»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.
sorear aha, found it 00:01
pugssvn r31008 | sorear++ | [viv] Allow .*? to backtrack in rules again. Oops. 00:09
00:16 whiteknight joined 00:21 Spreadsheet_ left 00:23 [Coke] joined 00:30 azert0x left 00:35 rgrau left 00:36 yinyin joined
[Coke] skips review. 00:38
00:38 kda left 01:00 meppl left 01:03 yinyin left 01:07 meppl joined 01:09 pausenclown left 01:13 pausenclown joined 01:16 tylerni7 left 01:31 snarkyboojum left, am0c joined 01:32 Guest86832 left, nothingmuch joined 01:34 tylerni7 joined 01:45 nothingmuch left 01:46 meppl left 02:08 synth left 02:09 synth joined
jnthn mornin' 02:19
02:23 whiteknight left
sorear yo. 02:27
02:28 [mark] joined 02:30 agentzh joined 02:31 atomicstack joined
sorear Hmm. 02:36
Fixing quantifier ratcheting doesn't seem to have helped performance much... 02:37
02:39 JimmyZ joined 02:43 bluescreen joined, bluescreen is now known as Guest3497 02:46 JimmyZ left 02:47 Guest3497 left, JimmyZ joined 02:48 pencilk joined 02:51 molaf joined
TimToady <masak> module A { module UNIT::A {} } # clash? 03:03
yes, clash 03:04
colomon jnthn: I'm spectesting Prakash Kailasa's hyper patch, but I'm going to go to bed before it finishes, I think.
TimToady std: module A { module UNIT::A {} }
p6eval std 31008: OUTPUT«===SORRY!===␤Illegal redeclaration of symbol 'MY:file</tmp/NdCFDuauHW>::<A>' (from line 1) at /tmp/NdCFDuauHW line 1:␤------> module A { module UNIT::A ⏏{} }␤Check failed␤FAILED 00:01 114m␤»
jnthn colomon: OK. :-) 03:05
colomon: I've gotta take care of $dayjob for the next little while anyways.
TimToady in fact, nearly all your questions about module nesting are already answered by STD, though you might have to parse a .pm and examine the resulting .syml file to find the answer
colomon It gets at least another 40 tests passing! \o/
jnthn colomon: \o/
colomon: We hit 33k yet? ;-) 03:06
colomon I dunno, but we're within 40 for sure, I think.
jnthn \o/
colomon currently having git issues. :(
jnthn Good chance of us making that in the next 24 hours then :-)
colomon yes 03:07
bother, I just erased the working patch. :()
:(
jnthn :( 03:08
colomon easy enough to rebuild, now trying compile and hyper.t again. 03:09
03:10 krakan joined
colomon ack, things are completely confused here. 03:10
think I will go to bed and tackle it again in the morning. 03:11
03:12 colomon left
TimToady the _reduced for <foo=.bar> is 'bar' in sTD 03:15
STD even
03:19 Eevee left
sorear TimToady: What are the most important optimizations in gimme5? 03:21
TimToady well, LTM itself prevents a great many false leads 03:22
the other biggie is avoiding lazymap
via ratcheting
sorear Huh. I had never thought of LTM as an /opt/imization 03:23
TimToady if you do the LTM at the termish level, you don't have to redo it at lower levels because the fate is already determined 03:24
sorear hmm. I wonder if that's working 03:25
(in viv)
TimToady should be able to tell from the logs what probes are made and what fates are piced 03:26
*picked
sorear on the plus side, LTM lexer caching does not appear to be magically broken 03:28
sorear has a hunch that <.ws> is the most expensive rule
how does HIGHEXPECT and MEMOS work? 03:29
TimToady which is why it is heavily optimized
you could run NYTProf on viv
I run it now and then on std 03:30
generally cursor_fate is the most expensive routine
since it's faking LTM
well, it's doing LTM but not always efficiently
@*MEMOS is indexed by pos in the original source 03:31
each entry is a hash that, beside mapping back to line number, allows any other keys, such as <ws> for the whitespace memoization
if <ws> is set, it's pointing back to the beginning of the whitespace here, and means "we already match ws here, don't do it again" 03:32
also we set some hints for possible future error messages
<nodecl> and such
sorear ponders experimenting with packratism
TimToady probably won't get much
because we don't backtrack much 03:33
HIGHEXPECT records the rules we attempted at the highwater mark 03:34
sorear gimme5: sub ws__PEEK { '' }
viv: sub ws__PEEK { '' }
er
TimToady it's the list of expectations in 'expecting any of' messages
sorear sub ws__PEEK { $_[0]->_AUTOLEXpeek('ws', $retree) }
aha, I was wondering how that worked 03:35
sorear thinks this __PEEK difference is important
03:35 cowyn left
TimToady I think ws is hardwired elsewhere to terminate LTM 03:35
so this PEEK shouldn't be important
RE_ast:214 03:36
218 rather
sorear in general, though
viv is currently hardwired to generate a single form of __PEEK
sorear doesn't follow the <<DECL>> etc stuff in gimme5 and just cloned behaviorally 03:37
ah, viv generates empty __PEEK for ws fail commit before after panic and anything declared using 'method' 03:39
s/viv/gimme5/
viv doesn't generate __PEEK *at all* in those cases
I'm also not generating EXPR__PEEK 03:40
hmm 03:52
die processing is pretty slow iirc 03:53
maybe adding __PEEK stubs for things would be a big improvement
except, no, that would only help autolexer generation
which is Not The Bottleneck
03:55 cli_ joined
sorear TimToady: Might changing the names of alternatives have broken anything? (ws_03 and ws_05 are now ws_0 and ws_1) 03:56
sorear quickly flips back and forth between two versions of "sub ws" looking for breakage 03:57
lue ohai 04:00
sorear TimToady: Can you tell me anything about the hacks involved in autolexing EXPR? 04:04
dmpk2k Does Perl6 have any method or function help at the REPL, like Python or Common Lisp? 04:09
sorear rakudo: Any.^methods.Str.say 04:11
p6eval rakudo dd0e5d: OUTPUT«can Numeric elems end reduce Str keys reverse isa uniq classify map pairs kv ACCEPTS min max pick first minmax does grep values join WHICH perl CREATE Capture PARROT bless WHENCE WHERE list notdef BUILDALL new Bool print say defined WALK item BUILD REJECTS clone␤»
sorear something like that?
dmpk2k Alas, nope. Fire up a python interpreter and type "help(dir)". 04:13
It shows you a description of what the dir() function does.
lue there's perldoc, but i'm not sure if that's part of the REPL itself. 04:14
04:15 cli_ left, cli_ joined
dmpk2k Perldoc is good stuff, but it'd very oh-so-nice if there were allowanced for it at the REPL. 04:16
I don't use Python, but I use it all the time with Factor.
lue rakudo: perldoc
p6eval rakudo dd0e5d: OUTPUT«Could not find sub &perldoc␤ in main program body at line 1:/tmp/TrJadigSqD␤»
sorear Perl6 is specced to have online help 04:19
dmpk2k Awesome. :)
sorear it's stored in the .WHY interrogative of variables
lue but last I heard, REPL still has an feeble "memory", in that it can't remember variables from the last line. It still has a waaays to go. 04:20
sorear there's a fork of the REPL from about two weeks ago with the ability to remember variables 04:21
pmichaud is working on a "real" version
04:22 molaf left
dmpk2k rakudo: say.WHY.say 04:24
p6eval rakudo dd0e5d: OUTPUT«␤Method 'WHY' not found for invocant of class 'Bool'␤ in main program body at line 11:/tmp/yFmMBZvYyP␤»
dmpk2k So much for that guess. 04:25
sorear there are no complete implementations of Perl 6 yet 04:26
pugssvn r31009 | sorear++ | [viv] Optimize .*? and .*! to SCANs with special meta nodes 04:28
r31009 | Blocks don't need additional cutting
lue rakudo: 3.WHO.say; 3.WHAT.say 04:30
p6eval rakudo dd0e5d: OUTPUT«Method 'say' not found for invocant of class 'Int'␤ in main program body at line 11:/tmp/eKMULuByqJ␤»
lue :O
that should work, right? 04:31
rakudo: say 3.WHO; say 3.WHAT
p6eval rakudo dd0e5d: OUTPUT«Int␤Int()␤»
04:40 am0c left
sorear oh, /nice/ 04:49
04:49 TiMBuS joined
sorear TimToady: apparently the .*? -> _SCANf optimization is Really Important 04:50
implementing it in viv made STD.pmc 36 times faster
04:56 lue left 05:04 PZt left 05:20 am0c joined
sorear looks through a 196MB viv.log 05:22
05:29 pencilk left
sorear STD-on-viv-on-STD can't seem to handle character classes 05:39
hmm
05:39 kaare joined
sorear /<[]>/ fails, tr/// succeeds 05:39
05:40 kaare is now known as Guest26513 05:42 dju joined
moritz_ good morning 05:49
PerlJam morning moritz_ 05:52
sorear good morning moritz_ 05:53
ok, I'm missing _PARAMS generation
moritz_ is very pleased with rakudo's progress in last 7 days or so 05:54
sorear hah, viv-compiled-STD doesn't grok balanced punctuation 05:56
pugssvn r31010 | sorear++ | [viv] Fix parameterized role autolexer generation (_PARAMS missing)
sorear it's willing to parse a character class if I spell it <[ a e i o u [>
dalek kudo: 5c09771 | moritz++ | t/spectest.data:
run whatever.t again
moritz_ sorear: :-)
pugssvn r31011 | moritz++ | [t/spec] test that method not found error contains method name and line number 05:59
05:59 mempko joined
sorear wanted: a web changeset service for pugssvn 06:00
06:00 am0c left 06:01 mempko left
sorear found it 06:05
moritz_ sorear: if you look at the IRC logs in /topic, all the r\d+ occurences are turned into links 06:06
as are many other things :-) 06:07
sorear excellent. moritz++ for telling me moritz++ for the logs 06:08
moritz_ #perl6++ 06:09
most feature ideas and layout patches came here
it was the first ugly-but-useful page I produced, and #perl6 people beautified it 06:10
same with perl6.org (perl6-proejects.org back then)
and same with proto.perl6.org
sorear wonders if #perl6 made mowyw 06:11
06:12 uniejo joined
moritz_ nope that was my own, pre-#perl6 invention 06:12
probably not really pre-#perl6, but shortly after I joined 06:13
pugssvn r31012 | sorear++ | [viv] Bracketize \xABCD in perl5-output
sorear yay, STD-viv can now parse bracketing delimiters
moritz_ \o/
sorear one little bugfix at a time 06:14
moritz_ rakudo: sub foo ($a where 5) { say "hi" }; foo(42);
p6eval rakudo dd0e5d: OUTPUT«Constraint type check failed for parameter '$a'␤ in 'foo' at line 11:/tmp/ld2jH1UvGy␤ in main program body at line 11:/tmp/ld2jH1UvGy␤»
sorear though I was rather in a quagmire not long ago; I had my ETA calculated as June 23, 2042
moritz_ ETA for viv compiling STD.pm6 without errors? 06:15
sorear no
06:15 atomicstack left
sorear that's already here 06:15
however, it's miscompiling STD 06:16
miscompilations are very problematic to debug
moritz_ with "without errors" I actually meant "correctly"
sorear ahhh 06:17
yeah. slow going.
