»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or rakudo:, std:, or /msg camelia p6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by masak on 12 May 2015.
dalek ast: 799f8f6 | TimToady++ | S32-exceptions/misc.t:
routine suggestions now leave off &
 01:03
dalek p: 3fc7722 | TimToady++ | src/vm/moar/QAST/QASTOperationsMAST.nqp:
tab damage
 01:10
p: 4888aa6 | TimToady++ | / (27 files):
Merge branch 'master' of github.com/perl6/nqp
p: c199c14 | TimToady++ | / (48 files):
Merge branch 'master' of github.com/perl6/nqp
p: 64f13fc | TimToady++ | / (8 files):
Merge branch 'master' of github.com/perl6/nqp
p: 237fd82 | TimToady++ | src/ (2 files):
Merge branch 'master' of github.com/perl6/nqp
p: 9cc4494 | TimToady++ | / (15 files):
Merge branch 'master' of github.com/perl6/nqp
p: 972432c | TimToady++ | src/QRegex/P6Regex/Actions.nqp:
Catch missing + or - between cclass elems
TimToady wow 01:11
dalek kudo/nom: c0fa4af | TimToady++ | tools/build/NQP_REVISION:
bump NQP to get missing + or - between cclass elems
TimToady those merges are all just from doing periodic ". config.status" 01:25
maybe there's a way to tune that out somehow? 01:26
dalek c: de9e7f3 | (Stéphane Payrard)++ | doc/Language/faq.pod:
tyop
 01:32
psch TimToady: that usually happen when you're not on origin HEAD, then commit and pull after 01:37
TimToady: that's assuming less git-familiarity than you might have, though
TimToady: rebasing usually purges merge commits
the funny bit is that git log only shows one merge commit 01:38
mrf urgh Its far to early.
psch so, yeah, something is weird in there
psch shrugs 01:39
probably not really something i know enough about - neither it being early nor why there's so much merge commit spam 01:40
mrf Its 1.45 in the UK. Thats early enough I think :D 01:44
mrf m: say "hello" ~~ /l ** 0..2/ 01:53
camelia rakudo-moar 273e89: OUTPUT«｢｣␤»
mrf m: say "hello" ~~ /l ** 1..2/ 01:54
camelia rakudo-moar 273e89: OUTPUT«｢ll｣␤»
mrf ^ is that a bug? I would expect the 0**2 to match 2 l's as well
TimToady first match takes priority over longest, and it matches '' at the beginning 02:05
mrf TimToady: How come the 1..2 doesn't only match 1 l in the second example? 02:06
timotimo mrf: if you want that, you'd need frugal instead of greedy matching 02:08
yoleaux 2 Nov 2015 16:42Z <nine> timotimo: found another issue with the for range optimization. Now there's only one left (last/next support) till we have a clean spec test again :)
mrf timotimo: Yeah I am just confused why 0..2 appears as non greedy but 1..2 is greedy 02:09
timotimo it's not greedy vs non-greedy
it's what TimToady said. early match trumps all
so when it can match 0 times at the beginning, it's satisfied
m: say "lhello" ~~ / l ** 1..2 / 02:10
camelia rakudo-moar 273e89: OUTPUT«｢l｣␤»
timotimo er
m: say "lhello" ~~ / l ** 0..2 /
camelia rakudo-moar 273e89: OUTPUT«｢l｣␤»
timotimo you can see it greedily matches as many ls as it can. just turns out the most ls it could match at the beginning of "hello" is 0
mrf ...Right! so if it matches without having to step further along the string then it stops becuase its all ready matched 02:11
timotimo that's right
mrf Thats likely to have a rather large impact on my patch allowing ^x syntax 02:12
becuase 0..* will always match surely
on the plus side it means that the two failing tests in my patch are just erroneous tests. Which means its good to go 02:18
TimToady it's relatively useless for scanning, sure, but most parsing doesn't scan, but anchors each match at the end of the previous
mrf TimToady: Yeah. I can see its use. Just a lack of understanding on my part 02:19
mrf TimToady: Don't worry about that earlier gist. I have pushed a PR for NQP to support using 1^ and ^2 for the ** quantifier 02:28
awwaiid Hello. Are any of the caller semantics for EVAL documented/standardized? I want to use things like use caller(...).line and have it use the 'context' passed in, but so far no dice 02:29
awwaiid as in EVAL('callframe().file ~ ":" ~ callframe().line') vs EVAL('callframe().file ~ ":" ~ callframe().line', context => CALLER::) 02:29
mrf timotimo: You might be interested in that PR tbh 02:33
awwaiid I'm prrreetttyyyy sure the answer is "caller as a param for EVAL kinda works and is a rakudo-only feature" 02:34
sprocket i’m writing a NativeCall binding to a struct whose structure differs on various platforms (ie. OS X vs Linux) 02:51
how can i write the definition so that the correct struct is used on different architectures?
at the moment i’ve got: if ($*KERNEL eq 'darwin') { … } else { …. }
but when i attempt to run, it complains that it “Could not locate compile-time value for symbol” 02:52
BenGoldberg m: my uint8 $x = 1; $x +<= 7; say $x; 03:03
camelia rakudo-moar 273e89: OUTPUT«-128␤»
BenGoldberg m: my uint8 $x = 1; $x += 255; $x += 0; say $x; 03:04
camelia rakudo-moar 273e89: OUTPUT«0␤»
dalek osystem: 23b1db8 | (Fayland Lam)++ | META.list:
Add Cache::LRU to ecosystem

See github.com/fayland/perl6-Cache-LRU		 03:11
psch mrf: fwiw, 1^ looks really weird to me 03:17
m: say 1^
camelia rakudo-moar 273e89: OUTPUT«5===SORRY!5=== Error while compiling /tmp/D2Luq2j7sD␤Missing required term after infix␤at /tmp/D2Luq2j7sD:1␤------> 3say 1^7⏏5<EOL>␤ expecting any of:␤ prefix␤ term␤»
psch mrf: i assume it's short for 1 ^.. *
mrf psch: yes.
psch well, it's not like i can parse every syntax in perl6 without having to think about it... :) 03:18
mrf I am not married to it. Was just easier to implement than catching another special case 03:19
psch i don't really object, it's probably a useful enough shorthand
considering /a{M,}/ is gone 03:20
mrf It certainly seemed useful enough as a 'more than' quantifier
psch so having /a ** M^/ as non-whatever-or-infix:<..>-for is neat
to rephrase, having /a ** M^/ as shorthand for /a ** M..*/ seems useful enough, considering perl5 allowed leaving either end out 03:21
timotimo except M^ would be M+1..*
with parens
i am late for bed 03:22
psch huh
yeah, that
also i apparently misremembered leaving M out in {M,N}
perl5 doesn't allow that
timotimo "perl5 doesn't do it" isn't a good reason to not do something, however, by now i'm afraid it might be a tiny bit more confusing to put ^ in there when it implies a +1 (or -1) 03:23
Hotkeys Also while @foo[^10] easily reads as "the first 10 elements of foo", @foo[10^] doesn't seem as intuitive 03:28
I'm sure one could get used to it though
(thanks zero based arrays) 03:29
mrf Hotkeys: I am not sure that the patch I have added would support that but it does sound like it would be something that could be added.
Hotkeys if @foo[10^..*] is possible then I don't see why the shorthand for it wouldn't work 03:30
then again I'm no parsing expert
mrf I would certainly argue that it should exist if it doesn't already 03:31
raiph m: say 10^..*
camelia rakudo-moar 273e89: OUTPUT«10^..Inf␤»
raiph m: say list 10^..*
camelia rakudo-moar 273e89: OUTPUT«(11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 …»
mrf but in a regex it seems to make send to want /<token> ** 10^/ for more than 10. with + being synonymous with 0^ I suppose 03:33
Hotkeys it seems indexing with infinite ranges on infinite sequences gets unhappy
m: my @foo = 1..*; say @foo[10^..*]; 03:34
m: my @foo = 1..20; say @foo[10^..*];
camelia rakudo-moar 273e89: OUTPUT«(timeout)»
rakudo-moar 273e89: OUTPUT«(12 13 14 15 16 17 18 19 20)␤»
Hotkeys but it would still be reasonable for 10^ to work the same in that case 03:35
mrf m: my @foo = 1..20; say @foo[^10]; say @foo[10]; say @foo]
camelia rakudo-moar 273e89: OUTPUT«5===SORRY!5=== Error while compiling /tmp/ywWJqbmNgW␤Unexpected closing bracket␤at /tmp/ywWJqbmNgW:1␤------> 030; say @foo[^10]; say @foo[10]; say @foo7⏏5]␤»
mrf m: my @foo = 1..20; say @foo[^10]; say @foo[10]; say @foo[10^..*]
camelia rakudo-moar 273e89: OUTPUT«(1 2 3 4 5 6 7 8 9 10)␤11␤(12 13 14 15 16 17 18 19 20)␤»
mrf m: my @foo = 1..20; say @foo[^10]; say @foo[10]; say @foo[10^] 03:37
camelia rakudo-moar 273e89: OUTPUT«5===SORRY!5=== Error while compiling /tmp/ERfC_OKXTl␤Missing required term after infix␤at /tmp/ERfC_OKXTl:1␤------> 3ay @foo[^10]; say @foo[10]; say @foo[10^7⏏5]␤ expecting any of:␤ prefix␤ term␤»
mrf Hotkeys: it would look neatly symetrical 03:38
PerlJam just spent about a minute staring at some P5 code wondering why it wasn't compiling ... 03:49
Turns out I had used the P6 ternary op rather than the P5 one 03:50
mrf :D
Hotkeys lol 03:52
atweiden m: class HOH is Hash {*}; my %h := HOH.new('a', 1); my %toml; %toml<a> = %h; say %toml<a>.WHAT; say %toml.perl; 03:54
camelia rakudo-moar 273e89: OUTPUT«(HOH)␤{:a(${:a(1)})}␤»
atweiden is there any way to customize Hash assignment so that the HOH<a> replaces the original %toml<a>? 03:54
mrf can anyone tell me how camelia outputs ␤ rathern than \n. Would really like if my tests could output a more readable form 04:04
PerlJam .u ␤ 04:08
yoleaux U+2424 SYMBOL FOR NEWLINE [So] (␤)
PerlJam mrf: by outputting the appropriate unicoe char 04:09
awwaiid mrf: github.com/perl6/evalbot/blob/mast...ot.pl#L289 and L291
PerlJam er, unicode
psch timotimo: i'm not saying that "perl5 doesn't allow it" is a good reason to not allow it. it just makes my attempt at drawing parellels fall apart. so that's more to be understood as "my argument isn't particularly stable under scrutiny", not "we shouldn't do that"
skids atweiden: I'm not sure what exactly you want to happen in the above. 04:12
atweiden skids: essentially i'm trying to label this as type HOH github.com/atweiden/config-toml/bl...ns.pm#L838 04:13
it's pretty hacky
mrf awwaiid: cheers. I was hoping there would be some magic. Regex is simple enough though
PerlJam: Yeah. Was hoping there was a method like .perl that would do it 04:14
awwaiid mrf: you could like: sub justsayin($x) { $x ~~ s/\n/␤/; say $x } . Could monkey-patch that sucker onto String 04:15
BenGoldberg m: operator postfix<^> { $n ^.. * }; m: my @foo = 1..20; say @foo[^10]; say @foo[10]; say @foo[10^]; 04:16
camelia rakudo-moar 273e89: OUTPUT«5===SORRY!5=== Error while compiling /tmp/pwcuCT1a9Z␤Unexpected block in infix position (missing statement control word before the expression?)␤at /tmp/pwcuCT1a9Z:1␤------> 3operator postfix<^>7⏏5 { $n ^.. * }; m: my @foo = 1..20; say @␤ …»
BenGoldberg m: operator postfix<^> { $n ^.. * }; my @foo = 1..20; say @foo[^10]; say @foo[10]; say @foo[10^];
camelia rakudo-moar 273e89: OUTPUT«5===SORRY!5=== Error while compiling /tmp/s5wImMF2M5␤Unexpected block in infix position (missing statement control word before the expression?)␤at /tmp/s5wImMF2M5:1␤------> 3operator postfix<^>7⏏5 { $n ^.. * }; my @foo = 1..20; say @foo␤ …»
mrf awwaiid: Yeah. I have a is_match and not_match test methods so it should be easy enough to just add it there for now. There is so much new p6 syntax I just wanted to check I wasn't missing something before I did
BenGoldberg m: sub postfix<^> { $n ^.. * }; my @foo = 1..20; say @foo[^10]; say @foo[10]; say @foo[10^];
camelia rakudo-moar 273e89: OUTPUT«5===SORRY!5=== Error while compiling /tmp/yAKqKcqW6k␤Missing block␤at /tmp/yAKqKcqW6k:1␤------> 3sub postfix7⏏5<^> { $n ^.. * }; my @foo = 1..20; say @␤ expecting any of:␤ new name to be defined␤»
mrf awwaiid: thanks though
awwaiid mrf: cool, have fun :)
mrf: interestingly I've been thinking that I might like an irssi pluging that un-does ␤, printing output nicely :) 04:17
mrf awwaiid: I will admit I am bored of writing tests for grammars 04:17
awwaiid: really. I quite like the concise-ness. to each their own I suppose 04:18
awwaiid mrf: maybe you should reverse it and write some grammars for tests. Then write a generator that uses the grammar to generate compliant output.
mrf awwaiid: :D 04:19
awwaiid ok. I should clearly go to bed
mrf I should as well. Just waiting for the game to finish then I will go back to bed 04:20
BenGoldberg debug-cat: this is a test/ 04:22
camelia debug-cat: OUTPUT«this is a test/»
BenGoldberg debug-cat: ooh, look, it thinks it's an echo-bot. 04:23
camelia debug-cat: OUTPUT«ooh, look, it thinks it's an echo-bot.»
BenGoldberg evalbotcontrol version 04:27
camelia This is evalbot revision 3235487
adu what's the difference between () and []? 04:33
adu m: ().Array 04:34
camelia ( no output )
adu m: say ().Array
camelia rakudo-moar 273e89: OUTPUT«[]␤»
adu m: say [].Seq 04:35
camelia rakudo-moar 273e89: OUTPUT«()␤»
Hotkeys m: say ().WHAT 04:45
camelia rakudo-moar 273e89: OUTPUT«(List)␤»
Hotkeys m: say [].WHAT
camelia rakudo-moar 273e89: OUTPUT«(Array)␤»
Hotkeys adu: ^
adu Hotkeys: thanks 04:46
m: say [1, 2] Z, [3, 4]
camelia rakudo-moar 273e89: OUTPUT«((1 3) (2 4))␤»
adu m: say [1, 2] >>,<< [3, 4] 04:46
camelia rakudo-moar 273e89: OUTPUT«[(1 3) (2 4)]␤»
adu it looks like one of those preserves type and the other doesn't, are there any other differences between Z and hyper? 04:47
moritz adu: hyper is eager (even to the point where it's allowed to parallelize), Z is lazy 04:49
adu interesting 04:50
Hotkeys also note that Z alone does Z, 04:51
m: say <1 2> Z <3 4> 04:52
camelia rakudo-moar 273e89: OUTPUT«((1 3) (2 4))␤»
adu I didn't know if >><< defaults to ,
basically, I only know about grammars, so I'm trying to learn about the rest of Perl6
Hotkeys I don't believe it does
m: say <1 2> >><< <3 4>
camelia rakudo-moar 273e89: OUTPUT«5===SORRY!5=== Error while compiling /tmp/dEdagzv9iO␤Missing << or >>␤at /tmp/dEdagzv9iO:1␤------> 3say <1 2> >><7⏏5< <3 4>␤ expecting any of:␤ infix␤ infix stopper␤»
adu you know, to smooth out my knowledge 04:53
Hotkeys smooth knowledge is the best knowledge
adu I know a little bit about classes, but only enough to get the parsers to run
TimToady 10^ is not gonna fly either inside or outside of regex 04:54
way too ambiguous
adu TimToady!
TimToady and it's just not all that useful
adu what's next, FFI? 04:55
TimToady hmm, camelia ain't recompilin' 04:58
psch probably broken jvm build 05:05
setinputlinesep{,s} is missing on jvm 05:06
adu how do I make a type alias?
psch someone recently brought up that camelia should build indepent for both backends, 'cause she apparently doesn't
adu like can I do type GString = OpaquePointer? 05:07
psch m: use NativeCall; constant GString = Pointer; my GString $x;
camelia ( no output )
adu psch: ah ok 05:08
psch hm, Pointer and OpaquePointer are probably different things. the concept still applies
adu constant works
psch m: use NativeCall; constant GString = Pointer; my GString $x; $x.VAR.of.say 05:09
camelia rakudo-moar 273e89: OUTPUT«(Pointer)␤»
psch adu: note, it still contains a Pointer, not a GString
(where "it" is the container)
dalek ast: dc98136 | TimToady++ | S05-metasyntax/charset.t:
test that cclass elems require + or - between
adu psch: what am I doing wrong: pastie.org/10525745 05:14
I got "(Mu)" but I was expecting "!dlroW ,olleH" 05:17
o wait, I needed "returns Str" never mind 05:18
adu wow 05:19
that was super easy
adu <3 NativeCall
adu you know what would be so great, like some kind of Advent Calendar, or something 05:20
TimToady we could, like, do it leading up to Christmas and all
adu I think people would like it 05:20
psch yeah, maybe host it on wordpress
cause that's easy
adu: it exists, perl6advent.wordpress.com/ 05:21
adu I know
I was being facesious, or sardonic
I'm not sure which
psch well, that went over my head :)
that second paragraph from last years 24th is still pretty amazing, TimToady++ 05:22
adu "facetious" sorry my sppeling is bad 05:23
psch well, actually the whole post is great, but i'm a sucker for silly puns
adu psch: have you ever heard of Spoonerisms? 05:25
psch adu: yes, i have. i speak them occassionally, sometimes accidentally 05:26
adu c2.com/cgi/wiki?WorstPunEver
but yes, TimToady++ 05:27
just because
psch now i'm getting a tad sad i can't share one of my favorite word plays 05:29
well, and expect people to understand it
"Das Leben gleicht in manchem Sinne einer Brille - man macht viel durch." 05:30
adu I wonder if there's going to be a post about exception handling this year.
psch adu: you can draft one :)
(the great thing about that quote is lost in translation... :P ) 05:31
adu well I've been wondering Python has 4 sections to it's handlers, (1) try (2) catch (3) else (4) finally, and 1, 2 and 4 are obvious in Perl6, but I'm not sure how one would do (3)
actually, maybe 4 isn't obvious 05:33
psch m: sub f { fail }; with f() -> $res { say "did it" } else { say "nyet" }
camelia rakudo-moar 273e89: OUTPUT«nyet␤»
psch m: sub f { 1 }; with f() -> $res { say "did it" } else { say "nyet" } 05:34
camelia rakudo-moar 273e89: OUTPUT«did it␤»
psch that's what i can make up for (3)
(4) is just a LEAVE in the block that has the CATCH, i think
adu so else is run if no exception is thrown, finally is run regardless 05:35
psch almost, with is if for definedness
TimToady there's also KEEP and UNDO
psch oh, right. UNDO is the better finally
well no 05:36
KEEP and UNDO seperate the possible cases for finally
so we have more granularity
m: if 0 { .say }; with 0 { .say }
camelia rakudo-moar 273e89: OUTPUT«0␤»
psch m: if 0 { say "if'd" }; with 0 { say "with'd" } # actually... 05:37
camelia rakudo-moar 273e89: OUTPUT«with'd␤»
psch m: if fail() { say "if'd" }; with fail() { say "with'd" }
camelia ( no output )
TimToady fail is a variant of return
psch oh, so i'm spreading intuited half-knowledge again... 05:38
i wanted to stop that, actually :/
adu psch: lol
TimToady just start spreading the ohter half instead
*other 05:39
adu I was thinking you could like set $cought = True in the CATCH block, which could be used to implement Python's else block 05:40
psch afair there's a bit missing still with exception handling 05:41
m: $!.pending 05:42
camelia ( no output )
adu it sounds like KEEP is similar to (try ... else), and KEEP and UNDO are like finally?
psch i think that doesn't actually hold the pending exceptions, if there were any
TimToady KEEP is like else, LEAVE is like finally
adu ah ok
TimToady iiuc
psch isn't UNDO like else? 05:43
i'm guessing python does it as "try this, if it dies do the else"
psch goes back to S04 05:43
hmm 05:44
TimToady maybe else is more like the default {} inside CATCH
psch is a block exit through a CATCH that handles an exception unsuccessful..?
psch yeah, i think else is CATCH { default {} } 05:45
(to answer my own question: "probably not" }
adu psch: no, "catch" is run if an exception is thrown AND it matches the symbol given (if any), and "else" is run if an exception is NOT thrown
psch adu: ah, okay, then TimToady was right
adu and actually, UNDO reminds me of Go's defer 05:46
Perl6 is like every language I love rolled into one :) 05:47
psch and the missing bits you can just slang in :)
adu or maybe it's more like LEAVE
psch (depending on how comfortable you are with extending the grammar...) 05:48
adu psch: I've been thinking of writing a new slang
adu <3 grammars
psch adu: did you look at the ones FROGGS++ implemented yet?
adu nope 05:49
psch github.com/FROGGS/p6-Slang-Piersing
adu psch: this is one of the grammars I wrote: github.com/andydude/p6-c-parser/bl...rammar.pm6
psch adu: oh, alright. i guess that means the missing bit is knowing which pegs to hang your tokens on 05:51
adu I think the missing bit for me is learning how all the $* vars work
psch because (currently at least) a slang depends on knowing which tokens already exist, and how to replace the parts you want to change 05:52
llfourn and figuring out the basics of QAST :) 05:53
psch although i can't imagine not having to know what part of the language one wants to replace...
well, for finding pegs we at least have the --target=parse 05:54
llfourn: yeah, that's definitely useful, although i'm not sure it's always neccessary
adu like I would want to make a dramatically different slang, like put "use css; " at the top and the rest of the file would be css 05:55
llfourn psch: thanks I didn't even know aboout --target=parse
psch adu: github.com/tony-o/perl6-slang-sql/ is sort of a step in that direction 05:56
adu: it's of course different in that it exposes a keyword to parse the rest of the statment as a sublanguage 05:57
llfourn adu: I think a slang is a superset of whatever the main language is where as use css would be just replacing the MAIN language
psch doesn't actually know how "use css;" could be useful, but maybe css is just not the best example 05:58
adu psch: that's pretty cool
perhaps
I mean, if you're making a web server, you probably want it at the statement level anyways 05:59
llfourn mmmm for CSS you would just want a grammar I think
psch i'd probably want a simplified subset, because all the newfangled magic css does nowadays seems a little silly to me :) 06:00
it's probably all the writing-a-compiler-in-java i do in my free time ;)
adu CSS is mostly dictionaryies anyways, except the occational gradient and/or url link
psch ...well, *helping* to write a compiler, don't want over-emphasize my contribution :P
+to
adu :)
llfourn I made a slang a while ago but I don't think it works anymore: github.com/LLFourn/perl6-slang-dotty 06:03
adu psch: there was some talk a few months ago of integrating NativeCall, Inline::C, and C::Parser into some kind of hybrid thing 06:04
psch adu: yeah, i remember 06:06
adu: as in, i remember the idea using C::Parser to autogen most of the NativeCall declarations
if not all of it
adu psch: something like that
llfourn: that's pretty involved 06:08
llfourn: where is Dotty?
llfourn adu: Dotty is a method call op?
. .? etc
adu Dotty::Slang::Grammar 06:09
Dotty::Slang::Actions
where are these defined?
llfourn github.com/LLFourn/perl6-slang-dot...g/Dotty.pm 06:10
I build them dynamically
inside EVALs -- quite ugly :\
notice .^add_method etc
adu *mind blown* 06:11
llfourn I think it's less complicated than it looks... I wonder if I could write it better now :) 06:12
llfourn ..maybe using parameterized roles 06:13
llfourn goes afk for a while 06:15
mrf TimToady: Will look at removing support for 10^ from my patch when I get in to work. 06:42
dalek line-Perl5: ffeb71a | (Stefan Seifert)++ | t/ (2 files):
Fix tests on platforms where objectids can be negative

