :) 19:26
EvanCarroll I think I'm going to find a different clever way to write a sql parser
masak EvanCarroll: Perl 6 grammars, for example :) 19:27
lue I thought Austrailia got negitrons :)
[or whatever the Antimatter of electron is]
masak lue: also known as "electrons"...
moritz_ lue: electrons *are* negatively charged already :-)
arnsholt lue: positron
Is the anti-particle for electrons
lue Yes. Off to send some positrons 19:28
masak lue: you should read more Asimov.
jnthn back from dinner and a nice stroll :-)
masak jnthn: #phasers!
TimToady EvanCarroll: also consider that TheDamian is probably furiously working on some new presentations for upcoming confs
19:28 sykes_ joined
jnthn masak: It's in 27 minuts? 19:29
PerlJam jnthn: now!
masak jnthn: -31
jnthn oh wtf
moritz_ :-)
PerlJam jnthn: good timing though :)
masak jnthn: I should have reacted earlier today when you said "10 pm" aloud.
19:29 jaldhar left
moritz_ you missed the heated discussion, and can now overthrow our concensus 19:29
masak jnthn: we're in the same time zone, and my #phasers began at 9 pm :)
lue finds out about phasers 19:30
jnthn masak: apart from, I'm not in the same timezone :-)
Kiev is +1 19:31
EvanCarroll I think the problem is that code like this: <delim2=((?{$MATCH{delim}}))> doesn't pop the capture off of the regex stack
so it remains and you can recapture it
masak jnthn: then I'm stumped.
jnthn Unfortunately, the clock I was looking at was...still showing me the time at home. :-/
EvanCarroll or it just writes the output of the code to delim2 and skips the capture entirely.
masak oh, that makes sense.
arnsholt Does rakudo handle block labeling? 19:32
moritz_ nope
arnsholt Right. That explains why I can't get it to work ^^
masak arnsholt: workaround: Bool variables, if statements, manual next/redo/last 19:33
19:33 azert0x left
arnsholt Yeah, that's my plain. A variable outside the inner block and then unlabeled last 19:34
plain -> plan
masak yup.
be glad you don't have nested loops. :/
also known as "cf GGE"
19:35 ShaneC joined
arnsholt I'm implementing CKY. There will be nested loops 19:35
19:35 snarkyboojum joined
Su-Shee likes date nr 4. then I hopefully get more dbi stuff done. 19:35
arnsholt But I won't need to control any of them with any kind of precision
masak arnsholt: it's not hard or anything. it's just a hassle.
arnsholt Yeah, I can imagine
masak one strongly wishes for labels on blocks.
lue afk 19:37
EvanCarroll stackoverflow.com/questions/3048407...t-on-match
there we go.
19:37 Chillance joined
masak EvanCarroll: there we go what? It's still a Perl 5 question. 19:38
this is #perl6.
19:40 ShaneC left
arnsholt What's the correct way to extract the last element of an array? 19:41
Su-Shee arnsholt: I use -1
masak @array.pop :)
Su-Shee: that's *-1
arnsholt masak: But I just pushed() it onto the array! =) 19:42
PerlJam arnsholt: there's also @array[@array.end]
Su-Shee "I use -1 first, then realize something is wrong and try all other options" :)
masak arnsholt: then what do you need it out again for?
rakudo: my @a; @a[-1]
p6eval rakudo 272191: OUTPUT«Cannot use negative index on arrays␤ in 'Array::postcircumfix:<[ ]>' at line 2612:CORE.setting␤ in main program body at line 11:/tmp/xMVnA3hVmo␤»
TimToady "extract" is underspecified 19:43
is it destructive?
.pop is destructive, .[*-1] is not
arnsholt Yeah, I meant non-destructive read
Su-Shee rakudo: my @a = ('cat', 'dog', 'mouse'); @a[*-2].say; 19:44
p6eval rakudo 272191: OUTPUT«dog␤»
masak arnsholt: seriously. if you just put it in, you don't need to extract it, destructively or non-.
