»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.
00:03 mattprelude left, tokuhiro_ left
timotimo were you able to get some kind of measurement for performance? 00:06
00:06 AlexDaniel joined
timotimo it seems to react faster on my end 00:10
00:22 timbunce_ left, timbunce__ joined 00:26 Dojk joined 00:27 rivarun left, Dojk left 00:28 rivarun joined
atweiden from inside a grammar's Actions class, is it possible to retroactively cause a regex match to fail? 00:30
or is that only possible to do in a closure inside the regex?
e.g. ix.io/laC 00:31
00:31 bjz left 00:42 leont left 00:49 bjz joined
ShimmerFairy atweiden: you can use <?{ ... }> inside a regex to test for a true condition (and use <!{ ... }> for a false condition) 00:52
m: say "foo" ~~ /(\w+) <?{$0 ~~ "foo"}>/
camelia rakudo-moar c52b0b: OUTPUT«｢foo｣␤ 0 => ｢foo｣␤»
ShimmerFairy m: say "bar" ~~ /(\w+) <?{$0 ~~ "foo"}>/
camelia rakudo-moar c52b0b: OUTPUT«Nil␤»
ShimmerFairy I'm not entirely sure how you'd do it from within an actions method, I suspect you'd probably have to mess with the actual Cursor object at that point. 00:53
atweiden ShimmerFairy: ty, what would manipulating the Cursor object look like in practice? 00:56
ShimmerFairy I don't know for sure, it'd probably end up being $/.CURSOR.'!fail_cursor'() to cause a failure. But I've never had to manipulate the cursor directly, so I'm not a good source of info for that :) 00:57
00:58 timbunce__ left
ShimmerFairy atweiden: most of the time though, if your regex still has a chance to fail, there's a way to do the check inside the regex, without having to do in the actions (at least, I've never had to resort to making a cursor fail in the actions side) 00:58
atweiden hmm, cursor appears undefined inside Actions class 00:59
00:59 bjz left, timbunce_ joined
ShimmerFairy note that you get the cursor via (Match object).CURSOR, and a lot of the methods on Cursor need to be quoted to call them, due to some oddity with how !private method names are handled in NQP. 01:00
ShimmerFairy afk for a while
01:01 vvz left 01:04 adhoc left 01:06 vendethiel joined 01:07 abaugher left 01:08 abaugher joined 01:15 ggoebel2 joined 01:16 ggoebel left, bjz joined 01:29 vendethiel left 01:31 roguelazer left, preyalone left, chansen_ left 01:32 adhoc joined, avalenn_ left, mephinet- left, avalenn joined 01:33 mprelude left, mephinet joined, samb1 left, roguelazer joined 01:34 preyalone joined, chansen_ joined 01:35 mprelude joined, mr_ron left, samb1 joined 01:46 mr_ron joined 02:00 dha joined, obra joined 02:01 rhr joined
dalek c: f5f915a | (David H. Adler)++ | lib/Type/ (2 files):
Added usage statements to Attribute.pod and Backtrace/Frame.pod
 02:04
c: 7a5339f | (David H. Adler)++ | lib/Type/Enum.pod:
Added usage statements to Enum.pod
 02:09
c: 5f093cd | (David H. Adler)++ | lib/Type/Enum.pod:
Merge branch 'usage_statements'

With usage statements for Enum.pod
02:14 aborazmeh joined, aborazmeh left, aborazmeh joined
dha So... the two signatures for C<await> in Channel.pod are "multi sub await(Channel:D)" and "multi sub await(*@)". I assume that second signature means it takes a list. Is it a list of Channels? Or could it be a list of anything? 02:17
The following docs indicate that it takes one or more channels, but the signature seems inspecific. 02:20