Fixes GH #46		 07:17
nine llfourn: github.com/niner/Inline-Perl5/issu...-149920242 is a different issue 07:21
llfourn nine: ok shall I make a new issue? :) 07:33
llfourn checks if Inline::Perl5 still doesn't work 07:34
nine llfourn: yes, if it still doesn't work, please open a new issue. Output of perl6 -Ilib t/use.t may help diagnosing 07:35
llfourn nine: sure thing 07:35
nine Does perl6-gdb-m work on OS X? 07:47
llfourn nine: like in general or on the test? 07:48
nine in general
llfourn nine: I find it's hit or miss
nine If it works in general, I'd of course be interested in its output on the test :)
llfourn I'll have a go :) 07:49
llfourn nine: nah debug aint working at all 07:51
[Tux] m:$*OUT.nl = "";
m: $*OUT.nl = Str;
camelia rakudo-moar 273e89: OUTPUT«Cannot unbox a type object␤ in block <unit> at /tmp/S17UPVihvy:1␤␤»
[Tux] so half of RT#123978 has been addressed 07:52
synbot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=123978
llfourn nine: would a core dump be useful?
nine llfourn: I doubt as I don't have OS X and Linux's tools probably won't be able to do anything with it. 07:53
[Tux] test 50000 37.191 37.078 07:55
test-t 50000 39.284 39.171
RabidGravy I think that's getting better 07:56
lizmat good *, #perl6! 07:57
nine Good morning lizmat! Thanks for the weekly :)
lizmat yesterday was better, I think
you're welcome!
[Tux] is the «$*OUT.nl = ""» fix a result of jnthn++'s grapheme work? 07:59
lizmat not grapheme, but IO work on MoarVM 08:00
m: &&::{}[];; # this used to fail, but doesn't anymore 08:01
camelia rakudo-moar 273e89: OUTPUT«5===SORRY!5=== Error while compiling /tmp/ePpNWTE8KT␤Undeclared routine:␤ & used at line 1␤␤»
lizmat ah, it does?
lizmat star-m &&::{}[];; 08:01
star-m: &&::{}[];;
camelia star-m 2015.09: OUTPUT«5===SORRY!5=== Error while compiling /tmp/PAXzNrMLiL␤Undeclared routine:␤ & used at line 1␤␤»
lizmat well, it doesn't fail for me locally anymore 08:02
test #19 in t/spec/integration/weird-errors.t
RabidGravy ew 08:06
abraxxa since upgrading to 2015.10 I can't run any DBIish test any more, they all fail with Could not find file 'DBDish::Oracle' for module DBDish::Oracle 08:14
when run with perl6 -Ilib t/45-oracle-common.t 08:15
moritz abraxxa: sounds like a 'require' gone wrong
abraxxa: did you merge from upstream DBIish?
it's fixed there, I believe
abraxxa just checked its repo if there are any commits that i'd need to merge but it doesn't seem so
this one? github.com/perl6/DBIish/commit/08d...cf98841fee 08:16
ok, merging
moritz yes, that one
lizmat hmmm... looks like 76391ca91b37e14b323d broke sub a ($a is rw) {$a}; a 42 08:17
lizmat test #5 in t/spec/S06-traits/misc 08:17
peschwa seems to have done that 08:18
RabidGravy naughty peschwa
moritz but that's supposed to fail, now?
s/now/no/
lizmat yes, and it doesn't anymore
moritz ah
lizmat m: sub a ($a is rw) {$a}; a 42
camelia rakudo-moar 273e89: OUTPUT«Parameter '$a' expected a writable container, but got Int value␤ in sub a at /tmp/2qiClYJJ6U:1␤ in block <unit> at /tmp/2qiClYJJ6U:1␤␤»
lizmat huh? 08:19
m: use Test; sub a ($a is rw) {$a}; throws-like { a 42 }, X::Parameter::RW
camelia rakudo-moar 273e89: OUTPUT« 1..2␤ not ok 1 - code dies␤ ␤# Failed test 'code dies'␤# at /tmp/sIxNEMjt9i line 1␤ ok 2 - # SKIP Code did not die, can not check exception␤ # Looks like you failed 1 test of 2␤not ok 1 - did we throws-like X::Parameter::RW?…»
lizmat m: sub a ($a is rw) {$a}; { a 42 } 08:20
camelia ( no output )
lizmat m: sub a ($a is rw) {say $a}; { a 42 } 08:20
camelia rakudo-moar 273e89: OUTPUT«Parameter '$a' expected a writable container, but got Int value␤ in sub a at /tmp/o7eih1KFDY:1␤ in block <unit> at /tmp/o7eih1KFDY:1␤␤»
lizmat so it apparently only doesn't fail if it's inside a block and nothing's done to it 08:21
moritz m: sub a ($a is rw) {say $a}; if 0 { a 42 }
camelia ( no output )
moritz m: use Test; sub a ($a is rw) {$a}; throws-like { sink a 42 }, X::Parameter::RW; 08:22
camelia rakudo-moar 273e89: OUTPUT« 1..2␤ not ok 1 - code dies␤ ␤# Failed test 'code dies'␤# at /tmp/1M3SzdJJlu line 1␤ ok 2 - # SKIP Code did not die, can not check exception␤ # Looks like you failed 1 test of 2␤not ok 1 - did we throws-like X::Parameter::RW?…»
RabidGravy okay, that's weird travis-ci.org/sergot/http-useragen...s/88943847 whichs seems to suggest that Encode is broken but it tests fine here 08:23
is there anything gone in the last day that could have broken it? 08:24
lizmat RabidGravy: doesn't ring a bell 08:36
RabidGravy yeah, odd. when my computer settles down I'll build a new rakudo and check again 08:37
lizmat .tell jnthn any idea what "get_boxed_ref could not unbox for the representation '20'" could mean ? 08:46
yoleaux lizmat: I'll pass your message to jnthn.
lizmat .tell jnthn nvm, needed some nqp::decont() sprinkling 08:58
yoleaux lizmat: I'll pass your message to jnthn.
dalek kudo/nom: 6e490dc | lizmat++ | src/core/Range.pm:
Move all of Range.pm into the GLR-era