Su-Shee rakudo: my @a = ('cat', 'dog', 'mouse'); @a[*-1].say;
p6eval rakudo 272191: OUTPUT«mouse␤»
arnsholt masak: Yeah, that's what I just changed my code to
sorear TimToady: as it happens, dfa/CursorBase already has everything in place to get character properties by number.
masak Su-Shee: parens in the lhs are so Perl 5 :P
PerlJam waits for the *-5 version
masak er, rhs.
arnsholt It's just that the code had the push before I realised I needed to do more stuff, so it felt natural to fiddle with the last element rather than do it before 19:45
TimToady sorear: \o/
Su-Shee masak: I still need to keep a balance between 5 and 6 or I go insane with daily javascript, sql, css on top of all that. ;)
masak Su-Shee: heh. :)
TimToady there is no probably with inserting parens for clarity, as long as they don't replace needed whitespace 19:46
masak Su-Shee: yes, I guess I'm too deep in Perl 6 to need that balance.
rakudo: my @a = <cat dog mouse>; say @a.pick
p6eval rakudo 272191: OUTPUT«dog␤»
TimToady <> are honorary parens :)
Su-Shee also, <> suck to type on a german kezboard.
sorear tadzik: Perl 6 doesn't have a very well-specced metaobject layer yet 19:47
moritz_ everything programming-related sucks on the german keyboard, except the #.
masak with the right software, all keyboards are US keyboards.
sorear tadzik: in particular, we don't have metaclass compatibility, or alternate representations
TimToady sorear: Perl 6 wants the implementations to fight over metamodels :) 19:48
tadzik mhm
Su-Shee moritz_: on all non-english essentially. ;)
tadzik really? 19:49
I have no problems on my Polish :)
„”, «», everything
Su-Shee tadzik: no altgr for {}?
tadzik no, just shift
Su-Shee and []? 19:50
tadzik nothing at all
Su-Shee "essentially non-english but polish.." ;)
tadzik all the fancy chars are on… chars
not on symbols
19:50 rindolf joined
rindolf rakudo: say "ש".chars() 19:51
p6eval rakudo 272191: OUTPUT«1␤»
sorear dfa/Cursor represents character classes using DNF.
This may be overkill
rindolf rakudo: say "Hello+ש".chars()
p6eval rakudo 272191: OUTPUT«7␤»
19:51 patspam left
masak rakudo: say "ש".uc 19:51
p6eval rakudo 272191: OUTPUT«ש␤»
masak huh!
arnsholt Is that a shin?
masak oh, RTL.
moritz_ arnsholt: yes 19:52
arnsholt Then that's the correct answer, AFAIK. No upper/lower case distinction in Hebrew
rindolf arnsholt: yes, it is. 19:53
masak arnsholt: I expected that. 19:54
sorear backlog over.
masak arnsholt: I didn't expect the ␤ to end up to the left. :) but maybe that only happened on my client.
arnsholt Ah, right. Bad interpretation of your huh then, my bad
Ah, right. In mine it's on the right =) 19:55
But which of our clients are in the wrong, I have no idea =)
masak it jumped over to the right here too when I marked it. :/
Emacs seems a bit undecided on the RTL thing.
or maybe it's just helpful in an inexplicable way.
arnsholt Heh. The joys of mixing RTL and LTR
Ah yes. Encoding and Emacs is loads of fun I've been led to understand =) 19:56
rindolf Apparently when szabgab demonstrated it in the console, the Hebrew letters were reported as occupying more than one char.
masak rindolf: results with -e can vary sometimes. 19:57
rindolf: because the encoding of the shell plays into it.
rindolf masak: it was the REPL.
masak rindolf: don't get me started on the REPL :(
actually, I don't remember if the REPL has encoding issues. maybe someone sle does. 19:58
moritz_ it's woefully unicode-unware 20:01
masak s/sle/else/ 20:02
20:03 eternaleye left 20:04 Psyche^ joined 20:07 colomon joined, Patterner left, Psyche^ is now known as Patterner 20:08 xabbu42_ joined, xabbu42 left, xabbu42_ is now known as xabbu42, broquaint joined
frooh is there a web based tryperl6 any more? 20:10
masak frooh: not yet. 20:11
frooh ok, I know there was one with pugs for a while
masak frooh: moritz_++ has plans to bring one back up for Rakudo.