moritz_ rakudo: my $x = "test"; given $x { say $_; m/(e.)/; say $/ } 06:18
p6eval rakudo dd0e5d: OUTPUT«test␤Any()␤»
sorear erm...
what does %<a> mean
moritz_ std: %<a> 06:19
p6eval std 31010: OUTPUT«ok 00:01 114m␤»
moritz_ std: % <a>
p6eval std 31010: OUTPUT«===SORRY!===␤Non-declarative sigil is missing its name at /tmp/NT9FIN0XhR line 1:␤------> <BOL>⏏% <a>␤Bogus statement at /tmp/NT9FIN0XhR line 1:␤------> %⏏ <a>␤ expecting twigil␤Parse failed␤FAILED 00:01 111m␤»
moritz_ has no idea
rakudo: class Foo { }; my $x = "Foo"; my $y = $x.new; $y.WHAT.say; 06:20
p6eval rakudo dd0e5d: OUTPUT«Foo()␤»
06:20 masonkramer joined
sorear STD.pm6 lines 3263-3265 contain 3 references to %<O><prec> syntax 06:20
gimme5 compiles this identically to $<O><prec>
moritz_ std: 1 R~= 2 06:21
p6eval std 31010: OUTPUT«ok 00:01 111m␤»
moritz_ rakudo: my $x = 3; 1 R~= $x; say $x
p6eval rakudo dd0e5d: OUTPUT«1␤» 06:22
moritz_ rakudo: sub foo { my $a = 5; say $^a }; foo(7)
p6eval rakudo dd0e5d: OUTPUT«===SORRY!===␤Multiple declarations of lexical '$a'␤␤»
sorear \o/ I'm 100 lines down 06:24
pugssvn r31013 | sorear++ | [viv] Translate the %<O> syntax found in STD
sorear that cuts about a year off my ETA at previous rate estimates :p
moritz_ somebody flagged rt.perl.org/rt3/Ticket/Display.html?id=62622 as [LHF] (pmichaud perhaps?) but I have no idea where to dig for it :( 06:25
sorear LHF? 06:26
moritz_ low-hanging fruit 06:27
06:31 masonkramer left 06:32 meteorjay joined 06:33 cognominal joined, tedv| left 06:35 Su-Shee joined
pugssvn r31014 | sorear++ | [viv] « » needs to generate an Array 06:41
sorear svn-- 06:42
moritz_ it's pugssvn that has problems with UTF-8 06:43
06:48 kda joined
sorear strange. now it's parsing :=, **, ++ fine, but not => 06:51
aha 06:55
it's parsing :sym« => » as having literal spaces in the name
screws up LTM
06:58 ejs joined
pugssvn r31015 | sorear++ | [viv] We need to trim leading and trailing spaces in word quotes used as adverbs 06:58
sorear jumped ahead *500* lines 06:59
3 years!!!
07:02 ejs left 07:03 ejs joined 07:09 atomicstack joined
Su-Shee good morning. 07:13
cognominal \o
07:14 cli_ left
sorear wishes he understood the contents of viv.log 07:18
mathw Morning 07:25
07:31 tomka joined
sorear TimToady: Why is $CTX only updated at certain places, instead of continuously in deb? 07:32
07:48 buubot left, buubot joined 07:56 cosimo left
pugssvn r31016 | sorear++ | [viv-ecosystem] Add a tiny script to make comparing lexers easier 07:57
08:01 plobsing left
pugssvn r31017 | sorear++ | [cleanlex] Also deal with FATE index skew. 08:04
08:08 proller joined
pugssvn r31018 | sorear++ | [viv] One hack and two bug-compatibilities to improve LTM 08:20
sorear it can now parse $<a> 08:21
there's something delightfully ironic about parsefailing in the parse error code. 08:22
(line 1307)
mberends heh 08:23
sorear revises his ETA from June 26, 2042 to next Thursday 08:24
mberends that's better than "to *the* next Thursday" ;) 08:25
sorear std: no Foo
p6eval std 31016: OUTPUT«Can't locate object method "explain_myster" via package "STD::P6" at STD.pm line 12432.␤FAILED 00:01 113m␤»
pugssvn r31019 | sorear++ | [STD] Fix a misspelled method name 08:26
mberends tried to access SQLite via zavolaj yesterday. It required some signatures that Parrot doesn't yet support. In other experiments, it *did* create an empty database file before segfaulting. Committed the WIP anyway.
sorear mberends: go talk to plobsing 08:27
mberends yes, will do that :)
sorear plobsing's parrot can handle any signature
mberends oh cool!
moritz_ sorear: local branch?
sorear er, wait 08:33
I forgot... it was an extension 08:34
08:34 kel_ joined
sorear github.com/plobsing/parrot-libjit-fb 08:34
08:35 Ross joined 08:37 cosimo joined 08:40 Ross left
pugssvn r31020 | sorear++ | [viv] Global protoregex trait tables simulate inheritance 08:40
08:40 Ross joined
pugssvn r31021 | sorear++ | [viv] Ignore specific token signature if a proto token provided one, to prevent double shifting and lossage. (Is this the right approach?) 08:46
sorear line 2438! 08:47
09:14 tomka left 09:29 PZt joined 09:30 Ross left
pugssvn r31022 | sorear++ | [viv] Add SUBSUME coalescing (acts like an optimization, but it turns out to be semantically critical) 09:33
sorear can now handle 1 !~~ 1 09:35
moritz_ hopes it returns False :-)
sorear no, it returns (3 pages of AST) 09:36
moritz_ ah well
sorear you seem to have STD confused with a perl 6 implementation 09:37
STD.pm, having been compiled with STD.pm, just successfully parsed STD.pm 09:39
\
o
/
that actually wasn't intentional entering... 09:40
moritz_ looks like a cool move :-) 09:41
now it would be interesting if also rejects wrong Perl 6 programs :-)
sorear: how long did it take (the parsing, that is)?
sorear out of scrollback, but 4-5 minutes 09:42
the same as gimme5 STD
09:43 Ross joined
sorear STD stage 3 contains several <<ERROR>> markers, indicating that unprocessable (i.e. different and wrong) AST is still being generated in some cases 09:45
however, diff -u (stage 3) (stage 2) is 10,000 lines
compared to a total of 60k lines
09:46 azert0x joined
moritz_ too much if I understand you 09:46
sorear I mean to say that STD-stage-2 has correctly compiled 80% of STD
and successfully compiled 100% 09:47
moritz_ ok
sorear -> sleep 09:48
moritz_ 'night
Su-Shee I have a sudden hunger for bacon. 09:49
moritz_ chunky bacon? 09:50
Su-Shee slices or chunks, but fried please. :)
09:57 agentzh left
Su-Shee *HAHA* 09:57
my co-worker is from poland. _he_ got the name of zavolaj immediately. ;) 09:58
09:59 masak joined
masak oh hai, #perl6 09:59
phenny masak: 30 May 19:03Z <sorear> tell masak : there is no 'package' statement in Perl 6. "package Foo;" is interpreted as meaning "use v5; package Foo;" to allow pure-Perl-5 modules to be used unaltered. STD implements this (partially)
masak: 30 May 21:58Z <mathw> tell masak I pushed some changes to Form which make it build on my system, but the tests fail. The first failure is in the parsing tests with some fields not parsing, so maybe Form::Grammar is now out of date.