This results in moderate speedups of ~25% and some bug fixes related to ranges where the max is smaller than the min.		 09:21
RabidGravy erk, travis for http::useragent all good again now. something a little weird going on there. 09:22
azawawi hi 09:27
RabidGravy marning 09:28
RabidGravy azawawi, so what cool and excellent things are you going to make today? 09:30
lizmat afk for a few hours& 09:31
azawawi im looking at the github stars (i.e. interest) in my modules lol 09:32
RabidGravy :) 09:33
azawawi funny thing farabi (my old Perl 5 in-browser) experiment has the highest starred github repo
azawawi RabidGravy: Net::Curl has the same interest as farabi6... lol 09:34
RabidGravy: the funny thing is that selenium p6 module has the lowest even though i worked on it the most along with Electron 09:36
RabidGravy: so the least effort has the highest interest lol
RabidGravy even some of the P5 modules I have (e.g. Term::Readkey) which are depended on all over the place have very few stars 09:37
RabidGravy stars some of azawawi's modules ;-) 09:40
RabidGravy I wonder, idly, if there is an API for starring things on github and whether that could be integrated into modules.perl6.org 09:41
azawawi RabidGravy: thanks... it is not about starring... it is about working on stuff that really matters to the Perl 6 community :) 09:44
RabidGravy: but the stars are cool also lol
RabidGravy azawawi, absolutely 09:46
:)
unfortunately my TODO list is getting longer and longer - at some point the dam will burst and I will release a massive bunch of modules as I'm blocking on a few very small things 09:50
RabidGravy I've got about six in flight with various deadlocks and blockers 09:53
jnthn morning, #perl6 09:55
yoleaux 08:46Z <lizmat> jnthn: any idea what "get_boxed_ref could not unbox for the representation '20'" could mean ?
08:58Z <lizmat> jnthn: nvm, needed some nqp::decont() sprinkling
mrf morning jnthn
azawawi RabidGravy: I want to use Perl 6 in my workplace. Hence I am writing more support libraries. 09:57
jnthn azawawi++ 09:58
RabidGravy I want to re-do github.com/jonathanstowe/Emitria as a Perl 6 application
;-) 09:59
right shower
azawawi irclog.perlgeek.de/perl6/2007-05-14#i_16193 # first hello in Perl 6 as far as IRC log gives me... it is a long time :) 10:00
jnthn awwaiid: The context arg to EVAL is decidedly "internal"; note that EVAL being uppercase is to indicate it's a special form, and the compiler is free to optimize away things if it doesn't see a literal call to EVAL in a given lexical scope. 10:01
.tell sprocket Define both structs, and then write constant TheStruct = $*KERNEL eq 'darwin' ?? TheOSXStruct !! TheLinuxStruct; 10:02
yoleaux jnthn: I'll pass your message to sprocket.
JimmyZ jnthn: do we have plans to expose fd to support some system call like github.com/MoarVM/MoarVM/issues/281 ? 10:05
jnthn Yeah, there's an RT that wants it too 10:07
JimmyZ that would be nice :P
not sure about handle though .. 10:08
file handle etc..
mrf m: say "ll" ~~ /l ** 3..4/ 10:09
camelia rakudo-moar 273e89: OUTPUT«Nil␤»
mrf nqp-m: say("ll" ~~ /l ** 3..4/) 10:10
camelia nqp-moarvm: OUTPUT«Substring length (-5) cannot be negative␤ at gen/moar/stage2/QRegex.nqp:1778 (/home/camelia/rakudo-inst-1/share/nqp/lib/QRegex.moarvm:Str:18)␤ from gen/moar/stage2/NQPCORE.setting:696 (/home/camelia/rakudo-inst-1/share/nqp/lib/NQPCORE.setting.moarvm:join:1…»
mrf given that Rakudo/moar seems to hide that error from nqp I assume that is an expected error 10:11
jnthn Rakudo's Match.Str has just been more smartly implemented than NQP's I guess :) 10:12
Oh, or actually it returns Nil on a failed match 10:13
And NQP doesn't have Nil
mrf jnthn: question out of the blue. Do you think that ** 10^ should throw a specific warning in NQP? context: I am adding support for 1^..^10 style range quantifier in nqp 10:15
jnthn: or do you think the 'Malformed range' error it currently throws is fine. 10:16
jnthn Malformed range seems fine to me
I've not really seen people try/expect 10^ to work in a ^10 way so far.
And every error we add bulks the compiler up a little more, so they should at least pull a little weight :) 10:17
Ven_ m: sub postfix:<^>($n){^$n}; 3.succ^
camelia ( no output )
Ven_ m: sub postfix:<^>($n){^$n}; say 3.succ^
camelia rakudo-moar 273e89: OUTPUT«^4␤»
mrf jnthn: yeah. It was something I was discussing very early this morning but TimToady has said no to it. So just wondering if we want to explict it
jnthn I don't think so, unless we see lots of people tripping over it, but so far they haven't.
mrf thats fine. It would have only really been a shortcut for 10^..* in any case which is not exactly a lot of chars. The only argument in its favour was the symetry with ^10 when getting things from arrays 10:18
RabidGravy JimmyZ: RT #126541 10:29
synbot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=126541
RabidGravy please feel free to chime in if you can think of more use-cases
JimmyZ oh, new RT! 10:36
mrf github.com/perl6/nqp/pull/249 10:37
Hopefully that meets approval now. :S
^ If there is someone this change should be documented then if someone can point that out then I will add them 10:38
dalek kudo/nom: 45e56b8 | jnthn++ | src/core/Rakudo/Internals.pm:
Hopefully unbust the JVM build.
 10:46
RabidGravy right off to the seaside to get some fresh air, play nicely now kids ;-)
see y'all later
mrf RabidGravy: \o 10:47
jnthn RabidGravy: Enjoy :) 10:48
mrf: It looks good to me. Will you add some spectests also?
mrf jnthn: I added some tests to t/qregex/rx_quantifier. Was there somewhere else needed? 10:49
jnthn mrf: Yeah, those cover the regex engine, but we should have some in roast too, so it's actually tested as part of the Perl 6 langauge definition 10:51
dalek p: a3ab8d8 | (Mike Francis)++ | src/QRegex/P6Regex/ (2 files):
Allow ^10 style ranges as a quantifier
p: 795fc8d | (Mike Francis)++ | src/QRegex/P6Regex/Grammar.nqp:
Support ^* Inf ranges in quantifiers

Actually test the important \n/\r\n case.
jnthn Merged the PR 10:51
mrf++
mrf jnthn: point me to where they need to be added and I will happily add them :D 10:51
mrf has now got a commit in 3 p6 repos :D 10:52
jnthn mrf: github.com/perl6/roast/blob/master...x/repeat.t 10:53
mrf ok. Will add some tests in there.
jnthn Thanks 10:56
mrf np 10:57
azawawi .tell RabidGravy to get the github stars on a certain github repo, get api.github.com/repos/rakudo/rakudo which returns a JSON response and use "stargazers_count" key :) 11:09
yoleaux azawawi: I'll pass your message to RabidGravy.
azawawi i think this will be useful on modules.perl6.org/ , number of stars and watchers a project has... 11:10
s/project/module/
azawawi forks github.com/perl6/modules.perl6.org/ :) 11:12
moritz azawawi: don't you have push access already? 11:13
azawawi moritz: true lol
moritz: i always forget it :) 11:14
moritz: i will work on my fork and then commit it once it is working... safer that way
one thing i dont understand, most LICENSE files in our repo have "Copyright (c) 2000-2006, The Perl Foundation." 11:15
is that correct?
moritz azawawi: the copyright of the license is by TPF, the copyright of the module typically not 11:17
azawawi I find MIT license to be simpler though 11:18
but that's me :)
tadzik +1 :)
yoleaux 1 Nov 2015 23:58Z <RabidGravy> tadzik: it would be a great boon to me if you could take a look at github.com/tadzik/JSON-Unmarshal/pull/9 :)
tadzik yoleaux: tell RabidGravy I just did, thanks a lot :)
.tell RabidGravy I just did, thanks a lot :)
yoleaux tadzik: I'll pass your message to RabidGravy.
tadzik thanks, yolo
dalek kudo/nom: d48f6ed | jnthn++ | src/core/Rakudo/Internals.pm:
Correct thinko in setting multiple separators.
 11:21
azawawi tadzik: :) 11:21
tadzik: i submitted a PR for rakudobrew build-zef... Can you check it out? 11:22
tadzik azawawi: it should be merged as of few minutes ago
azawawi tadzik: cool
hopefully it wont break anything :) 11:23
tadzik it didn't look like it was about to :P
azawawi cool 11:24
tadzik: re .bat for win32 panda, Nov 24, 2014 till now... the power of the followup :) 11:27
tadzik: thx :)
tadzik :) 11:28
I know a bit about open bugs hanging there forever :P 11:29
dalek ast: 80d5590 | lizmat++ | S03-operators/range-basic.t:
Some more Range tests
 11:36
kudo/nom: 9979bf2 | jnthn++ | src/core/ (6 files):
Move towards naving nl-in and nl-out.