moritz_ and you can help!
the plan is to split it up into byte-sized tasks
frooh well, I was just gonna show a coworker some examples :-)
TimToady in moderation, you can do them here with the evalbot 20:12
moritz_ rakudo: say ?open('README')
p6eval rakudo 272191: OUTPUT«1␤»
frooh eh, not now :-)
moritz_ you can also message the evalbot
sorear TimToady: see _get_unicode_map in dfa/CursorBase.pmc; it's pretty straightforward
frooh hasn't looked at things since nqprx was made master 20:13
TimToady btw, we should do an audit of all cursor functions and either _ or CAP all the ones that shouldn't really be used by users or conflict with their grammar rules
20:14 alester joined
masak frooh: we're sort of starting to see things get back to normal, and in some cases muchly improved. 20:14
Su-Shee starts watching perl6 Digest::MD5 on github.
colomon rakudo: say 'Y'...'z'
p6eval rakudo 272191: ( no output )
colomon rakudo: say ('Y'...'z').batch(10) 20:15
p6eval rakudo 272191: OUTPUT«YZAAABACADAEAFAGAH␤»
masak colomon: what I'd expect. and 'Y'...'z' never finishes.
TimToady colomon: do you think it makes sense to make an exception for 1 char on both ends
colomon TimToady: no idea.
tadzik Su-Shee: link? 20:16
colomon I was just trying to sort out what to do with Range and its tests
TimToady so that 'Y'...'z' will alwasy just march up codepoints and stop
it seems to be what the naive user expects
20:16 azert0x joined
Su-Shee tadzik: github.com/cosimo 20:16
20:16 pyrimidine left
TimToady since Perl is one of those languages that doesn't distinguish chars from strings much, when people write a single-character string, they're usually thinking 'char' 20:17
moritz_ in current Rakudo, can a module somehow access the lexical scope from which it was used? PIR magic is fine...
masak TimToady: I'm inclining every so subtly towards "yes".
TimToady I guess I'm asking if anyone sees a downside to forcing single chars to ignore the a..z ranges
masak s/every/ever/
moritz_ (I mean I'm fine if it's only possible with PIR magic)
20:17 moritz_ sets mode: +ooo masak colomon tadzik
masak \ooo/ 20:17
moritz_ :-)
masak look, a three-headed monkey! 20:18
20:18 Guest16581 left
masak s/>>/ behind you/ 20:18
TimToady I think I'll go ahead and spec the single-char exception then 20:19
moritz_ turns around
uh, I fell for the oldest trick ever!
masak moritz_: I seem to have used up my only time that trick works. :)
moritz_ :-)
tadzik github.com/cosimo/perl6-digest-md5/...est/MD5.pm
beautiful 20:20
moritz_ jnthn, pmichaud: I'm trying to get a mostly-working safe mode for rakudo again, but I don't know how to do it in a module...