masak clearly activity kept being high after I left yesterday :) 10:01
mathw: given that there's a new underlying grammar engine now, that sounds probable :) 10:02
moritz_ mathw: biggest pitfal with the current grammar engine: backslash escape sequences in char classes are NYI 10:03
masak sorear: if by 'package statement' you mean 'braceless package declaration', then I agree. I can see how it would be right for STD.pm6 to actually allow such a form, if it did enough p5. Rakudo, in its present form, should probably disallow it. 10:04
moritz_: I could probably implement those. but not before the Yapsi release. :)
moritz_ masak: in nqp-rx? that would be awesome 10:05
masak I've done it before in Perl 6, so... :)
moritz_ it's currently the biggest source of workarounds in JSON::Tiny
it also rates high in the often-discussed bugs on p6c 10:07
jnthn masak: We can make package an error, or a "is this Perl 5" kinda message, quite easily if that's the Right Thing to do. 10:10
mberends std: package Foo; 10:12
p6eval std 31022: OUTPUT«ok 00:01 114m␤»
masak alpha: say ?(chr(9) ~~ /<[\t]>/) # is this meant to work? 10:14
p6eval alpha 30e0ed: OUTPUT«1␤»
masak oh, it does :)
rakudo: say ?(chr(9) ~~ /<[\t]>/)
p6eval rakudo 5c0977: OUTPUT«0␤»
moritz_ rakudo: say ord("\t")
p6eval rakudo 5c0977: OUTPUT«9␤»
moritz_ rakudo: say ?('\\' ~ /<[\t]>/) 10:15
p6eval rakudo 5c0977: OUTPUT«1␤»
moritz_ rakudo: say ?('t' ~ /<[\t]>/)
p6eval rakudo 5c0977: OUTPUT«1␤»
masak notices with embarrassment that GGE doesn't implement this at all
mathw moritz_: no char classes in Form's grammar 10:16
masak: Yes. What I need is a grammar engine that tells me what it's doing :)
masak gnnnn. I really really need to write that GGE debugger. 10:17
mathw heh
masak maybe I should convince someone to help me do it? it's really LHF at this point.
mathw I'll probably take another look at it later
see if I can figure it out
right now I need to have some breakfast
moritz_ rakudo: '12 abc ' ~~ /<?DEBUG(1)> <alpha>+/ 10:18
p6eval rakudo 5c0977: OUTPUT«3/1: PASS at pos=6␤»
moritz_ not perfect, but better than nothing 10:19
mathw could be just the key
okay I'll look at it now
and then I'll have breakfast
okay now I'm confused 10:22
the same field that's failing... passes if I run just that
but doesn't run inside the test file itself 10:23
so maybe it's not the grammar
10:24 whiteknight joined 10:25 colomon joined
mathw oh 10:26
no my mistake
it is the parsing
bbkr rakudo: my Rat $a; $a += 0.1 for ^10 # chcking if 74626 is fixed 10:28
p6eval rakudo 5c0977: ( no output )
10:28 ejs left
masak well, it's resolved... 10:28
mathw rakudo: grammar A { regex B { a } }; "a" ~~ A::B; 10:29
p6eval rakudo 5c0977: OUTPUT«Can not find sub A::B␤ in main program body at line 1␤»
mathw pah
masak mathw: that's a bug. it was submitted yesterday.
mathw it's enormously unhelpful 10:30
masak mathw: workaround: rename the regex A::B
jnthn er
huh
How is that a bug?
regexes are has scoped by default 10:31
masak oh.
of course.
jnthn They're methods.
masak need to learn to think this new way.
mathw rakudo: grammar A { regex B { a } }; "a" ~~ A.B;
p6eval rakudo 5c0977: OUTPUT«Type objects are abstract and have no attributes, but you tried to access $!pos␤ in 'A::B' at line 11:/tmp/CT2pH7kqHL␤ in main program body at line 11:/tmp/CT2pH7kqHL␤»
masak mathw: that's calling the regex.
not referring to it.
jnthn /<A::B>/
Or our regex and than &A::B 10:32
masak rakudo: grammar A { regex B { a } }; A.B
10:32 tylerni7 left
p6eval rakudo 5c0977: OUTPUT«Type objects are abstract and have no attributes, but you tried to access $!pos␤ in 'A::B' at line 11:/tmp/ydvRI0xwHf␤ in main program body at line 11:/tmp/ydvRI0xwHf␤» 10:32
masak slightly unhelpful error message there.
mathw rakudo: grammar A { regex B { a } }; "a" ~~ /<A::B>/;
p6eval rakudo 5c0977: OUTPUT«===SORRY!===␤regex assertion not terminated by angle bracket at line 11, near "::B>/;"␤»
masak would it be possible to get something like "oh, you tried to call me as a method, don't do that"
?
10:33 ejs joined
masak mathw: I don't think /<A::B>/ is implemented yet... :( 10:33
colomon jnthn: seem to have a solid patch this morning....
jnthn masak: gah, now *that* is annoying (<A::B> NYI)
mathw any chance it's LHF?
masak jnthn: glad we agree :)
jnthn mathw: Well, probably if you're pmichaud 10:34
mathw: Maybe even if you're not. :-)
colomon: yay!
mathw because figuring out this Form thing without being able to call individual regexes in the grammar for testing is going to be a pain in the arse
it's going to be a pain anyway, but...
10:34 BrowserUk joined
jnthn masak: On the error, not quite sure what the best way to be more helpful would be. 10:35
masak: Perhaps though if it emits a signature with the invocant as :D when the binder supports that...
...then at least the dispatch fails if you try and call the regex on the type object.
masak hm, yes.
but until then, maybe we could have a check inside the regex method? 10:36
jnthn moze byt.
masak likes how 'may be' and 'moze byt' are built from the same components 10:37
10:37 atomicstack left, zamolxes left
jnthn masak: Well, it's more literally "can be" 10:37
masak and 'may' means...? :) 10:38
it comes from a proto-Germanic root *mag-, which indicates ability. 10:40
jnthn heh
Nice
Perhaps that's proto-indo european; Russian "I can" is могу for example. :-) 10:41
masak from proto-Indo-European *mogh/*megh meaning 'power', from Sanskrit mahan, 'great'. it's related to 'might'.
yes, they're probably related.
I never thought of that. 10:42
jnthn Cute.
Su-Shee "to understand the meaning of perl 6 modules, please refer to Linguistics 0.1 for beginners and consult a Sanskrit grammar of your choice. Knowledge in attic greek and japanese is helpful." 10:43
10:43 Lorn joined
masak even the etymology for the month May flows out of that same river. 10:43
jnthn Grr. If validating XML, helps to actually load the document you're planning to validate into the reader before calling .Validate().
On the upside, the first XML schema I ever wrote in my life...works straight off.
Su-Shee masak: I thought may comes from mars actually? 10:44
10:44 JimmyZ left
jnthn It's "Mars" the word for May in...some romance language? 10:44
Su-Shee yes. 10:45
jnthn is very rusty on romance
*romance languages
:)
masak Su-Shee: no, it seems to come from the greek goddess Maja/Maia, whose name has to do with might and power. 10:46
Su-Shee no, I'm wrong and confusing it with march ;)
just looked it up. :)
jnthn Oh!
:-)
Su-Shee jnthn: *HAHAHAHA* :)
masak Swedish for Match is 'mars' :)
jnthn masak: Match or March? :-) 10:47
masak er.
my typo disaster seems to not be limited to yesterday. :(
jnthn Jag har en Marsbjekt!
masak March. of course.
jnthn: 'ett objekt' :)
jnthn :-) 10:48
masak: Gradually adding words to my "these take ett" list. :-)
masak should probably have a list of 'these take en' instead... :)
it will be shorter. 10:49
jnthn masak: Huh? Aren't there a lot more common gender ones?
masak let me check.
jnthn I thought there were way more of those. :)
masak yep. you're right. common gender is around 75%. 10:50
jnthn Wow, even higher than I'd thought. 10:51
(I had thought maybe 2/3-ish)
masak according to Wikipedia.
and there's a handy rule for detecting the 'ett'-form nouns.
jnthn (IIUC, masculine and feminine genders merged to common a while back.)
A rule? :-D
masak if the plural indefinite form is the same as the singular indefinite form, it's a t-word.
'ett bord' -- 'flera bord'. t-word. 10:52
the converse of the rule doesn't hold.
jnthn Yeah. :-(
So unless you go learning the plural indefinite forms of everything first...that only helps so much.
masak sorry :/ 10:53
jnthn Nice to know though.