Now, nl-in is for input line separator, and supports being provided with an array of strings. The output line separator is nl-out. Plain nl is deprecated, but still has the effect of setting both for now. Note that Proc and IO::Socket::INET both need further updates.
dalek ast: f985b22 | jnthn++ | S (3 files):
Track nl => nl-in/nl-out change.
 11:38
Ven_ jnthn: is there no point to leaving :nl to be both an alias to :nl-in and :nl-out? 11:59
jnthn Ven_: I don't think so, 'cus it opens the "what if you pass that *and* nl-in" can of worms 12:01
dalek kudo/nom: eabc734 | lizmat++ | src/core/Range.pm:
Make Array methods throw an Awesome error message
Ven_ jnthn: yeah, sure, but that already exists for, say, open, with :r/:w and :rw
(I'd know, it took me 3 PRs to get it working correctly :P)
jnthn I'm not sure one existing awkward case justifies adding another ;) 12:02
Ven_ it might justify removing the original awkward case ;) 12:04
andreoss m: «(1 2 3)».WHAT.say 12:07
camelia rakudo-moar 273e89: OUTPUT«(List)␤»
andreoss m: «(1,2,3)».WHAT.say 12:09
camelia rakudo-moar 273e89: OUTPUT«(Str)␤»
jnthn m: «(1 2 3)».perl.say 12:12
camelia rakudo-moar 273e89: OUTPUT«("(1", IntStr.new(2, "2"), "3)")␤»
jnthn It splits on whitespace, and there ain't any in the second one. 12:12
pdcawley m: quasi { $bibble }.perl.say 12:13
camelia rakudo-moar 273e89: OUTPUT«5===SORRY!5=== Error while compiling /tmp/NZE8SMuuvL␤Variable '$bibble' is not declared␤at /tmp/NZE8SMuuvL:1␤------> 3quasi { 7⏏5$bibble }.perl.say␤»
pdcawley m: quasi { $bibble }.WHAT.say
camelia rakudo-moar 273e89: OUTPUT«5===SORRY!5=== Error while compiling /tmp/4kb2wCJeTN␤Variable '$bibble' is not declared␤at /tmp/4kb2wCJeTN:1␤------> 3quasi { 7⏏5$bibble }.WHAT.say␤»
pdcawley m: my $bibble = 1; quasi { $bibble }.perl.say 12:14
camelia rakudo-moar 273e89: OUTPUT«AST.new␤»
pdcawley Hmm... is that going to make a scope injecting macro a bit tricky to write? 12:14
lizmat masak might know, but he's probably having dinner about now 12:16
pdcawley lizmat: I'm just surprised that the quasi thinks it knows the scope in which it's going to be evaluated. 12:18
m: quasi { 1 } 12:19
camelia ( no output )
pdcawley m: quasi { 1 }.perl.say
camelia rakudo-moar 273e89: OUTPUT«AST.new␤»
pdcawley m: quasi { 1 }.eval.say
camelia rakudo-moar 273e89: OUTPUT«Method 'eval' not found for invocant of class 'AST'␤ in block <unit> at /tmp/cqcCUem9S2:1␤␤»
lizmat pdcawley: /me fully Perl 6 macro noob
pdcawley grins
lizmat pdcawley: it's EVAL nowadays
pdcawley m: quasi { 1 }.EVAL.say
camelia rakudo-moar 273e89: OUTPUT«Method 'EVAL' not found for invocant of class 'AST'␤ in block <unit> at /tmp/MUMHc92jXD:1␤␤»
lizmat ok, but not for AST's apparently :-) 12:20
pdcawley So, how do I evaluate an AST in an environment.
masak I'm here :)
jnthn Note that most macro exploration is actually being done by masak outside of the Rakudo repo, and that macro/quasi etc. are deferred to 6.future
Ah, there's masak :)
pdcawley I _think_ quasi { $bibble } failing to compile is a bug, if the AST doesn't have an environment attached.
masak short answer: quasis are not a get-out-of-jail-free card with regards to scoping.
pdcawley jnthn: fair enough.
masak: they bloody well should be. 12:21
masak if you want unhygiene, you'll want to refer to things (or declare them) as COMPILING::
pdcawley Since the AST they generate doesn't know its scope.
masak pdcawley: well, you're wrong about that.
pdcawley masak: Then why doesn't $anAST.EVAL do anything?
dalek ast: daaa141 | lizmat++ | S02-types/range.t:
Add some Immutable and some int-bounds tests
masak pdcawley: I'm sorry, I don't understand your question. could you rephrase?
you're aware that macros are not fully featured in Rakudo, right?
pdcawley masak: Am now :)
masak Qtrees should be able to do .EVAL, but I'm not sure .EVAL is a good name for that. 12:26
in 007 I've taken to calling it .melt, following some kind of "ASTs are solid, runtime values are liquid" analogy
andreoss if i have a grammar and an action class what should i do to make it work with EVAL ... :lang<Mygrammar>? 12:30
i've tried to figure it out from rubish example, but it seems broken and also in nqp 12:31
arnsholt From memory, you need to nqp::compreg a compiler object that implements the HLL::Compiler interface
awwaiid thanks jnthn. I'm using EVAL to build a REPL, you see. 12:36
dalek c: aa3c53f | lizmat++ | doc/Type/Range.pod:
Add some badly needed Range documentation
 12:37
llfourn andreoss: EVAL is a multi so you can export something with :lang<MyLang> 12:38
llfourn andreoss: are you introducing your language by messing with %*LANG? 12:39
andreoss` not really, i have a simple grammar that parses lisp code and building simple ast (nested arrays), i wonder how much effort would it take to turn it into a Slang 12:41
lizmat
.oO( Lisp In A Block: Wow! )
 12:42
llfourn andreoss: I haven't done a fully contains language in perl6 before but I guess exporting a multi EVAL is best place to start to call your language from p6
jnthn
.oO( It's my Lisp in a block, girl... )
andreoss`: I'd take a look at one of the existing slangs; getting it wired up to EVAL probably isn't enough for that. 12:45
jnthn has an NQP that is totally happy with rn as a single grapheme and is spectesting Rakudo at the moment
Couple of fails :/ 12:46
pink_mist (\r\n as a single grapheme)++ 12:47
jnthn Ah, first failing one is a test that does a silly workaround for Windows anyway 12:48
llfourn andreoss: if you create a slang by modifying %*LANG you won't get EVAL to work either from my experience. When EVALing p6 will use %?LANG to determine which language to use -- but modifying %*LANG doesn't do %?LANG. 12:49
I haven't seen a slang module that provides some way of using the slang in eval other than EVAL 'use MySlang; ...' 12:51
masak andreoss`: did you see github.com/masak/ipso ? 12:52
andreoss` masak: no, but seems like what i was trying to do 12:56
masak :)
Ven_ macro m { state $x++; quasi { $x } }; for ^3 { say m; } 12:57
m: macro m { state $x++; quasi { $x } }; for ^3 { say m; }; for ^3 { say m; }
camelia rakudo-moar 273e89: OUTPUT«2␤2␤2␤2␤2␤2␤»
Ven_ masak: ^ wat? why is my $x not captured? 12:57
masak Ven_: should be, AFAICS. 12:58
Ven_ so, "bug"?
masak yeah
feel free to submit
Ven_ nods
I wouldn't want to take that away from you ;-)
Odud Trting to install Inline::Perl5 using panda on Windows 10 and getting 'cl' is not recognized as an internal or external command error reported by nmake - any ideas/solutions 12:59
masak submits rakudobug
dalek p: c7bfa55 | jnthn++ | src/QRegex/P6Regex/Actions.nqp:
Smarter fix for \v under NFG.
p: db2fdbb | jnthn++ | t/qregex/01-qregex.t:
Harden test against \r\n -> 1 grapheme.
jnthn Odud: It can't find the compiler, it seems. Try running "vcvars64" (or maybe "vcvars32") to see if it sets up your environment enough for things to work. 13:00
(I'm assuming you actually do have a C compiler to hand, given you have nmake) 13:01
Odud Hi jnthm I have installed the community edition of Visual Studio so assumed the compiler is there, thanks for your suggestions also I will go and check the path looks ok 13:02
llfourn m: my $check = "a"; my $sub = sub ($ where $check) { }; $check = "b"; $sub("a") # any way to freeze the value of $check 13:03
camelia rakudo-moar 273e89: OUTPUT«Constraint type check failed for parameter '<anon>'␤ in sub at /tmp/MMlDsvqOLo:1␤ in block <unit> at /tmp/MMlDsvqOLo:1␤␤»
stmuk Odud: there is a link from the start menu called something like "developer prompt" for VS you could try that too 13:04
masak Ven_: huh -- no, wait
Ven_: it's `state`, so it's *shared* between both macro calls, no?
Ven_: that's what `state` does.
Ven_ I increment it.
jnthn masak: I strongly advise against using "state" in macros, fwiw. :)
Ven_ sub f { $++ }; say f, f, f; # shared but different 13:05
m: sub f { $++ }; say f, f, f; # shared but different
camelia rakudo-moar 273e89: OUTPUT«012␤»
masak jnthn: yes, but what *should* it do? :)
arnsholt masak: I guess the question is what the quasi {$x} is supposed to do
jnthn masak: The state vars will persist across compilations, and conflict across concurrent compilations...
masak Ven_: yes, but with the macro they all evaluate late.
arnsholt Should it freeze $x at the time of evaluation, or not
masak Ven_: and the increments happen early.
moritz $*W.sorry('"state" in macros is a really bad idea')
masak Ven_: so I no longer see the bug.
arnsholt Yeah, what moritz said =D 13:06
Ven_ masak: then I feel even stronger against this kind of scope mixups ;)
recipe for disaster
masak we've been down the road of trying to restrict things we don't see the use of before...
arnsholt Thinking some more towards what this would be in something like CL, I'm inclined to say it's not a bug
masak right
I think I can reason my way to that being correct 13:07
Rakudo++ # correct :)
Ven_ arnsholt: CL doesn't allow you to do crazy stuff like (let ((x 2)) `(x))
masak ENOTABUG
Ven_ perl6/007's macros allow this
masak the waves have gone high in #6macros about this today
Ven_ that they have. it's something that's going to trip up a lot of people ;) 13:09
masak it shouldn't be about what will trip up people, it should be about what you can use to build useful things
arnsholt Ven_: Sure it lets you let() over a defmacro(). But trying to write down the CL code, it doesn't actually make sense to do so for what you wrote, I think 13:10
masak imagine if someone came to regexes as they were being built and said "backtracking? that's going to trip up a lot of people! don't do that." :)
masak I will take arguments against Perl 6 scoping a lot more seriously when I feel that the arguments are starting from a general sympathy with Perl and Perl 6 thinking. 13:11
not with what's expected coming from a Lisp direction.
arnsholt Oh, I agree
masak and yes, I am reading all the required books. slowly.
I am paying attention. I am taking notes.
arnsholt I just find that all my thinking about macros ends up being in terms of Common Lisp 13:12
nine Backtracking? What doya wanna next? Recursion?
masak my main conclusion is that Perl 6 is not exactly Lisp :)
arnsholt Common Lisp being the only language with macros with I know
masak and trying to bolt Lisp macros onto Perl 6 would not end happily
arnsholt Yeah, I think something like Dylan is a better point of comparison
masak Dylan is on my list of things to look at, yes.
arnsholt (Assuming Dylan is the one with macros, which I think it is) 13:13
masak it is.
arnsholt Interestingly, I hacked up some CL code similar to what Ven_ wrote. Got an interesting result 13:14
CL-USER(1): (load "test.lisp") 13:15
; Loading /home/arne/test/test.lisp
1
2
3
4
T
gist.github.com/arnsholt/4d285d91f7ad8bb892b4
Whoops, a bit of a mispaste there
Although, I guess the weird result might be due to magic happening in the loop() macro 13:16
stmuk if I type "panda" it takes 5 secs to return the help args 13:17
masak m: my $c = 0; macro m() { my $n = $c++; quasi { say $n } }; for ^3 { m() }; for ^3 { m() } 13:18
camelia rakudo-moar 273e89: OUTPUT«0␤0␤0␤1␤1␤1␤»
masak works as advertised.
so if you want to *share* state across macro calls, you use `state`. 13:19
if you don't want that, you use `my` and the kind of copying used above.
easy :)
masak er, but that `= 0` in the beginning is probably bogus. ignore that. 13:20
it will be run *late*, not *early*.
masak likes this distinction more and more
m: BEGIN my $c = 0; macro m() { my $n = $c++; quasi { say $n } }; for ^3 { m() }; for ^3 { m() } 13:21
camelia rakudo-moar 273e89: OUTPUT«0␤0␤0␤1␤1␤1␤»
masak there. :)
stmuk does the --verbose flag on t/harness still work?
for roast
arnsholt I guess it's the comple-time vs. run-time distinction all over again 13:22
masak arnsholt: yes, but -- macros are runtime inside of compile time 13:22
arnsholt Yeah. And you need to keep track of the distinction between run-time for the *macro* code, and run-time of the code generated by the macro code 13:22
masak aye 13:23
masak and the things that pass into the macro "freeze" and are no longer values, they're trees 13:23
arnsholt Yup 13:25
masak and the quasi is a run-time value representing a compile-time AST that you put back into the program so it can run later
arnsholt Whee, fun! =)
dalek c: c16e4d8 | smls++ | doc/Type/List.pod:
expand &reduce documentation
 13:27
p: 1358834 | jnthn++ | src/vm/moar/QAST/QASTRegexCompilerMAST.nqp:
Remove special case not needed post-NFG.

This will regress a couple of tests until we bump to an NFG-supporting MoarVM.
p: 1a0ead0 | jnthn++ | / (12 files):
Bump Moar for \r\n as 1 grapheme; rebootstrap.
p: 568f06a | jnthn++ | src/QRegex/P6Regex/Actions.nqp:
Make \r\n in a regex mean the grapheme.
jnthn lunch & 13:30
mrf jnthn++ # You have just saved me writing a whole new grammar with that change
dalek kudo/nom: d02fede | lizmat++ | src/core/ (2 files):
Handle Array methods on List/Range similarly
 13:32
dalek ast: ae478c6 | lizmat++ | S02-types/ (2 files):
Handle testing Array methods on List/Range similarly
 13:33
cygx jnthn: I figured out that NativeCall was to blame for my inability to export a multi trait_mod:<is> - cf #126547 13:35
synbot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=126547
cygx going things with NativeCall also makes the bytecode sizes of precompiled modules blow up
*doing
abraxxa I want to store native typed values in a hash for passing to a NativeCall function but fail 13:40
cygx jnthn: including NativeCall to add a Pointer parameter declaration added 1.7m to the bytecode size...
abraxxa my long $valuep = $v; %!in_bind_for{$k} = $valuep; fails with Native call expected argument that references a native integer, but got P6opaque
the issue I'm facing is that Perl6 seems to garbage collect a variable I'm passing by reference to a C function 13:41
lizmat abraxxa: if you could golf that, that would be much appreciated
there's an nqp::op to force a gc, I forget what it was, maybe nqp::force_gc ? 13:42
abraxxa lizmat: ???
lizmat m: use nqp; nqp::force_gc # indeed, that's it
camelia ( no output )
abraxxa golf? Dieselgate? 13:43
lizmat en.wikipedia.org/wiki/Code_golf
llfourn abraxxa: reproduce it in a very compact way :)
abraxxa llfourn: i see, thanks 13:44
don't think that's easily possible as this involves Oracle OCI
lizmat perhaps the use of nqp::force_gc might help you to reproduce the problem better ?
abraxxa if I do %!in_bind_for{$k} := $valuep; the NativeCall error goes away but the problem persists
I still don't get the difference between = and := 13:45
lizmat m: my $a := 42; $a = 666 13:45
camelia rakudo-moar 273e89: OUTPUT«Cannot assign to an immutable value␤ in block <unit> at /tmp/CraQ1CHvD4:1␤␤»
lizmat m: my $a = 42; $a = 666
camelia ( no output )
llfourn m: my $a = 1; my $b := $a; $b = 42; say $a; 13:46
camelia rakudo-moar 273e89: OUTPUT«42␤»
lizmat m: my $a = 42; my $b := $a; dd :$a,:$b; $b = 666; dd :$a,:$b
camelia rakudo-moar 273e89: OUTPUT«Int a = 42␤Int b = 42␤Int a = 666␤Int b = 666␤»
lizmat := binds (aliases), = just assigns
m: my $a = 42; my $b = $a; dd :$a,:$b; $b = 666; dd :$a,:$b
camelia rakudo-moar 273e89: OUTPUT«Int a = 42␤Int b = 42␤Int a = 42␤Int b = 666␤»
abraxxa why does passing %!in_bind_for{$k} to the NativeCall function throw this error? 13:47
m: my %hash; my $k = "foo"; %hash{$k} = 3; say %hash{$k}.WHAT;
camelia rakudo-moar 273e89: OUTPUT«(Int)␤»
abraxxa m: my $v = 3; my long $valuep = $v; my %hash; my $k = "foo"; %hash{$k} = $valuep; say %hash{$k}.WHAT; 13:48
camelia rakudo-moar 273e89: OUTPUT«5===SORRY!5===␤Type 'long' is not declared␤at /tmp/XIpaN4h4X0:1␤------> 3my $v = 3; my long7⏏5 $valuep = $v; my %hash; my $k = "foo"; ␤Malformed my␤at /tmp/XIpaN4h4X0:1␤------> 3my $v = 3; my7⏏5 long $valuep = $v; my %hash; my $k…»
abraxxa m: use NativeCall; my $v = 3; my long $valuep = $v; my %hash; my $k = "foo"; %hash{$k} = $valuep; say %hash{$k}.WHAT;
camelia rakudo-moar 273e89: OUTPUT«(Int)␤»
abraxxa why is it Int and not long? 13:49
lizmat I think that's because Perl 6 hashes can only have P6 objects, not natives
m: my %h; my int $a = 42; %h<a> = $a; dd %h<a> 13:50
camelia rakudo-moar 273e89: OUTPUT«Int $var = 42␤»
lizmat so the native got upgraded to a HLL Int
abraxxa what do you propose to do to pass native typed vars to the NativeCall function and prevent gc to happen before the C library access the memory region by pointer? 13:51
lizmat well, the question is really I think: why do you want to use a hash ? 13:52
abraxxa my idea was to have a hash of in and out binds in the StatementHandle object so they don't get gc-ed
and as the placeholders are keyed a hash is better suited than an array 13:53
abraxxa i won't need to access the in binds later but need it for the out binds (data returned by a query) 13:54
lizmat I think you will get a better impedance match if you would define constants for indexes, and then use a native array 13:54
brb 13:55
[Coke] java build failed. 14:02
make 14:03
./perl6-j --target=jar --output=lib/Test.pm.jar -Ilib lib/Test.pm
java.lang.NullPointerException
in (gen/jvm/CORE.setting)
in (gen/jvm/ModuleLoader.nqp:380)
...
that's with a fresh build of nqp as well.
Ven_ arnsholt: part of it is because you're just dealing with a self-evaluating, non quoted value in your gist. Also, you're emulating "state", which wasn't the point of my ((let)) example 14:11
masak: using "my $foo; macro xx { $foo }" makes me think there's yet another time mixup
ZoffixW How would I convert a hex string to base10 int?
m: say "ff".base(10)
camelia rakudo-moar 273e89: OUTPUT«Method 'base' not found for invocant of class 'Str'␤ in block <unit> at /tmp/GBkYRd3P8L:1␤␤»
masak Ven_: not really. 14:12
Ven_ oh, you said it later
dalek href="https://modules.perl6.org:">modules.perl6.org: 9a665e2 | azawawi++ | web/build- (6 files):
chmod -x build-*.png
ShimmerFairy m: say :16<FF>
camelia rakudo-moar 273e89: OUTPUT«255␤»
Ven_ yes, because the macro is run at parse time, and $foo doesn't have the value then
masak Ven_: it's difference between static structure and runtime frames.
Ven_ how so? $foo's assignment is run later than the macro is, ain't it?
ZoffixW And actually, what I'm doing is trying to unpack hex string, say, "ffbbcc" into R G B ints, but I'm having a real trouble with Perl types getting in the way
masak Ven_: variables very much have values at parse time. in the static lexpads.
Ven_: that's how constants work.
Ven_: that's how BEGIN-time works. 14:13
Ven_ constants, yes, variables?
m: my $a = 5; BEGIN say $a;
camelia rakudo-moar 273e89: OUTPUT«(Any)␤»
ZoffixW ShimmerFairy, but how do I do that with variables?
Ven_ which is what I'd expect.
ZoffixW m: my $str = "ffbbcc"; my ( $r, $g, $b ) = map ( :16<$_> ), $str ~~ /(..)(..)(..)/; say [ $r, $g, $b ].perl
camelia rakudo-moar 273e89: OUTPUT«5===SORRY!5=== Error while compiling /tmp/lNBPz9adLs␤Malformed radix number␤at /tmp/lNBPz9adLs:1␤------> 3"ffbbcc"; my ( $r, $g, $b ) = map ( :16<7⏏5$_> ), $str ~~ /(..)(..)(..)/; say [ $r,␤ expecting any of:␤ number in radix …»
ShimmerFairy m: my $a = "2A"; say :16($a)
camelia rakudo-moar 273e89: OUTPUT«42␤»
ShimmerFairy ^^^ that's how :)
ZoffixW ShimmerFairy++ thanks 14:14
masak Ven_: me too.
masak Ven_: I did fix that late assignment later, using a BEGIN. 14:14
Ven_ masak: which is exactly what I pointed at around 15:12 (well, definitely not *your* $localtime) 14:15
Ven_ isn't really sure where masak++ is right now :) 14:15
perl6sum how to get autocompletion in perl 6 in interactive session 14:17
I am running rakudo star latest on windows 10
i get prompt when I type perl6 in cmd prompt
stmuk perl6sum: doc.perl6.org/language/faq#How_can_...REPL%29%3F 14:18
actually I don't think it works for autocompletion just yet, although there was an experimental attempt a few months back 14:19
stmuk just history 14:19
perl6sum stmuk I want autocompletion
ZoffixW Is there a nice way to get a list of matches of a regexp using a single statement? I'm having trouble with types: 14:20
m: my $str = "ffbbcc"; my ( $r, $g, $b ) = map ( :16($_) ), $str.match(/(..)(..)(..)/).list; say [ $r, $g, $b ].perl
camelia rakudo-moar 273e89: OUTPUT«Cannot call UNBASE(Int, Any); none of these signatures match:␤ (Int:D $base, Any:D $num)␤ (Int:D $base, Str:D $str)␤ in block <unit> at /tmp/DhQdHcjwKs:1␤␤»
ZoffixW Why am I getting "Any" there?
moritz ZoffixW: map { :16($_) } 14:21
ZoffixW: otherwise the :16($_) is evaluated before it's passed to map
ZoffixW moritz++ thanks 14:22
masak Ven_: I'm in UTC+08:00
'night, #perl6
ZoffixW Night 14:23
JimmyZ masak is in china? :)
jnthn cygx: I guess NativeCall is probably doing something that's bad for pre-comp. 14:23
Ven_ pdcawley: I'd also agree quasi{} shouldn't do that, but we've established that the current quasi{} isn't a quasi 14:26
jnthn m: my $str = "ffbbcc"; my ( $r, $g, $b ) = map { :16($_) }, $str.comb(/../); .say for $r, $g, $b 14:26
camelia rakudo-moar 273e89: OUTPUT«255␤187␤204␤»
lizmat wonder if $str.comb(2) would be a lot faster there 14:27
commute to Amsterdam.pm meeting&
ZoffixW Thanks. I'll read up on .comb 14:30
dalek kudo/nom: d598af3 | jnthn++ | src/core/Str.pm:
Prepare Str.lines for \r\n as 1 grapheme.