dalek c/usage_statements: 7a5339f | (David H. Adler)++ | lib/Type/Enum.pod:
Added usage statements to Enum.pod
 02:23
c/usage_statements: 088e9a8 | (David H. Adler)++ | lib/Type/C (2 files):
Added usages statements to Capture.pod and Channel.pod
c: 088e9a8 | (David H. Adler)++ | lib/Type/C (2 files):
Added usages statements to Capture.pod and Channel.pod
 02:25
c: 119dd07 | (David H. Adler)++ | lib/Type/C (2 files):
Merge branch 'usage_statements'

Adding usage statements for Capture.pod and Channel.pod
02:29 skids joined, salva left 02:30 salva joined 02:31 noganex_ joined 02:34 noganex left 02:35 dha left 02:36 dha joined, dha left
ShimmerFairy dha: Probably because you can't currently place a type constraint on a slurpy, but I'd put it in anyway for documentation purposes :) 02:39
02:48 cognominal left 02:49 kid51 left 03:00 aborazmeh left 03:11 Xor_ joined 03:16 lizmat_ joined 03:17 PerlJam left 03:18 rivarun left 03:19 Kcop left, perlpilot joined, pink_mist left, Exodist left, BenGoldberg left, lizmat left, Woodi left 03:20 pink_mist joined, Woodi joined 03:21 pink_mist left, pink_mist joined, raiph joined, Exodist joined, rivarun joined 03:37 mr_ron left 03:51 timbunce__ joined 03:52 timbunce_ left 03:54 laouji joined 04:10 mr_ron joined 04:24 jordman left, vvz joined, jordman joined 04:53 khw left 04:56 tokuhiro_ joined 04:57 mr_ron left 05:00 tokuhiro_ left 05:04 mr_ron joined 05:07 raiph left 05:12 vvz left 05:17 z448 joined 05:21 z448 left 05:26 kaare__ joined 05:29 bjz left 05:32 bjz joined
TimToady m: my $s = 'o⃗'; say $s ~~ m:m/o/; 05:48
camelia rakudo-moar c52b0b: OUTPUT«False␤»
TimToady something busted there
m: my $s = 'o⃗'; $s ~~ s:mm/o/p/; say $s 05:49
camelia rakudo-moar c52b0b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/yVsRpIk248␤Adverb mm not allowed on substitution␤at /tmp/yVsRpIk248:1␤------> 3my $s = 'o⃗'; $s ~~ s:mm/o/p/7⏏5; say $s␤»
TimToady and very busted there
ShimmerFairy star-m: my $s = 'o⃗'; say $s ~~ m:m/o/;
camelia star-m 2015.03: OUTPUT«5===SORRY!5=== Error while compiling /tmp/VgyqNhZ4ud␤Adverb m not allowed on m␤at /tmp/VgyqNhZ4ud:1␤------> 3my $s = 'o⃗'; say $s ~~ m:m/o/7⏏5;␤»
TimToady m: my $s = 'o⃗'; $s ~~ s:samemark/o/p/; say $s
camelia rakudo-moar c52b0b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/1A_uVYF5Ze␤Adverb samemark not allowed on substitution␤at /tmp/1A_uVYF5Ze:1␤------> 3my $s = 'o⃗'; $s ~~ s:samemark/o/p/7⏏5; say $s␤»
TimToady m: my $s = 'o⃗'; say $s ~~ m:ignoremark/o/; 05:51
camelia rakudo-moar c52b0b: OUTPUT«False␤»
05:51 mr_ron left
ShimmerFairy m: say "ä" ~~ m:m/a/ 05:52
05:52 mr_ron joined
camelia rakudo-moar c52b0b: OUTPUT«｢ä｣␤» 05:52
ShimmerFairy m: my $s = "ä"; $s ~~ s:mm/a/o/; say $s
camelia rakudo-moar c52b0b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/pXruGBiidi␤Adverb mm not allowed on substitution␤at /tmp/pXruGBiidi:1␤------> 3my $s = "ä"; $s ~~ s:mm/a/o/7⏏5; say $s␤»
ShimmerFairy TimToady: for :m at least, I think it doesn't handle certain graphemes very well (I very faintly recall seeing it a while back, but I'm not sure) 05:53
m: say "o".samemark("ä") 05:54
camelia rakudo-moar c52b0b: OUTPUT«Method 'samemark' not found for invocant of class 'Str'␤ in block <unit> at /tmp/I5qPW4ZtBk:1␤␤»
ShimmerFairy the method doesn't even exist, no wonder :mm doesn't work.