Though...it sounds familiar.
masak I've probably said it before.
jnthn I think I may have run into it and then discarded it as "not much use" :-)
I'm very much liking the "verbs in the present tense have just one form" thing though. 10:54
Rather than, like, six. :-)
Su-Shee you are clearly not talking about german. ;)
masak jnthn: there are two if you read slightly older texts.
but even that is pretty ok.
jnthn masak: Ah, OK. I haven't run accross that. 10:55
masak: I have run accross remenants of masc/fem though. 10:56
lillebror vs lillasyster
mberends Dutch is also a bit gender-confused. But I think French takes the cake for "un livre" and "une livre" meaning completely different things
masak jnthn: aye. and 'gamle'/'gamla', and 'käre'/'kära'. 10:57
mberends: that's not confusion, that's efficiency :P 10:58
jnthn masak: I love the whole grandparents thing though. 10:59
Su-Shee dutch is nice with their like three or four irregular verbs ;)
jnthn masak: mormor/morfar/farmor/farfar is so nice and logical.
masak aye. wouldn't have it any other way.
jnthn In English its ambiguous.
barnbarn is cute too :-)
masak jnthn: there's farbror and mo(sy)ster, too. 11:00
er, mo(rsy)ster.
jnthn Oh, wow.
masak lunch &
jnthn swedish++
11:02 ejs left
colomon jnthn: 32,983 passing tests, one fail. 11:03
rakudo: <a b> X, <1 2> 11:04
p6eval rakudo 5c0977: ( no output )
colomon rakudo: (<a b> X, <1 2>).perl.say
jnthn colomon: oh noes a fel!
p6eval rakudo 5c0977: OUTPUT«("a", "1", "a", "2", "b", "1", "b", "2")␤»
jnthn colomon: What fails?
colomon [~]«( <a b> X, <1 2> ) 11:05
I kind of think the test may be wrong.
jnthn Ah. :-)
colomon It expects the result to be a1 a2 b1 b2
11:07 ejs joined
colomon that's assuming <a b> X, <1 2> produces something like (a, 1), (a, 2), (b, 1), (b 2) 11:07
and you can then run [~] on each of those pairs.
jnthn Ah 11:08
Well, it probably should produce something like that, but then the [~] implies flattening list context.
I guess you'd need to explicitly stick the parcel into slice context (which is NYI in Rakudo) for that to work. 11:09
colomon but "dwimmy hyper operator works recursively for any object matching the Iterable role"
and (a, 1) is Iterable, isn't it?
jnthn I hope so. :-)
colomon I'm just going to fudge that one. and maybe TimToady can wander by and tell us if it correct or not. 11:11
jnthn Good plan. :-)
jnthn is still a tad hazy on all that area
colomon well, and you can get the same effect by doing X~ and skipping the hyper bit altogether. 11:12
11:12 ejs left
dalek kudo: aebd0c5 | (Solomon Foster)++ | src/core/metaops.pm:
Extend hyper to handle nested lists / arrays. Patch courtesy of Prakash
11:14
pugssvn r31023 | colomon++ | [t/spec] Refudging and tweaks for nested hyper. 11:15
11:17 ejs joined
colomon now to have a bit of noms and then try to knock off 20 more tests... 11:19
11:19 Ross left, Ross joined
jnthn colonom++ 11:20
.oO( noms sound like a good idea... )
11:22 zamolxes joined 11:42 azert0x left 11:43 alanhaggai joined
masak phenny: tell lue that you asked why trigonometry entered into the complex version of &exp. there's an exciting field called 'complex analysis' which holds the exact answers, but the short story is that 'it all fits'. :) the closeness between exp and sin/cos can be seen through their Taylor series. 11:45
phenny masak: I'll pass that on when lue is around.
moritz_ or the short explanation: exp(i phi) = cos(phi) + i sin(phi) 11:46
masak yes, but how to explain *that* without pulling out of your hat? 11:47
the explanation for that is that the (only possible) analytic extension of the real part of exp becomes just that. 11:48
which is slightly unstatisfactory until you learn the math to prove it.
moritz_ right :-)
I think I forgot that part of my math already :-)
11:48 DJ-DONGDOT joined
moritz_ I dimly remember the results, though 11:49
11:50 DJ-DONGDOT left
moritz_ but yeah, taylor expansion is a good way to show it 11:50
masak complex analysis ranks among the most beatiful subfields of math I've studied. it's fantastic, especially given that complex numbers are relatively non-obvious, yet almost as well-behvaed as reals.
it's like a cave of riches set up to reward curious mathematicians. :) 11:51
colomon has to admit he preferred real analysis.
11:51 envi^home joined
masak rakudo: say 3 «*» (1,2,3,4) 11:52
p6eval rakudo 5c0977: OUTPUT«36912␤»
masak why does that not give an error?
colomon why would it?
moritz_ never really got to chose, given that he usually picked subjects by physical interest, not mathematical
masak colomon: because the 'lists' are of different shape.
oh wait, the arrows are pointing outwards... :) 11:53
moritz_ rakudo: say 3 >>*<< (1, 2, 3, 4)
p6eval rakudo 5c0977: OUTPUT«Sorry, left side is too short and not dwimmy.␤ in 'hyper' at line 137:CORE.setting␤ in 'hyper' at line 150:CORE.setting␤ in main program body at line 1␤»
masak nice, error. it refers to two different things by 'left side' though :P
11:53 colomon_phone joined
masak s/,// 11:54
colomon_phone lost the last couple of lines there...
moritz_ colomon_phone: masak confused >>*<< with <<*>> 11:55
masak aye. :/
colomon_phone gotcha. :) 11:56
11:57 colomon left
jnthn Aye, that error could be slightly clearer, methinks. 11:57
colomon_phone I was just working on hyper on hashes, but ran out of battery with a couple of wee issues left.
11:58 azert0x joined
moritz_ wonders if working on Str.trans would make a good Weekly Challenge 12:01
colomon_phone I forgot to check aNd see if hyperhaSh wAs in alphA....
12:01 proller left
moritz_ rakudo: say {1 => 2, 3 => 4} >>*<< 3 12:02
masak moritz_: it would sure be a nice (re-)addition to Rakudo.
p6eval rakudo 5c0977: OUTPUT«Sorry, right side is too short and not dwimmy.␤ in 'hyper' at line 127:CORE.setting␤ in 'hyper' at line 150:CORE.setting␤ in main program body at line 1␤»
moritz_ alpha: say {1 => 2, 3 => 4} >>*<< 3
p6eval alpha 30e0ed: ( no output )
moritz_ masak: that much is for sure; other questions I have to ask myself is whether it's doable for a novice, or which parts of it, and how much to request 12:03
takadonet morning all 12:04
colomon_phone o/ 12:05
12:05 clintongormley joined 12:07 colomon_phone left 12:08 ejs left
masak moritz_: has any task proven too difficult yet, in the sense that no-one has attempted it? 12:14
moritz_ masak: nope
masak I must say I'm pleasantly surprised seeing *new* people contribute things each week. 12:16
moritz_ huh. Our federal president just stepped down. That's surprising. (off-topic, sorry)
masak that indicates that we should probably spread out as much as possible, task-wise, so as to tickle as many people as possible at least once.
Su-Shee moritz_: uhm what? *click* ;) 12:24
moritz_ Su-Shee: like www.tagesschau.de/inland/koehlerrue...tt100.html 12:25
npq: $_ := 3; say(.uc) # I know this blows up, just want to know at which point 12:26
nqp: $_ := 3; say(.uc) # I know this blows up, just want to know at which point 12:27
p6eval nqp: OUTPUT«Confused at line 1, near "say(.uc) #"␤current instr.: 'parrot;HLL;Grammar;panic' pc 528 (src/cheats/hll-grammar.pir:196)␤»
Su-Shee moritz_: already like 5 tabs open with news ;)
12:28 kda left 12:29 colomon joined
Su-Shee tststs. 12:31
12:31 JimmyZ joined 12:42 proller joined
masak what would be a fitting logotype for Yapsi? 12:59
jnthn Some kind of 6 looping back on itself somehow? 13:00
moritz_ a disc world standing on top of elephants on top of a giant turtle
masak all 6-es loop back on themselves :P
jnthn (since it's Perl 6 implemented in Perl 6)
moritz_ "it's turtle all the way down"
colomon shouldn't be a giant parrot?
jnthn masak: Well, you could think of a loopier one :-)
masak a turtle is kinda nice.
moritz_ jnthn: a mobius strip might symbol that very well 13:01
colomon camels on a parrot...
jnthn Ooh, yes.
colomon spectesting hyper hash patch now.
jnthn I think for a logotype, the camel would kinda dwarf the Parrot, if you did relative sizes. ;-)
masak likes the Möbius strip idea
jnthn: and the butterfly would be bigger than both :P
Juerd Möbius-6? 13:03
masak I can see how one could perhaps make a Klein bottle look like a 6...
a Möbius strip kinda looks like a 0 or an 8 no matter how you bend it. 13:04
jnthn Well, it'd need to be small for a logotype I guess...
;-)
Su-Shee please no animal.. I beg you.
masak :)
Su-Shee I'm so tired of all the cute cutesy open source cuddly thingies.. 13:05
masak Su-Shee: we could have a non-cuddly animal.
jnthn Dugong
Su-Shee no. rejected by commitee. :) moebius stripe is good!
jnthn Grizzly bear
Su-Shee naked molerat.
masak Velociraptor. 13:06
13:07 atomicstack joined
masak though that's more a Perl 5 symbol ;) 13:07
Su-Shee that's dino park and will trouble a) mozilla, b) disney and c) Warner Bros. ;)
masak dang. :P 13:08
Su-Shee design something escher like, afaik there's a couble of mathematically highly complex islamic ornaments.
masak yeah, because we want to trouble Islam rather than Disney. :P
Su-Shee the use of an ornament is perfectly fine. 13:09
was just an example islamic art has plenty of pretty ornaments.
moritz_ it's funny, when you read parrot-dev you'll see that allison and I proposed very similar syntaxes for matching PAST nodes, only did she express it pythony and I perly :-) 13:10
13:10 ejs joined
moritz_ and basically at the same time 13:10
(I saw here mail come when I finished writing mine, but didn't read it before sending)
masak multiculturalism++ # both re Islamic art and re Perl/Python
13:11 Guest26513 left
jnthn moritz_: It occured to me that Perl 6 signature syntax is perhaps powerful enough for that. 13:12
moritz_ woooh.