Leave it able to support JVM too.		 14:32
href="https://modules.perl6.org:">modules.perl6.org: 426a267 | azawawi++ | Build.PL:
IO::Socket::SSL 1.94+ is needed for Mojo::UserAgent
 14:37
ZoffixW 0.o we have a Mojo::UserAgent now 14:38
Oh. I thought it was a P6 module :P 14:39
dalek kudo/nom: 3e6bc1e | (Steve Mynott)++ | src/core/Kernel.pm:
fix S02-magicals/KERNEL.t on DragonFly BSD by upper casing signal names
 14:42
kudo/nom: a94389e | jnthn++ | src/core/Kernel.pm:
Merge pull request #571 from stmuk/nom

fix S02-magicals/KERNEL.t on DragonFly BSD by upper casing signal names
azawawi ZoffixW: hi 14:42
ZoffixW azawawi, hey 14:43
azawawi ZoffixW: im trying to edit module.perl6.org repo to add watchers_count and stargazers_count
ZoffixW: e.g. api.github.com/repos/rakudo/rakudo
ZoffixW: if we have those two columns, we can sort by them. What do you think? 14:44
ZoffixW: i.e. to find the most popular or used modules
ZoffixW: open_issues_count is useful also
dalek ast: fa4431a | jnthn++ | S16-filehandles/io_in_while_loops.t:
Update test to cope with Windows line ending too.
 14:46
ZoffixW azawawi, OK.
azawawi, that info should be in this hashref: github.com/perl6/modules.perl6.org...ub.pm#L117
So, $repo->{stargazers_count} etc 14:47
azawawi, and yeah, I think it would be good to have more info.
My only concern is we're making that table wider and wider.
dalek ast: 125cf07 | jnthn++ | / (17 files):
Fix Windows line ending workarounds.

The completion of the NFG work making \r\n a single grapheme, per the Unicode spec, means that the approaches these tests took to work on Windows became bogus. Note that we'll probably have to take a pass through these all again in the near future, when \n in a string will become \r\n on Windows by default (hopefully clearing up a lot of this cruft).		 14:48
ZoffixW Or rather, not wider, but tighter :P 'cause all of the extra info has to fit it on one row.
azawawi ZoffixW: hmmm, top 10 lists?
azawawi ZoffixW: e.g. metacpan.org/favorite/leaderboard 14:49
ZoffixW Woohoo Mojolicious is #1 :) Before even Perl :D
azawawi ZoffixW: Mojolicious is the new Perl 5 :) 14:50
ZoffixW azawawi, yeah, I think having something like this—a separate page—might be more useful. 14:50
dalek ast: 2d5a420 | (Steve Mynott)++ | S06-traits/misc.t:
correct typo
isBEKaml hoelzro: ping 14:51
azawawi ZoffixW: true since why do you want to sort them repos with no stargazers...
ZoffixW: /sort them/sort/
isBEKaml hoelzro: have you had enough success with vim-perl syntax highlighting for perl6?
hoelzro: I ask since I see this weirdness on vim right now: github.com/rakudo/rakudo/blob/3e6b...nel.pm#L13 14:52
dalek kudo/nom: 75b2451 | jnthn++ | tools/build/NQP_REVISION:
Bump NQP_REVISION for \r\n as 1 grapheme.

This may have some ecosystem fallout for code that has been written assuming that "\r\n".chars will be 2. Thankfully, since .chomp is used in most such cases, this should be relatively little.
isBEKaml hoelzro: It seems to treat the '|' as scope delimiters for s// :-)
hoelzro: that said, this should really go through: github.com/vim-perl/vim-perl/issues/194 14:53
azawawi ZoffixW: also badges column is redundant in the sense why do i need to see success badges, show me failed audits
PerlJam ZoffixW: as long as your making changes ... why does travis say "build passing" or "build failing" etc? The word "build" seems superfluous. Maybe remove that to gain a tiny bit more horizontal space.
s/your/you're/ 14:54
dalek ast: 302de05 | jnthn++ | S15-nfg/grapheme-break.t:
Unfudge test for \r\n as a single grapheme.
isBEKaml .tell hoelzro sorry, I left too many messages for you in the backlog. Please check messages around this timestamp :-)
yoleaux isBEKaml: I'll pass your message to hoelzro.
jnthn That's RT #125927 ticked off :)
isBEKaml jnthn: I was just looking at that ticket :-)
synbot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=125927 14:55
nine jnthn: congratulations!
hoelzro isBEKaml: I can look at this more in depth later, but first - are you using the latest version of syntax/perl6.vim from vim-perl? 14:56
yoleaux 14:54Z <isBEKaml> hoelzro: sorry, I left too many messages for you in the backlog. Please check messages around this timestamp :-)
isBEKaml hoelzro: yes
hoelzro I just tried it with the latest and it works for me =/
ZoffixW PerlJam, I think it was azawawi who was making changes :) But that's a good point. I'll open an Issue
hoelzro could you send me a screenshot?
hoelzro I have to get going to work, but I can have a quick look when I get there 14:56
PerlJam zostay: sorry :) 14:57
er, ZoffixW even
hoelzro wonders if one of the channel bots should detect github.com/rakudo/rakudo/blob/mast...nel.pm#L13 and rewrite it as github.com/rakudo/rakudo/blob/3e6b...nel.pm#L13
isBEKaml hoelzro: send you an email or upload somewhere? 14:58
ZoffixW azawawi, yeah, some of the badges might not be needed any more. I opened an issue for the topic: github.com/perl6/modules.perl6.org/issues/19
[Coke] jnthn++ 14:59
azawawi ZoffixW: i verified the hash and yes the information is already there... thanks :)
ZoffixW \o/ 15:00
isBEKaml hoelzro: d'oh - never mind.
hoelzro: I did :se ft=perl6 and it all looks fine now. The .pm confusion strikes again! :-| 15:01
isBEKaml hmm, my vim didn't read those editor markers for perl6... 15:04
azawawi is waiting for build-project-list.pl to finish 15:09
moritz that thing could use some caching, IMHO
azawawi moritz: yup it is way too slow 15:10
178/432...
ZoffixW azawawi, if you're just doing testing, it accepts a --limit option 15:11
to build just --limit X dists instead of all of them
azawawi wants to write a proper README.md after finishing :) 15:12
dalek rl6-roast-data: 4d0ec70 | coke++ | / (9 files):
today (automated commit)
[Coke] 37232 failing tests in rakudo-jvm today. :|
[Coke] starts a slow bisect.
jnthn :( 15:13
[Coke]++
colomon [Coke]: wouldn’t it be more time efficient if you started a fast bisect? ;) 15:16
stmuk renice -20 :) 15:17
[Coke] colomon: :P 15:19
[Coke] oneof the intermediate builds fails with: 15:25
Stage jast : Error while compiling op setinputlineseps (source text: "nqp::setinputlineseps(handle, endings)"), no registered operation handler
RabidGravy wahay! 15:26
yoleaux 11:09Z <azawawi> RabidGravy: to get the github stars on a certain github repo, get api.github.com/repos/rakudo/rakudo which returns a JSON response and use "stargazers_count" key :)
11:19Z <tadzik> RabidGravy: I just did, thanks a lot :)
RabidGravy azawawi++ # most excellent 15:27
pyrimidine jnthn++ # fix for RT #122971 15:29
synbot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=122971
RabidGravy tadzik++ # lets me do this 15:32
dalek osystem: 02d2303 | RabidGravy++ | META.list:
Fix branch on Perl6::Maven
osystem: 251b4d1 | RabidGravy++ | /:
Merge branch 'master' of github.com:perl6/ecosystem
osystem: 0a4e339 | RabidGravy++ | META.list:
Merge branch 'master' of github.com:perl6/ecosystem
osystem: ef1f0f1 | RabidGravy++ | META.list:
Add the JSON::Class

  github.com/jonathanstowe/JSON-Class
RabidGravy why didn't I take my own advice and rebase ;-) 15:33
mayuresh is [coke] around? 15:33
[Coke] Hio
RabidGravy there
mayuresh [coke] did your problem about that processing hanging get resolved? 15:34
i meant, process hanging. :)
[Coke] which process?
mayuresh weren't you the one running some tests under mac os x? 15:35
umnn, maybe it's just me being a jackass. ;-)
sprocket how would i be able to define different things for different architectures?
yoleaux 10:02Z <jnthn> sprocket: Define both structs, and then write constant TheStruct = $*KERNEL eq 'darwin' ?? TheOSXStruct !! TheLinuxStruct;
sprocket yoleaux: thanks :)
[Coke] I do lots of stuff. That's not enough of a callback for me to remember the specific issue, sorry. 15:36
mayuresh [coke], are you the same person running mac os x, after having worked under windows for years!
ilmari m: $*KERNEL.version 15:37
camelia ( no output )
ilmari m: say $*KERNEL.version
camelia rakudo-moar 273e89: OUTPUT«v3.16.6.2.default␤»
[Coke] no, I'm not a windows person.
mayuresh ah, so i am indeed being a jackass. ;)
i just can't remember who it was that got stuck with the process hanging issue under mac os x and couldn't find tools to inspect the problem. 15:38
mayuresh i think i might have some tips for that person. 15:38
mayuresh actually conversed with a bunch of mac os x developers 15:38
just to get that piece of information.
and now it's all useless. :(
just because i can't remember who it was i intended to help. 15:39
gah.
this is utterly disgusting on my part. 15:40
i should simply leave now.
[Coke] oh. Yes, I was trying to use that tool. I'm not a windows person, though. 15:40
but that was days ago. I have no idea how to reproduce the failing I was seeing.
hoelzro isBEKaml: which editor markers? 15:41
[Coke] please write that information down in a gist somewhere so I can use it in 3 weeks when I need it again. :)
*the failure
isBEKaml hoelzro: those # vim: ft=perl6... 15:42
hoelzro isBEKaml: do you have modeline set to true?
nine [Coke]: mayuresh already left 15:45
ilmari huh, the percentage bar at the top of github.com/perl6/nqp claims 66.8% parrot, but when you click it it finds nothing 15:46
[Coke] well, I hope they review next time. whoops.
isBEKaml hoelzro: ah, that's it. I have now set it in my .vimrc -- that's what happens when you move between machines not knowing which setting it was that worked on the other machine :-)
moritz ilmari: probably due to the .pir bootstrapping files
hoelzro isBEKaml: there's also a contrib/ file called perl11.vim that tries to detect which type the file is 15:47
isBEKaml hoelzro: but that would surely fail with a file like Kernel.pm?
hoelzro looks
ilmari moritz: I was more boggling at the actual search disagreeing with it
hoelzro isBEKaml: it should work 15:48
unless it changes =)
ilmari github.com/perl6/nqp/search?l=parrot
hoelzro it uses the heuristics set out in S02 to detect a Perl 6 file
ie. if the first non-comment line has a class/role/etc declaration
or if 'use v6' is found
s/non-comment/non-comment, non-empty/ 15:49
isBEKaml hoelzro: grr, I did the install the old way of doing 'make install' after extracting the tarball. perl11.vim is in ftdetect which is not part of the install 15:51
hoelzro you should be able to just drop it in under ftdetect/; I don't think it's meant to be installed by default 15:52
isBEKaml hoelzro: yeah, it works. but no matter, I got my modeline back =) 15:53
hoelzro \o/ 15:54
azawawi ZoffixW: nearly there :)
ZoffixW: generating image :)
isBEKaml thanks, hoelzro! 15:55
hoelzro happy to help!
dalek kudo/nom: 2c43078 | jnthn++ | src/core/core_prologue.pm:
Introduce a $?NL which represents the platform \n.
 15:58
kudo/nom: 0396397 | jnthn++ | / (3 files):
Add newline pragmatic module to control $?NL.
liztormato jnthn: why a separate file for that new pragma? Why not use the new approach for pragmas? 16:04
jnthn liztormato: What new approach?
liztormato: The %*PRAGMAS thing is there as a stopgap until we can write all pragmas in a "first class" way, not an ideal.
liztormato In World.nqp: do_pragma 16:05
jnthn Then we have to bulk up the compiler with it
There's no reason to do that; this one is trivially implementable as a normal pragmatic module. 16:06
liztormato Well having it as an external means all sorts of repo overhead
jnthn That's fine; I don't expect this to be commonly used :)
jnthn And I hope we can get that overhead down some. 16:06
The bigger issue with this is that now \r\n has a surprising meaning on Windows... 16:07
isBEKaml m: $*DISTRO.^methods.say # are these <anon>s meant to be there? 16:08
camelia rakudo-moar 273e89: OUTPUT«(BUILD cur-sep Str gist <anon> <anon> <anon> <anon> <anon> <anon> <anon> <anon>)␤»
jnthn Well, .^methods isn't meant to hide things 16:09
dalek kudo/nom: e98b114 | (Stefan Seifert)++ | src/Perl6/Actions.nqp:
Fix optimized for range loop messing up lexicals