TimToady coulda sworn that was implemented, but maybe it was pugs or niecza 05:56
ShimmerFairy I'm kinda surprised myself. I wonder if nqp has the opcodes to make it non-painful... 05:58
05:59 crux left
dalek ecs: 1847780 | ShimmerFairy++ | / (3 files):
Remove contents.pod

It's been very infrequently updated, nothing on design.perl6.org links to it, and while syntactically valid, wasn't a usefully formatted file. So it's been removed, and I suspect it won't bother anyone.		 06:06
06:07 atweiden left 06:13 AlexDaniel left 06:17 FROGGS[mobile] joined
FROGGS[mobile] .tell atweiden look at method _ws in rakudo/src/Perl6/Grammar.nqp 06:18
yoleaux FROGGS[mobile]: I'll pass your message to atweiden.
06:26 rurban joined
FROGGS[mobile] I am working on a patch that fixes 'my Int:D $foo'... should a 'Int:flubber' complain about an unknown colonpair/refinement? 06:36
jnthn: ^^ 06:37
ShimmerFairy FROGGS[mobile]: I should think so, I imagine it's only ignored as a lazy way of implementing :_ :P
06:38 domidumont joined
FROGGS[mobile] k 06:39
06:40 mr_ron left
ShimmerFairy FROGGS[mobile]: or in other words, I can't imagine why silently pretending it's :_ could be useful for anyone. 06:40
06:40 skids left 06:42 domidumont left 06:43 domidumont joined 06:49 labster left 06:50 labster joined 06:53 labster left 06:54 labster joined 06:57 tokuhiro_ joined 07:02 tokuhiro_ left, mr_ron joined 07:06 rindolf joined 07:20 vendethiel joined
moritz "unexpected type smiley :flubber" 07:40
07:40 darutoko joined
ShimmerFairy Does Perl 6 offer a way to know if a method is non-mutating? If not, then I don't really see how S26's proposed A<aliased-thing.method> can be made to work safely. 07:45
(Since A<> claims to allow the calling of 'non-mutating' methods)
[Tux] test 50000 35.839 35.723 07:48
test-t 50000 36.946 36.830
tux.nl/Talks/CSV6/test-t.jpg
tux.nl/Talks/CSV6/speed4.html 07:50
07:50 mr_ron left, mr_ron joined
moritz masak: fwiw 007 is on the hackernews front page 08:03
08:22 araujo left 08:25 xfix joined 08:28 vendethiel left 08:36 mr_ron left 08:38 mr_ron joined 08:43 mattprelude joined 08:49 rindolf left 08:58 rindolf joined 08:59 tokuhiro_ joined 09:03 tokuhiro_ left
lizmat_ good *, #perl6! 09:16
09:18 RabidGravy joined 09:19 spider-mario joined 09:23 vvz joined
moritz \o lizmat_ 09:24
09:25 rurban left, rurban joined 09:26 mr_ron left 09:28 darutoko- joined 09:31 darutoko left, darutoko joined 09:32 darutoko- left 09:33 rurban left 09:38 mr_ron joined 09:43 rurban joined 10:15 ggoebel joined 10:16 ggoebel2 left 10:20 leont joined 10:26 mr_ron left
jnthn ShimmerFairy: The "is pure" trait could be placed on a method to indicate that 10:27
FROGGS[mobile]: Are you just making `my Foo:D $x` just be a "NYI" error? But anyway, for things we don't have a clue what are I'd just have the error like "Don't understand :flubber in this context" or so 10:29
jnthn figures our NYI errors should get "in 6.christmas" added on to them
nine For those interested in the parse time regression: gist.github.com/niner/a66fb2caf1f0d0fb0c6f 10:31
jnthn nine++ 10:33
10:34 lizmat_ is now known as lizmat
lizmat hmmm... seems timotimo's patch causes some serious spectest issues? 10:36
leont I'm getting a «Cannot assign to a readonly variable or a value» exception on «make $/.Str.subst(/ <[\ \t]>+ $/, '');», I don't get what's going on 10:39
nine m: sub foo { my $s = 0; ($s += $_ for 1..3) }; say foo() 10:40
camelia rakudo-moar c52b0b: OUTPUT«(6 6 6)␤»