13:12 Kejohm joined
jnthn PAST::Op $node (:$returns where 'Pair', *%) 13:13
13:13 Kejohm left
moritz_ it gets clumsy when it comes to repetition though 13:14
jnthn Repetition? 13:15
moritz_ "5 child noeds that all return 'Int'"
masak .oO( junctions ) 13:16
moritz_ *nodes
masak: so how do you express the 5 as a junction?
masak moritz_: I don't. I measure the length of the list of children.
moritz_ no problem with where { ... } blocks. But that's not declarative anymore 13:17
13:18 [mark] left
colomon 33,019 passing tests! 13:28
moritz_ rakudo: say 2**15
p6eval rakudo aebd0c: OUTPUT«32768␤»
masak \o/
pugssvn r31024 | colomon++ | [t/spec] Unfudge the hyper hash tests.
moritz_ hugme: tweet rakudoperl Rakudo now passes more than 2**15 tests! #perl6
hugme hugs moritz_; tweet delivered
masak it would be interesting to research when Rakudo passed the other 2 ** $n thresholds... 13:29
and then blog about it.
colomon moritz_: we've actually been above 2**15 for a few days now. :)
moritz_ masak: should be a SMOP, given docs/spectest-progress.csv
masak oh! indeed.
moritz_ colomon: I know, but nobody twittered it so far :-) 13:30
colomon moritz_++
best part is, there are obvious further hyper hash tests needed.
and a couple of obvious extensions to the current functionality which will be both super-cool and need even more tests. :)
dalek kudo: 7a2ede9 | (Solomon Foster)++ | src/core/metaops.pm:
Simple implementation of hyper on hashes.
 13:32
Wolfman2000 Over 65K tests...man, that's a lot.
...wait, I added an extra exponent
masak we passed 2**8 on 2008-05-25, 2**9 on 2008-06-01, 2**10 on 2008-06-27, 2**11 on 2008-08-05, 2**12 on 2008-09-28, 2**13 on 2009-04-04, 2**14 on 2009-10-03. 13:33
that last one was the trig tests, I think :)
we almost reached 2**15 with alpha. so close. 13:34
jnthn suspects 2**16 is going to tkae us a little while. :-) 13:36
moritz_ well, the exponentials are hard to get by
masak not at all, there's an infinite supply of them. aleph zero. 13:38
they're hard to get *past*, though. :P
increasingly harder, even.
colomon masak: I have one word for you.
automatically-generated-tests
(okay, it's only one word in German. :) 13:39
jnthn Most things are. :-)
moritz_ not even there :-)
jnthn Aww.
moritz_ disappointing, I know 13:40
just when you need it
13:40 kaare joined, kaare is now known as Guest10356 13:43 kda joined
colomon Is there a non-modifying postfix numeric operator in p6? 13:46
moritz_ .succ (can't call that an operator though)
colomon should have a test for it too, though. 13:47
13:48 Ross left
jnthn ! is the obvious one to define and test. 13:48
cognominal mberends: reading baklog "But I think French takes the cake for "un livre" and "une livre" meaning completely different things"; yes, but I don't think they are so many homographs with meaning changed according to gender in French.
arnsholt There's only a handful 13:49
13:49 BrowserUk left
mberends oui, c'est vrai 13:49
13:49 ruoso joined
arnsholt There's un tour/une tour as well, but that's the only one that comes to mind right know 13:49
moritz_ very rare in German 13:50
colomon jnthn: but if I have to define it, that's an unrelated area of code being tested.
arnsholt Very rare in any language I think. It's just asking for confusion
13:50 uniejo left
jnthn colomon: Troo. 13:50
moritz_ can only come up with one example on the spot, and it's pretty weird 13:51
colomon jnthn: doing it anyway, cause can't think of anything better.
moritz_ (der Moment = the moment; das Moment = angular force)
jnthn (different meanings by gender) I think never happens in Slovak, especially given gender is very reliably predictable by the way the noun ends.
(Which was *such* a nice thing. :-)) 13:52
cognominal I had a thread here about some french adjectives changing meaning with place (before or after name, like "grand") or gender ("public", well when specifically used with "femme" or "homme). But these are mere curiosities too. 13:53
That will be fun when we will design Perl 6 grammars with appropriate among context to translate languages. 13:55
That will change from automated statistical methods in fashion today. 13:56
s/among/amount of/ 13:57
14:03 plobsing joined 14:09 PacoLinux joined 14:10 tylerni7 joined
arnsholt cognominal: I'm not sure Perl 6 grammars are the right tool for NLP 14:12
14:12 masonkramer joined
arnsholt AFAIK most people work with formalisms that are more unification-based rather than CFG-based 14:13
But LFG does use CFGs in c-structure
(But that's only a part of the formalism)
14:22 Ross joined 14:26 ejs left, Ross left 14:27 Ross joined 14:34 Ross left 14:35 JimmyZ left 14:39 Ross joined
moritz_ pmichaud++ # it would be hilarious to have Rakudo as a dependency for the PAST optimizer 14:40
14:43 Ross left
Juerd c 14:45
s/c//
cognominal googles for the TLAs arnsholt throws at me :) 14:47
masak pmichaud: I'm surprised at the lack of curlies around the where expression. can that work? even if it can, is it good style?
std: sub foo($a where $a and $a) {} 14:48
p6eval std 31024: OUTPUT«===SORRY!===␤Unable to parse signature at /tmp/8ESmlgwaku line 1:␤------> sub foo(⏏$a where $a and $a) {}␤Couldn't find final ')'; gave up at /tmp/8ESmlgwaku line 1:␤------> sub foo($a where $a ⏏and $a) {}␤ expecting any
..of…
14:48 Ross joined
masak thought so :) 14:48
moritz_ masak: can't work with 'and', would have to be '&' for a smart-matchable junction
masak yes. that'd work.
still, questionable to omit the closure there, IMO.
jnthn It's a precedence issue there.
std: sub foo($a where $a && $a) {}
p6eval std 31024: OUTPUT«===SORRY!===␤Unable to parse signature at /tmp/kwzajBi5pl line 1:␤------> sub foo(⏏$a where $a && $a) {}␤Couldn't find final ')'; gave up at /tmp/kwzajBi5pl line 1:␤------> sub foo($a where $a ⏏&& $a) {}␤ expecting any
..of:␤…
jnthn heh 14:49
moritz_ but then you'll probably need parens around the sub-smartmatch
jnthn std: sub foo($a where ($a && $a)) {}
p6eval std 31024: OUTPUT«ok 00:01 113m␤»
jnthn Aye, probably wise.
moritz_ std: sub f($x where $a & $a) { }
p6eval std 31024: OUTPUT«===SORRY!===␤Variable $a is not predeclared at /tmp/AxcNzZ2l6n line 1:␤------> sub f($x where $a⏏ & $a) { }␤Variable $a is not predeclared at /tmp/AxcNzZ2l6n line 1:␤------> sub f($x where $a & $a⏏) { }␤Check failed␤FAILED
..00:…
moritz_ std: sub f($x where $x & $x) { }
p6eval std 31024: OUTPUT«ok 00:01 113m␤»
masak I've used an &-junction of roles in a where clause. that feels OK, for some reason. 14:50
subtypes, too. 14:51
probably because it feels so declarative.
jnthn It's OK so long as you don't expect it to play into candidate ordering. 14:53
14:54 Ross left
jnthn uses Rakudo to generate test data sets for $dayjob 14:54
moritz_ \o/
another productive use of Perl 6 :-)
jnthn heh, good enough data set to crash the thing it was testing :-) 14:56
15:03 patspam joined 15:04 takadonet left 15:07 plobsing left, __2synth joined
dalek kudo: 4f9ca44 | (Solomon Foster)++ | src/core/metaops.pm:
Handle single-arg hash hypers as well.