The optimized code is created from the original op tree, not what p6for generates. So creating the closure for $block in p6for is not enough. Add the block_closure to the original op tree instead, so we get it in the optimized code, too.		 16:10
kudo/nom: f59b910 | timotimo++ | src/Perl6/Optimizer.nqp:
Bring back optimization of for $range loops

Only optimize for Range to while when looping in sink context. Only optimize unlabled for loops. Otherwise we miss the control exception handlers.
isBEKaml jnthn: so what would you recommend I use instead?
nine .tell timotimo it is done!
yoleaux nine: I'll pass your message to timotimo.
iv how does one have anonymous methods? 16:10
nine Bring forth those benchmarks!
PerlJam wonders if there's a :named to go along with :local and :global for .^methods
jnthn isBEKaml: What do you want to do?
isBEKaml jnthn: nothing! :P 16:11
timotimo nine: hooray! nine++
yoleaux 16:10Z <nine> timotimo: it is done!
isBEKaml jnthn: well, I'm mostly curious after being away from perl6 :-)
mr_ron m: say Nil.Str # docs.perl6.org/type/Nil#method_Str
camelia rakudo-moar 273e89: OUTPUT«Use of Nil in string context in block <unit> at /tmp/Q_n_kNl3GR:1␤␤»
jnthn isBEKaml: You may find .^method_table of interest
nine++ 16:12
timotimo nine: there's a commit there that's allegedly mine, but i don't remember making it :P 16:12
sprocket so the design docs call for $*GID / $*UID / $*EGID / $*EUID to be defined, but attempting to use them is giving me an error
is that something that’s coming eventually and just hasn’t been implemented yet?
nine timotimo: it's your WIP commit with two fixes added
timotimo sprocket: probably
llfourn sprocket: Not Yet Implemented :)
sprocket :) ok, i can work around that, thanks! 16:13
isBEKaml jnthn: huh, wow.
llfourn m: for ^10 { my $answer = 42; try { let $answer = 84; die "resetting $answer" if Bool.pick } ; print $answer }
camelia rakudo-moar 273e89: OUTPUT«42424242424242424242»
timotimo nine: also with commit message changes attached, i see ;)
llfourn Am I very unlucky?
mr_ron Did I misunderstand the docs or am I missing something else? 16:14
jnthn m: say Bool.pick
camelia rakudo-moar 273e89: OUTPUT«True␤»
jnthn m: say Bool.pick
camelia rakudo-moar 273e89: OUTPUT«False␤»
timotimo m: say Nil.Str.perl
camelia rakudo-moar 273e89: OUTPUT«Use of Nil in string context in block <unit> at /tmp/h6Oa4wgA79:1␤""␤»
timotimo mr_ron: it's exactly as documented. except, it also throws a warning :) 16:15
llfourn m: for ^10 { my $answer = 42; try { let $answer = 84; my $b = Bool.pick; say $b; die "resetting $answer" if $b } ; print $answer }
camelia rakudo-moar 273e89: OUTPUT«True␤42True␤42True␤42False␤42False␤42False␤42True␤42True␤42True␤42True␤42»
llfourn eh
timotimo :\
mr_ron Should I file an RT? I didn't see one for this. 16:16
llfourn m: for ^10 { my $answer = 42; try { let $answer = 84; my $b = Bool.pick; say $b; if $b { die "resetting $answer" } } ; print $answer }
camelia rakudo-moar 273e89: OUTPUT«True␤42True␤42True␤42True␤42True␤42True␤42True␤42False␤42True␤42False␤42»
timotimo perhaps the try { } around it makes the exception nonfatal and nonfailure-y enough to prevent let from resetting? 16:17
llfourn m: my $answer = 42; try { let $answer = 84; die} ; say $answer; 16:18
camelia rakudo-moar 273e89: OUTPUT«42␤»
llfourn timotimo: maybe...
m: my $answer = 42; try { let $answer = 84; } ; say $answer; 16:19
camelia rakudo-moar 273e89: OUTPUT«84␤»
jnthn .tell TimToady Turns out making \n magical mostly trips up over every \r\n usage out there. Should we special-case \r\n to mean The Right Thing also?
yoleaux jnthn: I'll pass your message to TimToady.
llfourn timotimo: but let *does* reset it 16:20
timotimo huh.
dalek kudo/magical-n: 95f66bb | jnthn++ | src/ (5 files):
Work-in-progress on making \n platform magical.
llfourn m: for ^10 { my $answer = 42; try { let $answer = 84; my $b = Bool.pick; say $b; if $b { die "resetting $answer" } } ; print $answer }
camelia rakudo-moar 273e89: OUTPUT«False␤42False␤42True␤42False␤42False␤42False␤42False␤42False␤42True␤42False␤42»
llfourn 42 means it was reset
mr_ron timotimo: sorry I misunderstood your answer, I think I understand now 16:21
timotimo isn't a change made by let supposed to survive leaving the block, though? 16:22
PerlJam m: for ^10 { my $a = 42; try { let $a = 84; fail if Bool.pick; 1 } ; say $a; }
camelia rakudo-moar 273e89: OUTPUT«84␤42␤84␤84␤42␤42␤84␤42␤42␤42␤»
llfourn timotimo: yes, so it's dieing every time
PerlJam llfourn: looks like a bug to me.
timotimo um, should "fail" in there really work?
fail is a sort of return, after all
m: for ^10 { say "test"; return }
camelia rakudo-moar 273e89: OUTPUT«test␤»
PerlJam m: for ^10 { my $a = 42; try { let $a = 84; die if Bool.pick; 1 } ; say $a; }
camelia rakudo-moar 273e89: OUTPUT«42␤84␤42␤42␤84␤42␤42␤84␤84␤42␤»
timotimo oh! 16:23
Skarsnik Hi there, where I can find the source code of the eval bot? x)
timotimo Skarsnik: probably perl6/evalbot or perl6/camelia
the else-block of that postfix-if returns a "interpreted as failure" value
PerlJam: good catch!
llfourn timotimo: eh the block has to return true?
llfourn m: for ^10 { my $answer = 42; try { let $answer = 84; my $b = Bool.pick; say $b; if $b { die "resetting $answer" }; 1 } ; print $answer } 16:24
camelia rakudo-moar 273e89: OUTPUT«False␤84False␤84False␤84True␤42True␤42True␤42False␤84False␤84False␤84False␤84»
timotimo llfourn: has to return a defined value, iirc
timotimo m: for ^10 { my $a = 42; try { let $a = 84; Bool.pick ?? 1 !! Any } ; say $a; } 16:25
camelia rakudo-moar 273e89: OUTPUT«84␤42␤84␤42␤84␤42␤84␤84␤42␤42␤»
llfourn Ah well that explains it. Thanks guys.
Skarsnik Also. I was wondering is there a way to have something similar to how package work in java. Like I define stuff in a MyModule package/namespace and I have access to then in a MyModule::OtherStuff. But not in a code that just do use MyModule. 16:26
[Coke] I didn't see an answer to the Nil.Str question - no, working as intended. 16:27
m: my $a = Nil.Str; say "alive"; say ":{$a}:" 16:28
camelia rakudo-moar 273e89: OUTPUT«Use of Nil in string context in block <unit> at /tmp/c1Ijzh3fs4:1␤alive␤::␤»
garu Skarsnik: you mean like use lib?
[Coke] ^^ it's warning you, but it's still returning the empty string.
timotimo [Coke]: i answered, but i'm not sure it "arrived" 16:29
Skarsnik garu, use lib?
[Coke] timotimo++ 16:31
dalek ast: 0a388d1 | jnthn++ | S16-io/handles-between-threads.t:
Test for file handle use between threads.
 16:32
PerlJam llfourn: maybe that should be part of a best practices or a FAQ or something ... if you're going to use "let" make *sure* you handle both success/failure in the block. 16:33
llfourn PerlJam: I'm just writing a quick doc for let now. Will be sure to mention it :)
PerlJam llfourn: your example was non-intuitive precisely because you were only concerned with the failure side and didn't specify success adequately
RabidGravy don't say "best practices" 16:34
llfourn PerlJam: I thought failure meant exception thrown not block return value
[Coke] are we not doing jenkins on the jvm builds?
mrf Anyone know how I force update of nqp in rakudobrew. (along with anything that would have to be updated as a result) 16:35
PerlJam llfourn: S04:1772 16:36
synbot6 Link: design.perl6.org/S04.html#line_1772
llfourn PerlJam: cheers.
jdv79 is someone starting a p6bp? 16:43
RabidGravy if they do I leave
jdv79 haha 16:44
PerlJam RabidGravy best practice: don't start a Perl 6 Best Practices
RabidGravy the people who have the loudest voices and the most dubious opinions form the "best practice" 16:45
PerlJam++ # precisely 16:46
TimToady m: say &[Z+].name 16:47
yoleaux 16:19Z <jnthn> TimToady: Turns out making \n magical mostly trips up over every \r\n usage out there. Should we special-case \r\n to mean The Right Thing also?
camelia rakudo-moar 273e89: OUTPUT«␤»
stmuk "there is more than one way to do but only my way is Right"
RabidGravy stmuk++
TimToady m: say &[Z+].name 16:48
camelia rakudo-moar 273e89: OUTPUT«␤»
PerlJam RabidGravy: are you sure you're not reacting to the evangelization of PBP?
jdv79 im imterested in what a p6 tidier will do
TimToady camelia is about 24 hours out-of-date now
stmuk I think most p5 progammers agree with about 2/3 of PBP but not which 2/3
PerlJam RabidGravy: I mean, Damian wasn't really loud and his opinions weren't dubious, it was just when people decided to take his word as gospel that things went sideways
RabidGravy partly that and partly to the newspeak of "best practices" wherever it is found 16:49
yes, blindly taking things as gospel
see also "Modern Perl"
timotimo did 2015.10 already have GLR in it?
nine timotimo: yes 16:50
timotimo OK
nine: i'm going to compare the benchmark results to 2015.09
jdv79 well, modern perl seemed a bit better thought out and based in reality
imo
RabidGravy PerlJam, totally agree fwiw
pmurias some of the opinions in PBP were dubious, like the inside out objects 16:51
dalek Heuristic branch merge: pushed 89 commits to rakudo/curli by niner 16:52
nine note to self: first push the merge, then push the new commit
PerlJam pmurias: How were inside out objects dubious exactly?
RabidGravy I'm only going to say this once but "Modern Perl" (in my mind) reduces Perl to a scripting language for a small set of cabal mandated frameworks
jdv79 its definitely not a best practice
jdv79 io objs that is 16:53
tony-o m: "\r\n".ord.say; 16:54
camelia rakudo-moar 273e89: OUTPUT«13␤»
stmuk inside objects were in fashion for about 6 months in 2005
grondilu is everything still inline for Xmas? 16:55
tony-o i think christmas is still going to happen this year grondilu
jdv79 amd how is a fad a best practice in any sense?
RabidGravy well we've worked out where we're going to buy a christmas tree this year
TimToady tony-o: that's still yesterday's camelia 16:56
.seen moritz
yoleaux I saw moritz 15:46Z in #perl6: <moritz> ilmari: probably due to the .pir bootstrapping files
sprocket hey all - i’ve done a first cut at a POSIX module containing some of the functions I needed for another project (ie. setuid, setgid, etc) and have pushed it out to github. how would i go about getting that added to the p6 module ecosystem?
xenu pbp spawned perl critic, which is pretty useful tool, but can be horribly misused. unfortunately usually that's the case 16:57
ugexe im getting strange behavior reguarding \r\n. the ord for it is changing depending on what is in the string 16:58
gist.github.com/ugexe/09cfce18f3ece8b6e83b 16:59
sometimes \r\n is converted as '13', sometimes '4294967309'
if i remove a line of text from that example, sometimes the 4294967309 changes to a '13' 17:00
RabidGravy sprocket, fork, clone the github.com/perl6/ecosystem add your META.info to the META.list (i.e the raw entry), make PR
TimToady ord should always report 13, and not the synthetic
jdv79 whats a pjpeg?
TimToady (and the synthetic should probably be reported as negative, not huge positive)
stmuk hahaha "make sandwich" 17:01
RabidGravy sprocket, if you're on a roll and making several modules someone may add you to the ecosystem project 17:02
I don't have that force
sprocket RabidGravy: thanks!
[Coke] onder if the synthetic is underflowing? 17:03
TimToady .tell jnthn either we accept \r\n as the same, or we blow up and say "you gotta say 'use newline :lf' for \r\n to work"
yoleaux TimToady: I'll pass your message to jnthn.
TimToady looks like a 32-bit number, so maybe 32 vs 64-bit problem? 17:04
tony-o dunno, check this out 17:07
tonyo@imac ~$ perl6 -e '.ord.say for "\r\n".comb;'
13
timotimo nine: t.h8.lv/p6bench/2015-11-02-rakudo_m...t_glr.html
jnthn If synthetics are somehow leaking, that's a bug 17:08
yoleaux 17:03Z <TimToady> jnthn: either we accept \r\n as the same, or we blow up and say "you gotta say 'use newline :lf' for \r\n to work"
jnthn TimToady: I think we'll go with "make it work as it did"
TimToady: Least surprise, for msot definitions
Note that if you .ord a synthetic you get the ord of the base char. 17:09
dalek c: 89362d6 | (Lloyd Fournier)++ | doc/Language/ (2 files):
let example (duplicated)
 17:10
tony-o wonder how that works with trans 17:10
llfourn PerlJam: ^^ for your reviewing pleasure :)
tony-o ah, doesn't transform at all with trans 17:11
nine timotimo: the results for for loops look awesome! But why would while have become so much slower?
timotimo good questoin 17:11
... question
i'm currently working on #job, so can't delve too deep
perhaps you can figure out something about [+] 1 .. SCALE, though?
that's become 14x slower 17:12
nine Well, I'm off to a Calexico concert now. 17:13
timotimo oh
enjoy! :)
nine Thanks :)
PerlJam llfourn: if Bool.pick returns True, the die is executed and $answer will be 42. Your text has it backward. 17:16
llfourn PerlJam: ah thanks
PerlJam llfourn: I find the sentence "Like the temp prefix, but only restores the value if the block exits unsuccessfully" tedious to parse 17:17
well ... and understadn
llfourn PerlJam: Agreed
sprocket RabidGravy: the ecosystem META.list has the urls as “raw.githubusercontent.com/….” - but i’m not able to view my META.info with the url yet 17:18
RabidGravy: will that appear eventually?
PerlJam llfourn: sorry I've only got criticism, my brain is occupied elsewhere enough that I don't have suggestions for better wording. 17:18
ugexe how do i construct a string with a \r\n that matches a regex with \r\n but not \n? 17:19
RabidGravy sprocket, go to the file in github, there's a "raw" button to the top right, click that copy the URI
sprocket RabidGravy: ah! thank you
TimToady a Str cannot contain \r\n except as a synthetic
RabidGravy top middle right, that is 17:20
llfourn PerlJam: nw looks good now thanks. It's never the last commit!
TimToady a Buf.new(13,10) is a CRLF
RabidGravy if it isn't we ain't having no web clients ;0)
ugexe right, but in this case i want to write a test for a http header. its not very convienient to store the test data as a Buf instead of as a human readable string 17:21
dalek c: 9fc68ca | (Lloyd Fournier)++ | doc/Language/ (2 files):
fix let example and clean up first sentence
ilmari huh, doesn't Buf have a .join method? 17:23
otherwise you could have done ('Foo: bar','Baz: zot').map(*.encode('ascii')).join(Buf.new(13, 10))
dalek osystem: 926458a | cspencer++ | META.list:
Added Perl 6 POSIX module to ecosystem.
 17:24
osystem: 1ffbc3d | (Zoffix Znet)++ | META.list:
Merge pull request #86 from cspencer/master