I *could* add a WEIRDLY_NAMED_SUB to the setting, which removes offending symbol from the caller's lexical scope
20:20 xabbu42_ joined, xabbu42 left, xabbu42_ is now known as xabbu42
moritz_ would that work for you? 20:20
pugssvn r31269 | colomon++ | [t/spec] Refudge to work with list branch. 20:23
pmichaud moritz_: its fine with me. You could also stick a sub into another package. 20:24
moritz_ right; I just can't call it automatically from 'use'
pmichaud in some ways, Safe.pm just sounds like another setting :) 20:25
20:26 Su-Shee left
sorear run STD as a pre-verifier, since it refuses to parse Q:PIR, pir::, and panics on any unknown function 20:27
remove run/runinstead/open/etc from CORE.setting first 20:28
moritz_ sorear: that's a nice idea, but *very* slow
anyway, getting late here, and it seems Safe.pm isn't a LHF
will investigate more soon 20:29
bbl&
erm
bb tomorrow :-)
20:31 Guest16581 joined
sorear App::Persistant 20:31
the only slow part of STD is running the Perl5 parser over 60,000 lines of generated code
once it's running, it's a bit faster than the rakudo parser
20:33 eternaleye joined 20:34 xabbu42_ joined, xabbu42 left, xabbu42_ is now known as xabbu42
pugssvn r31270 | lwall++ | [S03] guarantee a monotonic function when single characters are used in a series 20:35
20:38 PZt joined
colomon TimToady: next question: is there any real reason to have RangeIter now, if $a .. $b always generates $a ... $b? 20:39
20:39 hercynium joined
colomon (Though hmmm, what about $a ..^ $b?) 20:39
pmichaud colomon: I thought about that also, and decided that the endpoints were a reason to switch
another reason is that we might need RangeIter.perl someday
and that's much easier to handle than an arbitrary gather/take block 20:40
colomon "easier to handle" is my #1 reason. 20:41
but it starts getting ugly if Range needs to duplicate a bunch of series code.
hmmm, maybe there's a clever refactor waiting there.
pmichaud: getting a Junction infinite recursion in S03-metaops/cross.t 20:42
pmichaud it's always possible that RangeIter is really SeriesIter
tadzik hmm, does MAIN params work in rakudo
?
pmichaud and that Range.iterator just creates a SeriesIter
colomon: okay, I'll look at the infi recursion 20:43
20:43 armagad joined, xabbu42_ joined
pmichaud I looked at it briefly last night but was too tired to do much about it, and now I've forgotten if I found anything (or what I found :) 20:43
20:43 xabbu42 left, xabbu42_ is now known as xabbu42
colomon Don't stress too much about it for my sake, just wanted to ping you on it. 20:44
pmichaud colomon: I'd be really really really happy if you could continue RangeIter from where I left off
I don't plan to tackle it much more myself
colomon pmichaud: that's what I'm trying to figure now. :)
it never was right before, and I thought I'd try to get it a bit better this time around.
pmichaud colomon++ 20:45
20:45 ash__ left
colomon and of course, the relevant rules have changed about ten times since I implemented RangeIter back in January. :) 20:46
pmichaud ten is a lot. 20:48
colomon you know how it is; having an implementation helps find problems in the spec.
;)
TimToady we do define ...^ somewhere 20:51
though not ^...
rindolf BTW, what is the :step/:by/etc. adverb to the range operator? How one should write it? 20:52
1,3...11
TimToady that works
rindolf rakudo: (1,3..11).say
p6eval rakudo 272191: OUTPUT«134567891011␤»
TimToady but only with 3 dots
rindolf rakudo: (1,3...11).say
p6eval rakudo 272191: OUTPUT«1357911␤»
TimToady there is no :by anymore for .. 20:53
you must use ... for a non-1 step
masak std: 1 ...^ 5
p6eval std 31269: OUTPUT«ok 00:01 108m␤»
TimToady ranges are just "intervals" these days
but we let you use an interval for a counting sequence 20:54
rindolf TimToady: we discussed it in szabgab's presentation.
colomon TimToady: really? ...^ is news to me.
masak to me as well.
why ... and ...^ but not the other two? 20:55
TimToady it's there at the end of the series operator sectoin 20:56
*io
20:56 Targhan_ joined, dual left
TimToady mainly because ...^ seems easier, and it's not clear what you're excluding on the front if you say 1,2,3 ^... 100 20:57
masak oh yes, there it is. S03:2086.
20:57 rv2733 joined, justatheory left
masak TimToady: thought so. sounds sane. 20:57
TimToady your line number is pre-last-checking 20:58
*in
S03:2107 now
masak ok.
does it still say "The compile\nmay complain"? 20:59
should there be an -r on "compile"?