 15:09
pugssvn r31025 | colomon++ | [t/spec] Single-op hyper hash tests. 15:10
15:11 synth left 15:12 __2synth left 15:13 tedv| joined
bbkr rakudo.org is down :( 15:16
cosimo while (! ordered(@list)) { @list = shuffle(@list) } 15:18
how do i do that in perl6?
and hi everyone :) 15:19
moritz_ @list.=pick(*) while ! [<=] @list
cosimo moritz_: wow, cool 15:20
moritz_ I see you want to implement RandomSort, with average exponential run time :-)
cosimo moritz_: i want to add a code example here
en.wikipedia.org/wiki/Bogosort
moritz_ rakudo: my @l = 3, 1, 2; @l.=pick(*) while ![<=] @l; say @l.perl 15:21
p6eval rakudo 7a2ede: OUTPUT«[1, 2, 3]␤»
moritz_ wow, there's even a name for it... didn't know that :-)
rakudo: my @l = 3, 1, 2; @l.=pick(*) until [<=] @l; say @l.perl 15:22
p6eval rakudo 7a2ede: OUTPUT«[1, 2, 3]␤»
moritz_ even nicer
cosimo amazing
moritz_ wonders if .classify would make radix sort in Perl 6 really awesome 15:27
I'm sure we can beat the python implementation by a factor of two in code length :-) 15:28
15:29 Trashlord left
masak rakudo: my @a = "!", a => "1", b => "2", c => "3"; my ($name, %opts) = @a; say %opts.perl # RT #74302 15:30
p6eval rakudo 7a2ede: OUTPUT«{"a" => "1"}␤»
masak please fix this. :/
15:30 Trashlord joined
moritz_ masak: you can hope for pmichaud's magical list refactoring 15:30
15:30 BrowserUk joined
cosimo en.wikipedia.org/wiki/Bogosort#Perl_6 15:31
\o/
moritz_ rakudo: my @l = 3, 1, 2; @l .= pick(*) until [<=] @l; say @l.perl 15:32
p6eval rakudo 7a2ede: OUTPUT«[1, 2, 3]␤»
moritz_ rakudo: my @l = 3, 1, 2, 4, 5; @l .= pick(*) until [<=] @l; say @l.perl
p6eval rakudo 7a2ede: OUTPUT«[1, 2, 3, 4, 5]␤»
cognominal tries to wrap his head around NativeCall.pm6 in zavolaj
moritz_ wow, still within the time limit
cosimo moritz_: wikipedia doesn't have perl6 syntax highlighting, guys :-)
we need to tell them
moritz_ and then they'll ask how to parse Perl 6 15:33
and we point them to STD.pm6
and they say "ugh" and move along
hugme: tweet rakudoperl Bogosort #perl6 example at en.wikipedia.org/wiki/Bogosort#Perl_6
hugme hugs moritz_; tweet delivered
cosimo :) 15:34
pmichaud (parens or curlies in where clause) yeah, I knew I was potentially mangling the syntax. otoh, I wanted to make it look less complex and line-noisy 15:35
moritz_ cheating! :-)
15:35 rgrau joined
pmichaud besides, sometimes the spec itself changes in response to use cases like this :-P 15:35
15:36 meppl joined
pmichaud masak: the list refactoring fixes RT #74302 15:36
right now I consider list refactoring to be highest priority -- higher than closures or anything else, if only because Rakudo's current broken implementation is leading people down far too many false paths. 15:37
masak agreed.
moritz_ +0.9
(the other 0.1 parts scream "but @other_stuff is important too!!11!" :-)
masak closures are easily workaroundable and not something people stumble on until after a while.
15:38 takadonet joined
moritz_ pmichaud: if you have a plan for void-context-detection in PAST, could you write a few lines about it, and hope that somebody else takes it up? 15:38
pmichaud moritz_: it's faster for me to write it than to explain it. 15:40
I mainly need to decide what to name the node type. 15:41
moritz_ ok
pmichaud so, suggestions there would be helpful. :)
I've already somewhat rejected PAST::Context
because "context" is often too overloaded a word.
moritz_ it really depends on the use case 15:42
do you want it to return a boolean value?
pmichaud no
moritz_ or more a IfVoidContext?
pmichaud it's a node that says "if the caller expects xyz type, return the result of compiling self[0], otherwise return self[1]" 15:43
but it's a compile time return, not a run-time one.
(thus it's not really the same as PAST::Op(:pasttype<if>, ...), which is a runtime test. 15:44
masak lol i blogged: use.perl.org/~masak/journal/40371
moritz_ masak: undef is gone. You probably meant Nil or Mu or something. # wouldn't that be s/Mu/Any/? 15:46
it's really rare that Mu is actually needed
masak moritz_: I've only ever needed Nil, but since Nil misbehaves in some cases, I used Mu instead. but I can change it to Any. 15:48
cognominal Masak, I am not sure you say anywhere in your blogs what CGE is
GGE!
masak cognominal: I could provide a link.
the README explains what it is.
cognominal I guess that this suppose cloning or looking to a git repository? 15:49
masak sorry, what? no, you can read the README online. 15:51
moritz_ GGE is not in proto's projects.list :( 15:52
15:53 zostay joined
masak o.O 15:53
please, someone feel free to add it.
mberends hokay, in a few minutes 15:54
masak is a bit too scattered at the mo' 15:55
moritz_ could resist the urge to point out that masak++ has commit access to proto :-)
masak :) 15:57
moritz_ time to go home...
masak I usually add my own projects. GGE was a secret project at first, so that's how it avoided being added, I guess. 15:58
16:04 ruoso left
pmichaud wonders when masak++ will register sekritprojects.org. Or did he do that already and not tell anyone?!? ;-) 16:05
afk for a while 16:09
masak pmichaud: I'm not sure you've grokked the essence of a sekrit project :P
(1) it doesn't need a website :) 16:10
16:19 ruoso joined
cosimo tried to summarize my interpretation of the bogosort snippet, it was a fun ride. hope it's not too much bullshit 16:22
my.opera.com/cstrep/blog/2010/05/31...t-in-perl6
mberends has a sekrit website at 192.168.1.2 16:24
moritz_ cosimo: FYI, @list.pick(3) would pick three random elements from @list
cosimo: so .pick(*) could be read as "pick all"
cosimo oh, ok, now it makes sense 16:25
masak swimming & 16:31
16:31 masak left 16:34 avar joined
avar duuudes 16:35
16:36 synth joined 16:37 lest_away joined
avar does rakudo really need to run spectests with --jobs? It churned up to load ~40 before I managed to kill it, hovering at load ~3 with just one test process 16:37
16:37 lest_away is now known as lestrrat, athomason joined
moritz_ do you have, like, very little memory available? 16:38
anyway, you can disable it with an env variable (see README)
16:38 mdxi joined
avar I have ~500MB of RAM free, since some tests seem to take 200-300MB of memory and I have 4 cores spectest seesm to require at least 1.2GB to be comfortable with --jobs 16:39
Anyway. Yes I can tweak it. It's just a bit unfriendly to have that as the default when rakudo is eating this amount of resources, that's all. 16:40
Maybe it could adjust the --jobs count based on free memory
16:40 mdxi left
moritz_ I wouldn't oppose it, but I don't know how to do it in a way that doesn't break everythiing except linux 16:41
so the proverbial "patches welcome" holds, in lack of a friendlier way to phrase it (I don't mean to be rude)
Su-Shee dinner, code. 16:45
16:47 pmurias joined 16:50 mdxi joined
sorear I just expect everything to run -j1 unless instructed otherwise 16:58
17:01 finanalyst joined
colomon btw, 33,043 passing tests, and I'm just getting started on the hyper hash stuff, there are a lot of neat cases to add code and tests for. ;) 17:02
partying with family now, however. 17:03
17:04 envi^home left 17:06 takadonet left, skids joined 17:11 Ross joined
bbkr rakudo: say (none() ~~ all()); say (none() ~~ any()); # what is the logic behind junction comparision in this case? if none() is the whole universe then it should be "1" in both cases, am i right? 17:12
p6eval rakudo 4f9ca4: OUTPUT«1␤0␤»
moritz_ rakudo: say all() ~~ none() 17:13
p6eval rakudo 4f9ca4: OUTPUT«1␤»
moritz_ that's easy to understand at least
17:15 orafu joined
moritz_ ./parrot_install/bin/parrot --hash-seed 1 perl6.pbc t/spec/S05-capture/caps.rakudo 17:15
fails
bbkr i understand it this way: in "set theory" none() can be represented as a whole universe with excluded empty element. while all() is empty element. so all() ~~ none() should be 0. 17:17
moritz_ do you means s:g/element/sset/ ?
for all x element A $predicate is true for empty A in mathematics 17:18
because it's defined as
not exists x element A !$predicate
(uhm, set, not sset)
bbkr indeed, thanks, it's clear now. moritz_++ 17:19
17:20 fda314925 left 17:21 fda314925 joined
moritz_ rakudo: 'abc' ~~ /<alpha>/; $<alpha> = 5; say $<alpha> 17:27
p6eval rakudo 4f9ca4: OUTPUT«Cannot assign to readonly value␤ in '&infix:<=>' at line 1␤ in main program body at line 11:/tmp/xJObQUaCPT␤»
jnthn That "at line 1" is bull. 17:29
17:29 patrickas joined
patrickas sabaho perlsixers 17:29
jnthn o/ patrickas 17:30
moritz_ oh hai
patrickas really enjoyed the perl6 bogosort thingy :)
hello moritz_, jnthn
jnthn There's a name for even the silliest algorithms. :-) 17:31
moritz_ actually it's a non-algorithm
at least I've learned in my CS classes that an algorithm has to terminate
jnthn It'll terminate when a Rakudo memory leak causes the process to be killed. 17:32
;-)
moritz_ wow, I get random parrot + rakudo behaviour even if I set hash seed by command line option 17:35
jnthn On chunks.t? 17:37
moritz_ caps.t 17:38
but it doesn't need .caps or .chunks to reproduce
Su-Shee mberends: Unable to find module 'FakeDBD::CSV' in the @*INC directories. Did I miss something? 17:42
(while make-ing)
17:44 pmurias left
patrickas I'm going over the perl6 advent calendar making sure all the example work on rakudo.... 17:45
in day 2 the one liner at the bottom does not work on windows (quotes need to be inverted) which is not rakudo's fault, but could be mentioned in the post 17:46
mberends Su-Shee: oops, that's a bit of the future that hasn't been backported to the present yet ;)
please comment out any mention of FakeDBD::CSV
also, this points out a need to 'soft fail' on unmet dependencies. hmm. 17:47
Su-Shee ah, there it is. 17:48
17:54 cdarroch joined, cdarroch left, cdarroch joined, Jedai joined
avar moritz_: The current behavior is obviously how the devs prefer it. I don't see how to patch it without mucking up the current behavior. 17:55
moritz_: I just try to check out perl biannualy. Tested it on a linode and had to hard-kill it from a console because it locked up the box in ~30 load because I ran spectest and went away for tea ;/ 17:56
running it on one core was pretty cpu/memory heavy too, but it got through it
Su-Shee ha. I knew it. Positional of Str.