Added Perl 6 POSIX module to ecosystem: github.com/cspencer/perl6-posix
TimToady we probably need better ways to write Buf/Uni literals
RabidGravy sprocket++ # I'll look later, but damn good stuff ;-) 17:25
jnthn \r\n in a regex won't match a \n 17:26
RabidGravy which would be quite weird 17:27
ugexe "\x[d]\x[a]\x[d]\x[a]".encode will encode to 13, 10, 13, 10
jnthn \n in a regex is logical
sprocket RabidGravy: appreciate it, thanks for all your help! 17:28
ugexe yea, but i want to write a string with a \r\n that doesnt match a \n token in a regex
jnthn \n in a regex has always matched \r\n 17:29
You need to write something like \xA in the regex if you really mean "exactly this char"
TimToady but that won't work either if the Str is NFG
jnthn How so? 17:30
TimToady not unless you ignore the \n "mark"
er, wait, \xa is linefeed
ugexe yea i already use \x0D \x0A 17:31
TimToady so you're matching only the "combiner"
is \xd\xa in a pattern somehow smart enough to match a single grapheme?
jnthn ugexe: Can you give me a golfed example of what you're doing and geting unexpected results with? I'm not quite following what you want to do.
jnthn TimToady: Yes 17:31
TimToady okay 17:32
ugexe ill try, but its for Grammar::HTTP which is probably the poster child for something that would choke on such a change 17:33
ugexe but it seems to boil down to me writing the headers as a string for tests, and those \r\n get turned into \n 17:34
jnthn say so qq{\r\n} ~~ /\r\n/ # True 17:36
say so qq{\n} ~~ /\r\n/ # False
This is without the Windows magical \n in place yet 17:37
garu Skarsnik: nevermind, I thought you meant something else. That said, I'm not entirely sure what you do mean :) 17:39
raiph Have folk discussed the idea of `has $foo` being the mid ground between 100% private `has $!foo` and 100% public `has $.foo`? 17:41
TimToady the middle ground is a bog 17:42
jnthn raiph: What would that mean, exactly?
Skarsnik garu, I written a module (Gumbo) it's a binding to a c lib. It export a single function but define internally lot of cstruct/ccall. I want to write a Gumbo::Parser class in a Gumbo/Parser.pm6 file and have access to all these internal cstruct. But I don't want then exposed to an user of the module
tony-o Skarsnik: just don't export the struct classes.. 17:43
mr_ron m: grammar G { token TOP { a+ } }; my Nil $x = G.parse("zz") 17:44
camelia rakudo-moar 273e89: OUTPUT«Type check failed in assignment to $x; expected Nil but got Any␤ in block <unit> at /tmp/r2t2se6KEB:1␤␤»
raiph m: class c { has $foo }; c.new(:foo) # would initialize the private foo attribute # jnthn 17:44
camelia ( no output )
PerlJam raiph: and how to we retrieve the private $foo value? 17:45
tony-o you'd have to write your own getter/setters for it.. 17:45
raiph PerlJam: same as currently
JimmyZ m: has $foo; 17:46
camelia rakudo-moar 273e89: OUTPUT«5===SORRY!5=== Error while compiling /tmp/9XkvUyLB19␤You cannot declare attribute '$foo' here; maybe you'd like a class or a role?␤at /tmp/9XkvUyLB19:1␤------> 3has $foo7⏏5;␤ expecting any of:␤ constraint␤»
mr_ron m: grammar G { token TOP { a+ } }; my Nil $x = G.parse("zz") # docs.perl6.org/type/Grammar#method_parse "Returns a Match object on success, and Nil on failure." 17:47
camelia rakudo-moar 273e89: OUTPUT«Type check failed in assignment to $x; expected Nil but got Any␤ in block <unit> at /tmp/M2IcZbVvcg:1␤␤»
jnthn raiph: Ah, initializable publicly but then not public. No, we've not considered that, and I don't think it's likely.
mr_ron What am I missing again ? 17:47
raiph jnthn: thanks 17:49
mr_ron m: grammar G { token TOP { a+ | <bs> }; token bs { b+ } }; say G.parse("zz").WHAT; say G.parse("aa")<bs>.WHAT 17:56
camelia rakudo-moar 273e89: OUTPUT«(Any)␤Nil␤»
garu Skarsnik: what tony-o said... looks as if you'd have to do something if you wanted to expose the cstruct, not the other way around 17:58
mr_ron Is G.parse() returning the wrong thing? It looks to me like it is. 18:05
tony-o m: grammar G { rule TOP { a+ } }; G.parse('zz').perl.say; 18:06
camelia rakudo-moar 273e89: OUTPUT«Any␤»
tony-o m: grammar G { rule TOP { a+ } }; G.parse('zz').WHAT.say;
camelia rakudo-moar 273e89: OUTPUT«(Any)␤»
tony-o m: say Nil ~~ Any;
camelia rakudo-moar 273e89: OUTPUT«True␤»
tony-o m: say Any ~~ Nil;
camelia rakudo-moar 273e89: OUTPUT«False␤»
tony-o m: my Nil $a = Any.new;
camelia rakudo-moar 273e89: OUTPUT«Type check failed in assignment to $a; expected Nil but got Any␤ in block <unit> at /tmp/eTOiFilv7h:1␤␤»
mr_ron m: grammar G { token TOP { a+ } }; my Nil $x = G.parse("zz") 18:07
camelia rakudo-moar 273e89: OUTPUT«Type check failed in assignment to $x; expected Nil but got Any␤ in block <unit> at /tmp/HMNfyPJjvI:1␤␤»
tony-o mr_ron: appears either the doc is wrong or .parse is returning the wrong type
mr_ron The definition of Nil is the 'Absence of a value' suggesting to me the doc makes sense anyway 18:09
Skarsnik :m my @t = 1..5; @t.tail(1); 18:10
m: my @t = 1..5; @t.tail(1);
camelia ( no output )
tony-o mr_ron: agreed
Skarsnik I get Method 'tail' not found for invocant of class 'Array' ~~ 18:11
tony-o update rakudo 18:13
Skarsnik it's a recent addition? 18:14
gfldex Date: Tue Oct 27 23:02:09 2015 +0100 18:15
Introduce .head($n)/.tail($n)
Skarsnik: ^^^
ugexe this \r\n problem seems it may be related to submatches and matching on \r\n 18:21
ugexe my $request = q{GET /http.html HTTP/1.1} ~ "\r\n"; my $u = $request ~~ /\r\n/; say $u.Str ~~ /^\n/; 18:21
the actual token is `token status-line { <HTTP-version> <.SP> <status-code> <.SP> <reason-phrase> <.CRLF> }`. By passing in a string that ends with "\r\n" the <CRLF> is left to match against ord 13, and not \x0d \x0a, so it fails 18:25
dalek kudo/lines-vm-chomp: afef47d | jnthn++ | src/core/IO/Handle.pm:
Use nqp::readlinechompfh for .get and .lines.

Needs the JVM backend to support this op before we can merge it, but pushing it so nobody duplicates the work, and so we can see how much it helps performance.		 18:43
jnthn ugexe: A string "\r\n" will match /^\n/ 18:44
ugexe: But if we're really getting the ord 13 in there, yeah, something's not right
ugexe: But still not seeing how that happens. Alas, time to make dinner now. 18:45
jnthn & 18:47
flussence looks like panda needs \r\n-fixing... explodes in the projects.json HTTP request stuff. 19:09
dalek c: 7a6dc7d | (Steve Mynott)++ | doc/Language/faq.pod:
minor typos and tweaking
 19:11
kmel hello everybody 19:29
vendethiel ahoy! 19:30
PerlJam kmel: greetings!
kmel if i defined a sub with optional arguments, how can i check if the user passed arguments or not? 19:31
hi vendethiel PerlJam
vendethiel kmel: as in, did not pass Nil explicitly? 19:32
kmel vendethiel, the user just called a sub without any arguments 19:33
vendethiel m: sub foo($a?) { say $a.defined }; foo; foo(3); # kmel
camelia rakudo-moar 273e89: OUTPUT«False␤True␤»
PerlJam m: sub foo($a?) { say $a.defined }; foo(Any); foo(Nil); # note this too
camelia rakudo-moar 273e89: OUTPUT«False␤False␤»
kmel aha .defined will do the job 19:34
vendethiel PerlJam: yeah, hence my prior question
kmel thanks
vendethiel PerlJam thanks 19:35
PerlJam vendethiel: right, kmel's response led me to believe this was a foreign concept
flussence if I do «$file.print: @stuff», rakudo writes @stuff as-is, but if I do «$socket.print: @stuff», the elements get joined with spaces between. Is that meant to happen? 19:36
PerlJam flussence: I hope not.
grondilu is not sure what flussence means by "as-is" 19:44
flussence if @stuff is ^10, the output would be exactly 10 chars long 19:45
grondilu ok, a concatenated output then 19:46
PerlJam flussence: I'd probably change IO::Socket.print into multi methods just like IO::Handle does. 19:58
Though that would mean that passing Cool things to print would call the listy version which would stringify and call the stringy version whereas now, it only calls the stringy version. 20:02
cognominal I got something in a Buf, how to see it as a repr(CStruct). How do I make it independant of the indianness? I would like to avoid unpack. 20:20
for info, that's a .git/index
bartolin r: use nqp; say nqp::backendconfig<osname> 20:28
camelia rakudo-jvm 273e89: OUTPUT«(Any)␤»
..rakudo-moar 273e89: OUTPUT«linux␤»
cygx cognominal: for the first part of the question, cf:
class Foo is repr<CStruct> { has int8 $.a; has int8 $.b; }; my $buf := Buf[int8].new(42,2,37); say nativecast Foo, CArray[int8].new($buf.list); # => Foo.new(a => 42, b => 2)
cygx it's not efficient as copying to the CArray happens element (ie byte) by element 20:29
it would be better if you could get at a CPointer to the data
bartolin r: use nqp; say nqp::atkey(nqp::backendconfig(), "osname") # that prevents JVM build atm
retupmoca cygx: you can't nativecast directly from a blob? 20:30
camelia rakudo-moar 273e89: OUTPUT«linux␤»
..rakudo-jvm 273e89: OUTPUT«java.lang.NullPointerException␤ in block <unit> at /tmp/tmpfile:1␤␤»
retupmoca cygx: I know you can nativecast from a Blob -> CArray (nativecast CArray[uint8], $stuff;)
cygx retupmoca: my bad, you actually can
retupmoca err, from a blob to a carray at least
cygx no idea why I thought you could not... 20:31
ie nativecast Foo, $buf in the example works as expected
cognominal: ^ 20:32
retupmoca ah, ok. Good, that's much faster :)
cygx now, does anyone know how 'int' gets its nativesize from? 20:32
I've been ack-ing through the source, but no dice yet 20:33
retupmoca pretty sure int is always 64bit on a 64bit build
and 32bit on a 32bit build
cognominal thx, cygx.
cygx retupmoca: NativeCall thinks int should map to long, which is 32-bit on windows 20:34
retupmoca cygx: that may have changed then - when I was first writing Compress::Zlib, 'int' didn't work on windows because it was 64bit 20:36
I had to end up using 'long'
star: use NativeCall; my long $x = 32; say $x;
camelia star-m 2015.09: OUTPUT«32␤»
cygx well %type_map in NativeCall.pm says int translates to long, but nativesizeof int is 8 (whereas a proper long is size 4 here) 20:37
zengargoyle how often is doc.perl6.org built from repo sources? 20:38
cygx why on earth someone who's interested in supporting both *nix and windows would use long as a default type for int is beyond me anyway 20:39
zengargoyle m: say <a a b b b c c>.repeated 20:40
camelia rakudo-moar 273e89: OUTPUT«(a b b c)␤»
zengargoyle m: say <a b b c c b a>.repeated
camelia rakudo-moar 273e89: OUTPUT«(b c b a)␤»
[Coke] java bisect is done. the skipped commits here all complained about an NQP op- I was testing for the NPE in the current build.
gist.github.com/coke/bf0196891ef9dc9b8ada
zengargoyle m: say <a A B b c b C>.repeated(:as(&lc))
camelia rakudo-moar 273e89: OUTPUT«(A b b C)␤»
AlexDaniel m: grammar Test { token TOP { ^ <custom=x> $ }; token x { .* } }; say Test.parse(‘foo’); 20:42
camelia rakudo-moar 273e89: OUTPUT«｢foo｣␤ x => ｢foo｣␤ custom => ｢foo｣␤»
[Coke] ah. so 45e56b8ed7f403e2c45b3bbedab41a4a41afb710 looks like what probably changed one error to another.
AlexDaniel Is it a bug?
cygx for now, I'll just do a `constant int = int32` to make my tests pass ;)
retupmoca cygx: assuming a 64bit build, 'int' is always 64 bits within the perl6 world. I'm not sure what NativeCall does with it, but you probably want to use 'long' or 'longlong' instead of 'int' for NativeCall 20:43
zengargoyle Type/List doc has the first two of those wrong. and a couple Perl5 POD =over/=back lists that don't render correctly. have to look up Perl 6 POD version of lists...
[Coke] I'm guessing that nqp::setinputlineseps(handle, endings);
is a problem on the jvm, since we're now giving it a list.
testing...
AlexDaniel I was expecting to see “custom” without “x”…
retupmoca cygx: basically, if you use 'int' for NativeCall code, it's probably wrong
bartolin [Coke]: irclog.perlgeek.de/perl6/2015-11-03#i_11477695
PerlJam m: grammar Test { token TOP { ^ <custom=.x> $ }; token x { .* } }; say Test.parse(‘foo’); 20:44
camelia rakudo-moar 273e89: OUTPUT«｢foo｣␤ custom => ｢foo｣␤»
cygx retupmoca: sure - but it nevertheless should be consistent
PerlJam AlexDaniel: ^^^ 20:44
cygx currently, it is not
AlexDaniel oh right!
PerlJam: I forgot about that dot!
thanks
[Coke] sighs, he missed that bartolin found anything.
[Coke] r: say $*VM.config<osname> 20:45
camelia rakudo-jvm 273e89: OUTPUT«(Any)␤»
..rakudo-moar 273e89: OUTPUT«linux␤»
lizmat waves from the Amsterdam.pm meeting
jnthn: something came up here wrt to using multiple line endings 20:46
ilmari r: say $*KERNEL.name
bartolin [Coke]: looks like it reads the values from nqp/bin/jvmconfig.properties -- and there is no osname :/
camelia rakudo-{moar,jvm} 273e89: OUTPUT«linux␤»
lizmat suppose we have 4 different line endings. and we're chomping. how would you find out *which* line ending was used ? 20:47
to mark the end of a line
[Coke] bartolin: that line is never invoked on the jvm.
it's warpped in an #if moar block ?
*wrapped
lizmat jnthn: so perhaps, we need something like nqp::chomped
[Coke] j: nqp::p6box_s(nqp::atkey(nqp::jvmgetproperties(), 'os.name'));
camelia rakudo-jvm 273e89: OUTPUT«===SORRY!===␤java.lang.NullPointerException␤» 20:48
[Coke] -there- we go.
lizmat which would give you the line ending of the last line that was read with nqp::linechomp (or whatever it is / will be called)
cygx retupmoca: to clarify, there are basically 3 types of 'int' - p6-land int (normally 64-bit), C-land int (normally 32-bit) and NativeCall-parameter-land int, which is a long, ie corresponding to p6-land int on linux and C-land int on win32 20:50
bartolin [Coke]: src/core/core_prologue.pm line 24
[Coke]: at least that looked like the source of the problem to me
cygx I've got a failing testcase which I suspect indicates that somewhere, we assume P6-land and NativeCall-parameter-land agree
[Coke] bartolin: ah, I was looking in src/core/IO/Spec.pm 20:51
PerlJam lizmat: it's too bad chomp itself can't tell you what it chomped. 20:56
lizmat PerlJam: we want to prevent chomping in Perl 6 land
and it should become available, *especially* if we support muiltiple simultaneous line endings 20:57
[Coke] j: use nqp; say nqp::p6box_s(nqp::atkey(nqp::jvmgetproperties(), 'os.name')); 20:57
camelia rakudo-jvm 273e89: OUTPUT«Linux␤» 20:58
[Coke] ok, the NPE we got on that before was for something else. :)
fixing...
lizmat m: $*DISTRO.name 20:59
camelia ( no output )
bartolin [Coke]++
lizmat m: say $*DISTRO.name
camelia rakudo-moar 273e89: OUTPUT«opensuse␤»
lizmat [Coke]: isn't that what you're looking for ?
[Coke] lizmat: Yes, but if you look in backscroll, that gave me an NPE earlier. 21:00
so it sent me down an alley for a few minutes. I'm fine now. 21:01
lizmat j: say $*DISTRO.name
camelia rakudo-jvm 273e89: OUTPUT«opensuse␤»
lizmat ok
[Coke] We now have 2 places where we are keying off jvm vs. moar to get the os name. we should probably not do that.
bartolin another option seems to be to add 'osname=$Config{osname}' to 'nqp/tools/build/gen-jvm-properties.pl'
moritz but we want the OS name of the host system, not where it was compiled 21:03
and java code is supposed to be portable, no?
bartolin oh!
[Coke] I'm testing a patch using the same logic in IO/Spec
[Coke] bartolin++ for tracking that down. no doubt faster than doing a bisect over 30 versions with 2 very different forms of breakage. :) 21:09
dalek c: c1140e7 | lizmat++ | doc/Type/List.pod:
Fix problems found by zengargoyle++
[Coke] still more NPEs.. 21:10
bartolin :-/
[Coke] and of course the stack trace starts out with no line number or sub name. 21:11
in (gen/jvm/CORE.setting)
in (gen/jvm/ModuleLoader.nqp:380) 21:12
in load_setting (gen/jvm/ModuleLoader.nqp:370)
it's only 34287 lines...
AlexDaniel So if I want to have a constructor, do I really have to do this? “method new(&callback, *@dependencies) {” or is there something simpler? 21:14
moritz AlexDaniel: you get a constructor by default 21:15
AlexDaniel moritz: I know, but what if I want to do stuff? :)
moritz AlexDaniel: you can do stuff with the default constructor
AlexDaniel moritz: hmmm
moritz AlexDaniel: you seem to want a custom constructor, but to not write a custom constructor. That confuses me. 21:16
AlexDaniel moritz: nah, that's just my non-perl6 thinking, I guess 21:17
moritz: thanks! I'll try that
I mean, I'll try to see if I really need one
AlexDaniel m: my regex foo { ‘test’ } 21:19
camelia rakudo-moar 273e89: OUTPUT«5===SORRY!5===␤Unrecognized regex metacharacter ‘ (must be quoted to match literally)␤at /tmp/CZeBhJ1YlJ:1␤------> 3my regex foo {7⏏5 ‘test’ }␤Regex not terminated.␤at /tmp/CZeBhJ1YlJ:1␤------> 3my regex foo { ‘7⏏5test’…»
AlexDaniel ‘’ do not work in regexes?
moritz m: my $x = ‘foo’ 21:20
camelia ( no output )
moritz seems not
AlexDaniel bug or not?
inconsistency for sure
moritz dunno; I haven't seen fancy quotes in S05
zengargoyle lizmat++ thanks for fixing, i was off reading up on POD6 :) 21:21
[Coke] They were added to the language much later by timtoady directly. I wouldn't mind if they also existed in rules.
bartolin: any idea how to find that exception? 21:22
AlexDaniel submits rakudobug then
[Coke] r: say $*VM.config<osname>; say $*VM.config<os.name>; 21:23
bartolin [Coke]: I'm looking, but without a good idea .
camelia rakudo-jvm 273e89: OUTPUT«(Any)␤(Any)␤»
..rakudo-moar 273e89: OUTPUT«linux␤(Any)␤»
[Coke] r: say Any eq 'MSWin32' 21:24
camelia rakudo-{moar,jvm} 273e89: OUTPUT«Use of uninitialized value of type Any in string context␤Any of .^name, .perl, .gist, or .say can stringify undefined things, if needed. in block <unit> at /tmp/tmpfile:1␤False␤»
dalek kudo/nom: 6cb35b4 | coke++ | src/core/core_prologue.pm:
Fix one potential NPE on the jvm.
 21:25
c: 3ebb144 | (Christopher Bottoms)++ | doc/Language/faq.pod:
Minor edits