21:00 tedv left, clintongormley left
TimToady either the compile or the compiler may complain :) 21:01
masak :)
just checking that I'm getting this right: S03:2048 says `'9'..'0'` -- is that an example of an empty range still being empty under the new one-char exception? 21:02
pugssvn r31271 | lwall++ | [S03] typos spotted by masak++ 21:06
masak having spotted today's typos, I withdraw to my dwelling for the evening. 21:07
see you tomorrow, #perl6.
TimToady \o
colomon o/
21:08 justatheory joined 21:09 masak left
colomon Mouth is starting to unnumb, and it hurts. :( 21:11
jnthn Aww.
21:11 foodoo left 21:13 tadzik left, whiteknight joined
colomon on the plus side, the largish hole in that one tooth is now filled in. 21:13
jnthn In the long run, I think the plus side will win out. :-) 21:14
TimToady if it doesn't you'll be non-plussed 21:15
21:17 dual joined
colomon I'm just hoping that bassoon playing doesn't hurt tonight. 21:17
21:18 bphillips left
pmichaud nopaste.snit.ch/21272 # lazy slurpies 21:25
colomon pmichaud: crazy! \o/ 21:27
rindolf Yes, "ש".chars() in the REPL yields 2. 21:28
TimToady I assume this all naturally fixes the two-gather bug
colomon does it work the same atm if you take make it non-slurpy?
TimToady or wait, that was the closure thing
pmichaud TimToady: yes, that's the closure issue. That comes next :)
colomon: yes, it works the same with non-slurpies
rindolf "é".chars() too. 21:29
TimToady eventually it will say 1
when we switch to really processing graphemes
pmichaud it's possible the repl is forgetting to set utf-8 encoding again.
and... what TimToady++ said
TimToady and yes, what pmichaud++ said, these should already have been doing 1 as codepoints 21:30
rakudo: say "ש".chars() 21:31
p6eval rakudo 272191: OUTPUT«1␤»
21:31 Mowah left, Trashlord left
TimToady rakudo: say "𠯭".chars 21:32
p6eval rakudo 272191: OUTPUT«1␤»
TimToady at least it doesn't fib about plane 1 chars
21:34 azert0x left, rindolf left
TimToady nap & 21:34
21:34 Ross left 21:48 jferrero joined 21:55 skids left
lue ohai [ooh unicode discussion!] 22:02
Tene pmichaud: you understand the closure issue and know how to fix it? 22:03
pmichaud Tene: yes. 22:06
it's already been fixed in nqp-rx.
Tene Ah.
pmichaud it's the same fix for rakudo, but the fact that rakudo wraps all of its codeblocks (subs) in rakudo-specific objects makes it trickier on the codegen side
because it's not sufficient to clone the codeblocks -- one has to have matching wrappers to go with them. 22:07
Tene nodnod
colomon > my @a = 1, 2, 3; @a.push: <a b c>; say @a.perl 22:10
[1, 2, 3, "a", "b", "c"]
that's list branch
rakudo: my @a = 1, 2, 3; @a.push: <a b c>; say @a.perl
p6eval rakudo 272191: OUTPUT«[1, 2, 3, "a", "b", "c"]␤»
colomon hmmm.
oh. 22:11
rakudo: my @a = 1, 2, 3; @a.push: [5, 4]; say @a.perl
p6eval rakudo 272191: OUTPUT«[1, 2, 3, [5, 4]]␤»
colomon > my @a = 1, 2, 3; @a.push: [5, 4]; say @a.perl
[1, 2, 3, 5, 4]
pmichaud ugh
colomon there we go, that's the bit that's giving me trouble.
pmichaud yeah, okay, I know what's wrong there.
actually, I don't. hrm.
lue I don't see it, unless it should be [1, 2, 3, [5, 4]] 22:12
pmichaud it should be [1,2,3,[5,4]]
22:13 Guest23195 left
pmichaud testing fix. 22:13
> my @a = 1,2,3; @a.push: [5,4]; say @a.perl 22:16
[1, 2, 3, [5, 4]]
better?
(I love fixes that make code shorter.)
22:17 eternaleye left
pmichaud ah, but it points to a different problem. 22:19
hrm.