mberends yes, all you probably needed was to call the same lib func once per column. 17:58
Su-Shee yes. darn. ;)
17:59 pmurias joined
sorear well, you see, the devs are running on real computers 18:01
with 3+ GB of memory and 4+ cores
patrickas rakudo: my $b = Nil; $b //= 7; say $b; 18:04
p6eval rakudo 4f9ca4: OUTPUT«␤»
patrickas rakudo: my $b = Any; $b //= 7; say $b;
p6eval rakudo 4f9ca4: OUTPUT«7␤»
18:06 atomicstack left
Su-Shee mberends: that kind of sub with returns Positional of Str - what is that exactly? why doesn't it have ; and why is { ... } ok? 18:06
mberends Su-Shee: it's how zavolaj registers native functions as subs 18:08
Su-Shee mberends: it looks nice. 18:09
mberends the data types determine how values are sent to / returned from the lib
jnthn++ for zavolaj
jnthn Su-Shee: Esssentially, it uses Perl 6 signature introspection to map args to C types and, but the Parrot NCI does most of the hard work really. :-) 18:10
s/ and//
Su-Shee what it does I understand basically - I was wondering about the style.
jnthn Su-Shee: { ... } is just a stub - essentially the body is replaced by something that magically forwards the call to the C library
It'd actually work fine with just an empty block too. 18:11
18:11 gbacon joined, TiMBuS left
jnthn But I like it with the ... as "something to come here" :-) 18:11
Su-Shee jnthn: that's zavolaj specific or a basic p6 feature?
jnthn Su-Shee: Which bit? The ... ?
Su-Shee: That's general syntax for declaring stub code.
Su-Shee jnthn: looks like "..." as in "well you know what to do" ;)
jnthn rakudo: sub foo() { ... }; say foo()
p6eval rakudo 4f9ca4: ( no output )
jnthn ...huh
oh, locally it says "Stub code executed" 18:12
Don't know what's up with p6eval.
mberends rakudo: ...
p6eval rakudo 4f9ca4: ( no output )
Su-Shee I'm always envious when ruby folks have that pretty shortened code.
18:14 molaf joined, kda left
Su-Shee submethods? when did that happen? 18:17
pmurias pages.cs.wisc.edu/~liblit/pldi-2009-b/ # a really cool way of automatically generating bindings to c libraries 18:21
mberends Su-Shee: the ... spec: perlcabal.org/syn/S03.html#line_2223
Su-Shee ah, nice. 18:22
"stubby exception generators" - noone will believe me if I say this. :) 18:23
jnthn Just wait until you have to tell them about phasers. :-) 18:25
Su-Shee aaah I just forgot a ++ on saturday.. I'm such an idiot. :)) 18:26
jnthn: I already tried. ;)
moritz_ jnthn: in a parameter trait, will I be able to get a reference to multi candidate it's in? 18:27
18:27 tedv| left
moritz_ ie for multi foo($x is autothreaded, ... ) I want to grab the current candidate to add a few more of them 18:27
in the 'autothreaded' trait handler 18:28
jnthn moritz_: Hmm...not quite sure how that'd work. 18:33
moritz_: Maybe it could be made available as some contextual.
moritz_ that might work, yes
jnthn Dunno, I can see the use case. Probably needs more thought. Nobody implemented parameter traits yet. 18:34
(as in, custom ones)
They're not so hard to do, mind.
patrickas rakudo: my $b = Nil; $b //= 7; say $b; #Is that supposed to print 7 ? 18:35
p6eval rakudo 4f9ca4: OUTPUT«␤»
moritz_ I think so, yes 18:36
rakudo: say (my $b = Nil).defined
p6eval rakudo 4f9ca4: OUTPUT«1␤»
moritz_ rakudo: say (my $b = Nil).perl
p6eval rakudo 4f9ca4: OUTPUT«()␤»
patrickas rakudo: say Nil.defined; 18:37
p6eval rakudo 4f9ca4: OUTPUT«0␤»
moritz_ it should be the other way round :-)
Nil.defined should be True, and assignment to a variable should turn it into Any
jnthn nomtime! 18:38
moritz_ already had some tasty nom
Su-Shee had egg salad with green asparagus.
moritz_ had white asparagus yesterday - 0.5kg per person :-) 18:39
pugssvn r31026 | moritz++ | [t/spec] fudge randomly failing tests for Rakudo 18:40
Su-Shee minimum :)
18:46 alanhaggai left, masonkramer left, alanhaggai joined
moritz_ rakudo: pir::sleep(Inf); say "alive" 18:47
p6eval rakudo 4f9ca4: OUTPUT«===SORRY!===␤The opcode 'sleep_p' (sleep<1>) was not found. Check the type and number of the arguments␤»
moritz_ rakudo: pir::sleep_vN(Inf); say "alive"
p6eval rakudo 4f9ca4: OUTPUT«error:imcc:syntax error, unexpected PREG, expecting '(' ('$P45')␤ in file 'EVAL_1' line 65␤===SORRY!===␤syntax error ... somewhere␤»
moritz_ rakudo: pir::sleep__vN(Inf); say "alive"
p6eval rakudo 4f9ca4: ( no output ) 18:48
18:49 clintongormley left 18:51 clintongormley joined
Su-Shee \o/ someid somedata something somethingelse ;) 18:51
moritz_ not the classical 'fooid foo bar baz'? :-) 18:52
18:52 masak joined
Su-Shee it's from fetchrow_hashref :) 18:53
patrickas rakudo: sub lsay(@a) { @a.perl.say }; lsay (1, 2, 3, 4)>>++; 18:54
p6eval rakudo 4f9ca4: OUTPUT«Cannot assign to readonly value␤ in '&infix:<=>' at line 1␤ in 'hyper' at line 5466:CORE.setting␤ in main program body at line 1␤»
patrickas Is that the right behaviour ? 18:55
moritz_ yes
just like 1++ is forbidden
patrickas yea that makes sense ... it it strage that it is used as an example in the advent calendar ... maybe alpha was buggy .... 18:56
moritz_ it was
avar moritz_: Just t/spec/integration/advent2009-day17.t takes ~600MB to run. It seems it (and maybe tests like it) are actually the main problem 18:59
I can't run it to completion without locking up my box
moritz_ avar: we'll probably move some of those tests to an extra test target 19:00
avar should I file a bug for it?
moritz_ if you feel like, yes
avar (is there a bug reporting tool?)
moritz_ nope
avar email? 19:01
moritz_ [email@hidden.address]
19:01 tri1 joined
masak heh. the Wikipedia user who made the bogosort addition has made one edit in total. :) 19:02
pugssvn r31027 | moritz++ | [t/spec] test for RT #61772, *@a is copy messed up in Rakudo 19:03
moritz_ 100% Perl 6 contributions!
patrickas rakudo: my @a = (1, 2, 3, 4) ; @a >>/=>> 2; #I know this is the same case but it feel like it should dwim 19:04
p6eval rakudo 4f9ca4: OUTPUT«Cannot assign to readonly value␤ in '&infix:<=>' at line 1␤ in 'hyper' at line 143:CORE.setting␤ in 'hyper' at line 150:CORE.setting␤ in main program body at line 1␤»
dalek kudo: cde6abe | moritz++ | src/core/system.pm:
enable argumentless sleep(); closes RT #57294
moritz_ patrickas: that one should work
masak ah, it's cosimo++. :)
patrickas oh ok in that case I'll add a test for it 19:05
moritz_ when it's assigned to an array first, it has a mutable container
that's why this case is supposed to work, but not the one before
19:06 masonkramer joined
avar moritz_: ok, sent 19:06
moritz_ avar++
19:08 finanalyst left
moritz_ rakudo: sub f($x) as Int { "12" }; say f(3).perl 19:10
p6eval rakudo 4f9ca4: OUTPUT«===SORRY!===␤Could not find sub &trait_mod:<as>␤»
moritz_ rakudo: sub f($x as Int) { say $x.perl }; f("42") 19:14
19:14 bluescreen joined
p6eval rakudo 4f9ca4: OUTPUT«42␤» 19:14
19:14 bluescreen is now known as Guest56962
moritz_ where's that implemented? in the binder somewhere? 19:14
jnthn moritz_: Yeah
moritz_: The compiler knows about that one. 19:15
moritz_ when I have a type object, how do I coerce to that type in current Rakudo? 19:16
ooh, I know
my $typename = $type.WHAT.perl()
$obj."$typename"()
Su-Shee I have a stupid question.. "self" without $ .. where does that come from? 19:17
moritz_ rakudo: sub f($x) { $x }; &f.wrap({ say "foo" }); f(3)
p6eval rakudo 4f9ca4: OUTPUT«No such attribute 'proxy' in class 'Sub'␤ in main program body at line 1␤»
moritz_ Su-Shee: from the compiler
Su-Shee: a bit like 'this' in C++
Su-Shee wait, am I just confused now.. 19:18
moritz_ bah. If &sub.wrap was working, implementing 'as' coercion would be a piece of cake
Su-Shee I'm just confused. 19:19
perl 5 and 6 and javascript is too much.