Couple of grammar edits, plus replacing space with nonbreaking space in "Perl 6".		 21:26
c: 909ac44 | RabidGravy++ | doc/Language/faq.pod:
Merge pull request #183 from molecules/patch-1

Minor edits
AlexDaniel moritz: I want to write something like “has $.foo = initfoo();”, but I can't do that (where initfoo is a method). What would be the right way to get it to work? 21:33
or should I probably have one .init() method and call that manually? Not sure if I want that 21:34
lizmat AlexDaniel: do you know about BUILD 21:35
?
AlexDaniel lizmat: well, I've seen that, but it's a bit hard to type 21:36
lizmat B U I L D :-)
AlexDaniel lizmat: what about (:&!callback, :@!dependencies) ? Or should I just omit that?
yes, indeed 21:37
lizmat: right!
lizmat: thanks
MadcapJake Why am I getting "Can't use unknown trait 'is native' in a sub declaration"? 21:50
[Coke] m: sub foo is native { * }; 21:51
camelia rakudo-moar 273e89: OUTPUT«5===SORRY!5=== Error while compiling /tmp/t5gKqcM7U9␤Can't use unknown trait 'is native' in a sub declaration.␤at /tmp/t5gKqcM7U9:1␤ expecting any of:␤ rw raw hidden-from-backtrace hidden-from-USAGE␤ cached pure default DEPRE…»
[Coke] m: use Nativecall; sub foo is native { * };
camelia rakudo-moar 273e89: OUTPUT«===SORRY!===␤Could not find Nativecall in any of:␤ file#/home/camelia/.perl6/2015.10-134-g273e895/lib␤ inst#/home/camelia/.perl6/2015.10-134-g273e895␤ file#/home/camelia/rakudo-inst-1/share/perl6/lib␤ file#/home/camelia/rakudo-inst-1/share/pe…»
ilmari m: use NativeCall; sub getuid() returns uint32 is native(Str) {}; say getuid(); 21:52
camelia rakudo-moar 273e89: OUTPUT«1012␤»
ugexe m: class Foo { has $.bar; method new(*%_) { my $self = self.bless(|%_); $self.init(); $self; }; method init { $!bar = 1; }; }; my $x = Foo.new( :bar(3) ); say $x.perl; # not ideal, but sorta works 21:54
camelia rakudo-moar 273e89: OUTPUT«Foo.new(bar => 1)␤»
MadcapJake oh i need to declare `use NativeCall` thanks!
AlexDaniel is anybody using HTTP::Client ?
cygx there's a weird bug for you: I pass a block to use so I can execute it in the corresponding EXPORT; if the block contains calls with named parameters, their value gets passed as positionals
the workaround: use a sub() {} instead 21:55
AlexDaniel it does not work for me and travis says ｢build | unknown｣
cygx (or rather a sub () {} - whitespace matters) 21:55
ilmari m: class Foo { has $.bar; submethod BUILD { $!bar = 1 } }; say Foo.new(:bar(3)) 21:56
camelia rakudo-moar 273e89: OUTPUT«Foo.new(bar => 1)␤»
ugexe but then you have to declare all your variables inside the BUILD signature
ilmari only the ones that you want to initialise from a method call 21:57
ugexe yes, thats what he wanted to avoid
MadcapJake is there an environment variable that perl6 uses to find native libraries?
ilmari ugexe: it's a lot less than your example of overriding new 21:58
ugexe not if you have multiple variables
with long variable names at that
ilmari ugexe: how is your method init { $!bar = 1; } different from submethod BUILD { $!bar = 1; }?
RabidGravy AlexDaniel, the build unknown just means that travis is turned on for it, but hasn't actually run yet 21:59
ilmari other than the fact tht you have to override new and call it manually, that is
ugexe because you can pass mine any number of parameters without having to declare a huge BUILD signature
AlexDaniel RabidGravy: interesting. Looking at the log, I see that on my machine all tests were successful
so it should work
ilmari m: class Foo { has $.bar; method new(*%_) { my $self = self.bless(|%_); $self.init(); $self; }; method init { $!bar = 1; }; }; say Foo.new(:bar(3),:baz(42) ); 22:00
camelia rakudo-moar 273e89: OUTPUT«Foo.new(bar => 1)␤»
ilmari ugexe: you still have to declare the attributes for them to be stored anywhere, no? 22:01
ugexe yes
has $.whatever; instead of has $.whatever; submethod BUILD($!whatever)
RabidGravy MadcapJake, look at LibraryMake that has an example
ilmari ugexe: eh, you don't need a signature on BUILD
m: class Foo { has $.bar; has $.baz submethod BUILD { $!bar = 1 } }; say Foo.new(bar=>3, baz=>7) 22:02
camelia rakudo-moar 273e89: OUTPUT«5===SORRY!5=== Error while compiling /tmp/rpDKau3RCf␤Two terms in a row␤at /tmp/rpDKau3RCf:1␤------> 3class Foo { has $.bar; has $.baz7⏏5 submethod BUILD { $!bar = 1 } }; say Fo␤ expecting any of:␤ infix␤ infix stoppe…»
ilmari m: class Foo { has $.bar; has $.baz; submethod BUILD { $!bar = 1 } }; say Foo.new(bar=>3, baz=>7)
camelia rakudo-moar 273e89: OUTPUT«Foo.new(bar => 1, baz => Any)␤»
AlexDaniel RabidGravy: “Unsupported protocol, 'https'.” uh!
RabidGravy boo
ugexe aha i did not know that 22:03
AlexDaniel “if ($request.protocol ne 'http') { die …”
I wonder if it will work without this check
RabidGravy AlexDaniel, if it doesn't use OpenSSL unlikely 22:04
AlexDaniel okay :(
I'll just create an issue on github then
RabidGravy HTTP::UserAgent *does* support https 22:05
AlexDaniel RabidGravy: OH!
RabidGravy: I did not notice it. Thanks!
though I wonder if I should create an issue anyway?
RabidGravy do it 22:06
AlexDaniel ok, thanks
ilmari ugexe: oh, that appears to inhibit the default bindig to attributes 22:07
Skarsnik I was wondering, should LWP::Simple support cookies? I mean for me it like the equivalent of writing 'wget mywebsite' 22:18
tadzik HTTP::UA does, I think 22:19
ilmari well, wget supports cookies, and even has --load-cookies, --save-cookies 22:20
by default it won't load or save, but it'll use them if it does multiple requests in a session
AlexDaniel Skarsnik: does it support https? :)
Skarsnik Yes, but if I just want to get some html from a webpage once. I don't necessary need to create a ua object 22:21
AlexDaniel try require IO::Socket::SSL; # seems so
Skarsnik ilmari, I think wget does. I have an exemple where requesting a page without cookies just fail (or does weird thing) and wget work on it 22:22
ilmari Skarsnik: yes, it sets a cookie and redirects, wget will send that cookie on the second request, if it applies 22:23
s/, it/, if it/
vendethiel mmh, do we have a module for terminal color codes around? 22:25
Skarsnik could be cool if lwp:simple act a similar way like "give me my html" I don't care for details
* and it leave the dev out of details
vendethiel yay \o/. I implemented a little script to sync all my repos, push the correct branch, and then ping on some remote at $work. wanted to give it nice output
sprocket how do i declare a typed array in p6? 22:26
ie) an array of Str
Skarsnik my Str @array
sprocket oh, that’s straightforward then :)
thanks!
RabidGravy Skarsnik, "use HTTP::UserAgent :simple; say get('some-url')" 22:34
Skarsnik hm 22:35
I think it will not work, since by default cookie are not activated 22:37
RabidGravy if it doesn't send an isuue with the headers of the request, as it should 22:38
jdv79 the header parsing code is buggy if thats what you mean 22:45
Skarsnik hm interesting it work fine now. I had a really weird behavior from H:UA 2 months ago (like it took 0 sec or 50 sec to get this page) 22:48
RabidGravy Skarsnik, quite a lot of fixes gone in the last few months 22:52
Skarsnik Unrelated, is there some tricks to speed up the creation of new object? 23:01
caymanboy is looking forward to Perl6 1.0 next month 23:02
jdv79 yes, H:UA had a bunch of bugs more than it still has:) 23:04
ugexe jnthn: my problem was defining token CRLF { [<.CR> <.LF>] }, so the match on <.CR> would fail. token CRLF { \x0d \x0a } fixed it. grammar::tracer++ 23:05
ugexe that behavior is still slightly confusing though 23:06
AlexDaniel caymanboy: oh hi again :) 23:07
japhb .ask pmurias Do you need help getting perl6-bench working for nqp-js? If you PM me configure/build instructions, I can do the necessary patch to build and benchmark it along with the other backends. (timotimo knows how too, but I don't want to speak for his time.) 23:27
yoleaux japhb: I'll pass your message to pmurias.
RabidGravy right bed time 23:39
zengargoyle does anybody have a recommendation for/against any of various the github CLI tools? anything to make it easy to keep a cloned fork up to date and submit pull requests and such without going through the web gui? 23:44
Hotkeys there's always git 23:53
dalek osystem: b3c8a7b | retupmoca++ | META.list:
Add Crust::Middleware::Syslog

  github.com/retupmoca/P6-Crust-Middleware-Syslog		 23:57

Powered by Cro and the Raku® Programming Language.

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

Thank you!