22:20 dolmen joined
pmichaud tries a different fix. 22:22
22:25 Wolfman2000 left 22:26 Guest16581 left 22:27 rv2733 left
pmichaud and with this fix, S32-array/push.t now passes! 22:30
pushed as 1fe15ce 22:31
lue \o/
22:35 Util joined
Tene perl6: say (1,3 ... 9).perl 22:41
p6eval rakudo 272191: OUTPUT«(1, 3, 5, 7, 9)␤»
..pugs: OUTPUT«*** ␤ Unexpected "9"␤ expecting operator or ")"␤ at /tmp/SG0A6VTNd3 line 1, column 14␤»
..elf 31271: OUTPUT«Parse error in: /tmp/mWug_UqEtX␤panic at line 1 column 0 (pos 0): Can't understand next input--giving up␤WHERE: say (1,3 ... 9).perl␤WHERE:/\<-- HERE␤ STD_red/prelude.rb:99:in `panic'␤ STD_red/std.rb:76:in `scan_unitstopper'␤ STD_red/std.rb:224:in `comp_unit'␤
..STD_red/std.rb:21…
22:44 dolmen left
lue rakudo: my @a = en.wikipedia.org/wiki/Special:Searc...3],[4,5,6; say @a[(1,2)] 22:45
p6eval rakudo 272191: OUTPUT«===SORRY!===␤Confused at line 11, near "my @a = ht"␤»
lue rakudo: my @a = [[1,2,3], [4,5,6]]; say @a[(1,2)]
p6eval rakudo 272191: OUTPUT«␤»
lue rakudo: my @a = [[1,2,3], [4,5,6]]; say @a(1,2)
p6eval rakudo 272191: OUTPUT«invoke() not implemented in class 'Array'␤ in main program body at line 11:/tmp/a4AjVhQMMc␤»
lue rakudo: my @a = [[1,2,3], [4,5,6]]; say @a[1,2] 22:46
p6eval rakudo 272191: OUTPUT«␤»
lue rakudo: my @a = [[1,2,3], [4,5,6]]; say @a[0,1]
p6eval rakudo 272191: OUTPUT«1 2 3 4 5 6␤»
lue something isn't right here... 22:47
pugssvn r31272 | sorear++ | [dfa/Cursor] Small refactor of nfamanip routines to clarify item/list context 22:48
sorear that's called a slice.
you are probably looking for multidimensional lookup
lue yes I am. 22:49
sorear rakudo: my @a = [[1,2,3], [4,5,6]]; say @a[0;1]
p6eval rakudo 272191: OUTPUT«␤»
sorear rakudo: my @a = ([1,2,3], [4,5,6]); say @a[0;1]
p6eval rakudo 272191: OUTPUT«4 5 6␤»
sorear rakudo: my @a = [1,2,3], [4,5,6]; say @a[0;1]
p6eval rakudo 272191: OUTPUT«4 5 6␤»
sorear eh, it won't work until pmichaud++ makes lists stop sucking 22:50
lue alright, at least I know to code the semicolon :)
22:51 mjk joined
sorear pugs: my @a = [1,2,3], [4,5,6]; say @a[0;1] 22:52
p6eval pugs: OUTPUT«*** ␤ Unexpected ";"␤ expecting "x", octal digit, "o", "b", fraction, exponent, term postfix, operator or "]"␤ at /tmp/3UFMqf4EnX line 1, column 35␤»
sorear alpha: my @a = [1,2,3], [4,5,6]; say @a[0;1]
p6eval alpha 30e0ed: OUTPUT«Confused at line 10, near "[0;1]"␤in Main (file <unknown>, line <unknown>)␤»
sorear rakudo: my @a = (1,2,3), (4,5,6); say @a[0;1]
p6eval rakudo 272191: OUTPUT«2␤»
sorear oh good, *something* works
23:04 eternaleye joined
pugssvn r31273 | sorear++ | [dfa/Cursor] extract _nfa_to_dfa from _AUTOLEXgenDFA 23:11
sorear the automaton converter spans 6 functions and 230 lines... it's a lot bigger than I expected it would be 23:12
23:13 xabbu42_ joined, xabbu42 left, xabbu42_ is now known as xabbu42
pugssvn r31274 | sorear++ | [dfa/CursorBase.pmc] Per TimToady++'s recommendation, remove all usage of characters as 1-length strings from the DFA execution engine 23:14
TimToady std: 74 23:16
p6eval std 31272: OUTPUT«ok 00:01 107m␤»
diakopter std: r31272: ::OUTPUT«ok 00:01 107m␤» 23:18