masak Su-Shee: consider 'self' as a special term that is defined within the confines of a method body. 19:20
pugssvn r31028 | patrickas++ | Added test for in-place operators
19:22 masak left
jnthn moritz_: Yeah, wrap needs re-doing for master. 19:30
pugssvn r31029 | patrickas++ | Added test for map form and hyperoperator form. 19:34
r31030 | moritz++ | [t/spec] test for RT #64310, $^a after my $a 19:38
19:39 patrickas left 19:43 molaf left
moritz_ rakudo: my @a=1,2,3; say ( ^@a ).perl 19:45
p6eval rakudo 4f9ca4: OUTPUT«0..^3␤»
moritz_ rakudo: say :(3).params[0] 19:47
p6eval rakudo 4f9ca4: OUTPUT«Parameter()<0x1d90c70>␤»
moritz_ rakudo: say :(3).params[0].constraints 19:48
p6eval rakudo 4f9ca4: OUTPUT«3␤»
jnthn rakudo: say :(3).params[0].type
p6eval rakudo 4f9ca4: OUTPUT«Int()␤»
jnthn nice :-)
pugssvn r31031 | moritz++ | [t/spec] Test for RT #70720, :(3) has the 3 available to introspection 19:51
19:54 lue joined
lue hello! 19:55
phenny lue: 11:45Z <masak> tell lue that you asked why trigonometry entered into the complex version of &exp. there's an exciting field called 'complex analysis' which holds the exact answers, but the short story is that 'it all fits'. :) the closeness between exp and sin/cos can be seen through their Taylor series.
moritz_ rakudo: say '4' ~~ /<-alpha>/
p6eval rakudo 4f9ca4: OUTPUT«4␤»
moritz_ rakudo: say '4' ~~ /<!alpha>/
p6eval rakudo 4f9ca4: OUTPUT«␤»
20:01 tri1 left 20:02 Psyche^ joined
pugssvn r31032 | moritz++ | [S05] correct think in .trans example 20:03
r31032 |
r31032 | A zero-width assertion doesn't do us any good here.
moritz_ I'm currently thinking about how to implement Str.trans 20:04
and I find that I have a hard time coming up with an efficient scheme 20:05
that handles both regexes and literals as patterns well
any ideas? 20:06
20:06 Patterner left, Psyche^ is now known as Patterner
moritz_ rakudo: my @x = <a b d>; say 'a' ~~ /@x/; say 'c' ~~ /@x/; 20:10
p6eval rakudo 4f9ca4: OUTPUT«␤␤»
moritz_ huh? I thought that was working...
lue heads for S32::String [Or whatever it is]
jnthn moritz_: me too... 20:12
lue wow. The spec is a big help on str.trans [at least in S32]
</sarcasm>
moritz_ lue: perlcabal.org/syn/S05.html#Transliteration
jnthn rakudo: my $x = <a b d>; say 'a' ~~ /$x/; say 'c' ~~ /$x/;
p6eval rakudo 4f9ca4: OUTPUT«␤␤»
jnthn rakudo: my $x = 'a'; say 'a' ~~ /$x/; say 'c' ~~ /$x/;
p6eval rakudo 4f9ca4: OUTPUT«a␤␤»
jnthn :-S 20:13
lue lets see... if I were to type "hi".trans("i" => "o") I would get "ho" ? 20:16
moritz_ rakudo: tr/a/b/
p6eval rakudo 4f9ca4: OUTPUT«===SORRY!===␤Confused at line 11, near "tr/a/b/"␤»
moritz_ alpha: say "hi".trans("i" => "o")
p6eval alpha 30e0ed: OUTPUT«ho␤»
lue I would imagine some sort of translation. The easiest I see is literal -> regex, then all you deal with is a regex. 20:18
moritz_ and then you have to determine which regex matched 20:19
lue alpha: say "Help keep our e's on earth!".trans("e" => " ")
p6eval alpha 30e0ed: OUTPUT«H lp k p our 's on arth!␤»
moritz_ so either you need some cleverness involved there, or match each regex separately (expensive)
lue or, if you want to be crafty, the other way around 20:20
regex -> literal
alpha: say " something or other about whitespace".trans( / \s+ / => ' ' ) 20:21
p6eval alpha 30e0ed: OUTPUT« something or other about whitespace␤»
sorear moritz_: { make $replacement_str }
moritz_ sorear: so... compile each alternative to a separate proto regex?
20:22 Su-Shee left 20:23 Su-Shee joined
sorear moritz_: or just put them in a single alternation 20:24
moritz_ with eval? or how do I get the { make $replacement} into each branch?
perlgeek.de/blog-en/perl-6/contribu...trans.html lolihazblogth 20:29
jnthn moritz_++ 20:31
20:31 justatheory joined
moritz_ having a simple variant would already help me for JSON::Tiny :-) 20:32
(probably could come up with such a thing in 10 minutes... but others should have fun too)
Su-Shee had her fun for today and pushed it out already. ;) 20:34
20:34 eternaleye left
lue what is the difference between s/// and tr/// ? 20:34
moritz_ s/ab/cd/ will replace 'ab' by 'cd', but leave 'a' alone 20:35
tr/ab/cd/ will replace each a by c, and each b by d
also tr does global transliteration, s/// defaults to one substitution only (needs :g for all)
lue alpha: say "abra".trans("ab"=>"cd") 20:36
p6eval alpha 30e0ed: OUTPUT«cdrc␤»
lue ah. Well, there goes using s/// :P
20:39 kda joined
sorear tr/// also does left-to-right longest token matching 20:44
there is no reasonable way to change \" into " in a string while allowing backslashes to be escaped using s/// 20:45
with tr, it's trivial
20:46 jaldhar left
moritz_ example? 20:46
sorear "\\\\" -> "\\\\", "\\\"" -> "\""
20:46 jaldhar joined
moritz_ I mean, tr example? 20:46
sorear that was the tr example
moritz_ I don't see a tr/ in there 20:47
sorear I don't know the tr syntax offhand
lue rakudo: say '"'
p6eval rakudo cde6ab: OUTPUT«"␤»
moritz_ as I understand S05, you can't do that with tr directly, only with the method form 20:48
(ok, it's nit-picking, I admit)
20:49 eternaleye joined
Su-Shee <-- bed. tomorrow cleanup. good night everyone. :) 21:02
jnthn o/
21:07 Su-Shee left, kel_ left 21:08 Guest56962 left
lue rakudo: $/<a><b>; say "Alive!" 21:09
p6eval rakudo cde6ab: OUTPUT«Method 'postcircumfix:<{ }>' not found for invocant of class ''␤ in main program body at line 1␤»
sorear viv can handle that :> 21:10
lue My question is: where'd the { } come from?
moritz_ from the < >
see S02 and S05
lue afk 21:12
21:12 Ross left, Guest10356 left 21:26 kda left
sorear TimToady: How does ./std precompile modules? Looking at it it seems to just be a thin wrapper over STD->parsefile 21:26
pugssvn r31033 | sorear++ | [viv] In $<foo> = (bar), for gimme5 compatibility, disable binding to $0. 21:32
21:38 pnate joined
pmurias sorear: would it be difficult to expose viv as a library so i could conviently get access to VAST for mildew (without resorting to evil tricks to subvert the executable)? 21:42
sorear The VAST-generating part of viv already is a library 21:43
use STD; use Actions; STD->parsefile(..., actions => 'Actions')->{_ast} 21:44
21:45 rgrau left
sorear 'Actions' dynamically generates VAST::foo classes; the VAST:: classes "defined" in viv are most correctly viewed as supercede class VAST::foo { ... } 21:45
pmurias ok thanks will attempt to remove the hackery tommorrow 21:48
21:52 kda joined, kda left
sjohnson std: my $a; 21:57
p6eval std 31032: OUTPUT«ok 00:01 114m␤»
21:58 ruoso left 22:02 pmurias left 22:11 pnu joined 22:20 pyrimidine joined
pyrimidine regarding Str.trans, I had an implementation working on my github fork 22:21
22:24 pnu left 22:26 pnu joined
[Coke] phenny: tell patrickas to check the advent tests in t/spec/integration 22:26
phenny [Coke]: I'll pass that on when patrickas is around.
pyrimidine moritz_: here is the latest version for Str.trans that was passing tests (including closures) gist.github.com/420337 22:27
moritz_: the alternation capture is really slowing things down, though 22:28
could be optimized
moritz_: In the gist, it's called trans2 (I implemented a version of .trans in Cool-str.pm that broke at some point). 22:33
22:40 pyrimidine left 22:46 pyrimidine joined
pyrimidine phenny: tell moritz_ that there is a partially working .trans implementation (including closures and :d, :s) at gist.github.com/420337, but there are issues with regex name mapping. Also, odd bug with array version mapping (tests with "&nbsp;&lt;&gt;&amp;" string are not parsing). 22:51
phenny pyrimidine: I'll pass that on when moritz_ is around.
23:01 orafu left, orafu joined 23:29 azert0x left 23:31 pyrimidine left 23:38 masonkramer left 23:54 clintongormley left

Powered by Cro and the Raku® Programming Language.

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

Thank you!