p6eval std 31272: OUTPUT«ok 00:01 104m␤»
TimToady sorear: STD doesn't make from clean currently for me 23:19
oh, wait, might be unchecked in code 23:20
23:20 xabbu42_ joined, xabbu42 left, xabbu42_ is now known as xabbu42
TimToady yes, my check for dup protos caught a dup infix:<=> in the setting :) 23:21
23:24 cdarroch left
pugssvn r31275 | lwall++ | [STD] catch duplicates involving protos as well as onlys 23:26
r31275 | [CORE] remove duplicate infix:<=> found by above check :)
23:27 jferrero left 23:28 xabbu42_ joined, xabbu42 left, xabbu42_ is now known as xabbu42
pugssvn r31276 | lwall++ | [STD's lib/Text.pm6] remove dup proto 23:29
23:30 alester left 23:32 xabbu42 left
TimToady er, s/Text/Test/ 23:32
rakudo: say [][0-1].WHAT 23:34
p6eval rakudo 272191: OUTPUT«Cannot use negative index on arrays␤ in 'Array::postcircumfix:<[ ]>' at line 2612:CORE.setting␤ in main program body at line 11:/tmp/k8RkENnTx9␤»
TimToady um, that should be a fail, not a die
lue what is the difference between a fail & die? 23:35
TimToady rakudo: say [][0-1] //"Caught it" 23:36
p6eval rakudo 272191: OUTPUT«Cannot use negative index on arrays␤ in 'Array::postcircumfix:<[ ]>' at line 2612:CORE.setting␤ in main program body at line 11:/tmp/Qw7pqjFpEg␤»
TimToady that should say "Caught it"
23:36 armagad left
lucs Wow. Look at what pronik++ has produced: img267.imageshack.us/img267/2687/booka4.pdf 23:37
23:38 mikehh left 23:39 mikehh joined
TimToady we'd like to be able to say, for a given grapheme $g, if @alnum[$g] // extended_NFG_is_alnum($g) { say "It's alnum" } 23:39
23:39 pronik joined
TimToady assuming $g is a positive or negative integer 23:39
sorear what sort of object is @alnum?
a 256 element bit array?
TimToady bitmap
for important ones, 0x10ffff big 23:40
but can be done with inversion lists or whatever for less important ones
point is, the code is the same
rakudo: say 0x10ffff / 4 23:41
p6eval rakudo 272191: OUTPUT«278527.75␤»
sorear 4?
TimToady rakudo: say 0x10ffff / 8
p6eval rakudo 272191: OUTPUT«139263.875␤»
TimToady sorry, my finger slipped
so 140MB is the largest needed bitmap and most will be smaller 23:42
sorear 140 *k* B
140MB is half my RAM :(
why will most be smaller?
TimToady most people won't be using higher planes 23:44
sorear so the bitmap is lazily extended?
TimToady well, could be
sorear personally thinks the best way to handle the database is to represent it as a position-independant blob and let the OS demand-load it 23:45
23:45 gbacon left
TimToady or we just decide which bitmaps are worth 140k and which aren't 23:45
sorear: that may well be
that will tend to work well with locality, as the swashes do 23:46
makes it easy for different processes to share pages as well 23:47
assuming they're marked readonly
of course, this may not work out well for your typical phone OS... :) 23:49
lue ...which is why you use Meego :) 23:59
pugssvn r31277 | lwall++ | [CORE] add CallFrame and callframe()
diakopter ooo

Powered by Cro and the Raku® Programming Language.

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

Thank you!