nine m: sub foo { my $s = 0; for 1..3 { $s += $_ } }; say foo()
camelia rakudo-moar c52b0b: OUTPUT«Nil␤»
10:41 mr_ron joined
nine m: sub foo { my $s = 0; ($s++ for 1..3) }; say foo() 10:41
camelia rakudo-moar c52b0b: OUTPUT«(0 1 2)␤»
psch hi #perl6 o/ 10:48
something funny about sigil'd constants: gist.github.com/peschwa/f07328a66ecaecb470bb
FROGGS[mobile] jnthn: I'm making it work 10:49
psch it still works right with BEGIN, which makes me stumble 'cause i have no idea where to look 10:50
maybe the way i'm trying to get a compile time value is wrong, i suppose
yeah, unbox to the rescue \o/ 10:56
leont Huh, now I'm really confused 10:57
My method TOP is being run, yet parse returns a Any? 10:58
moritz leont: then most likely it did not match 10:59
leont Inside the method TOP, the match object is looking fine, yet next step in the debugger it's not
11:01 AlexDaniel joined
moritz nopaste the code? 11:02
leont Ah, now I understand. It matches, but fails because parse wants to match until the end of the string
psch m: sub infix:(+-) { say $^a, $^b }; 5 +- 6 # misparsed as signature..? 11:04
camelia rakudo-moar c52b0b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/e_YNMY6cek␤Malformed parameter␤at /tmp/e_YNMY6cek:1␤------> 3sub infix:(+7⏏5-) { say $^a, $^b }; 5 +- 6 # misparsed ␤ expecting any of:␤ constraint␤»
11:05 pmurias joined
psch hm, the infix:() form only appears on RT, can't find it in the specs repo 11:06
moritz psch: even if it were designed and implemented, it would need tobe infix:('+-')
because () don't quote
psch moritz: yeah, right 11:07
moritz: well, i'll not-implement it as it's not designed... :)
m: sub infix:('+-') { say $^a, $^b }; 5 +- 6 # probably still should complain about *something* here 11:08
camelia rakudo-moar c52b0b: OUTPUT«WARNINGS:␤Useless use of "+" in expression "5 +- " in sink context (line 1)␤»
11:09 ely-se joined, Peter_R joined
psch m: &infix:('+')(1,2).say # this also doesn't appear in specs anymore, which means #68024 can be closed (or converted to LTA if we want a different error there...) 11:14
camelia rakudo-moar c52b0b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/ojxvvSZuB7␤You can't adverb ␤at /tmp/ojxvvSZuB7:1␤------> 3&infix:('+')7⏏5(1,2).say # this also doesn't appear in ␤»
11:24 aborazmeh joined, aborazmeh left, aborazmeh joined
leont Is there a way to match true iff the current rule has matched >0 characters? 11:26
moritz leont: . 11:27
leont: or .+
or an assertion that checks $/.chars
leont $/.chars is probably what I need
ShimmerFairy m: "foo" ~~ / f { say $/.chars } oo / 11:28
camelia rakudo-moar c52b0b: OUTPUT«1␤»
ShimmerFairy I'd be more inclined with $/.to or $¢.pos myself :)
11:29 Ybro joined, Ybro left, mr_ron left
leont ShimmerFairy: I don't think those would be correct in a multi-rule grammar 11:30
11:30 mr_ron joined
ShimmerFairy leont: why not? 11:30
leont It'd give a non-zero position if a previous rule already has matched characters
11:31 cognominal joined
ShimmerFairy ah, right 11:31
jnthn FROGGS[mobile]: Ah, I was gonna leave it for 6.d 'cus I wanted to do it by actually having :D/:U be first-class types. Trouble is that otherwise you end up with holes, like what happens when assigning through a pseudo-package, etc. 11:32
ShimmerFairy
.oO(emotypecons)
 11:34
11:41 patrickz joined
moritz .tell colomon smoke.perl6.org/report shows a T for IO::Notification::Recursive, even though that repo has no tests; should that be a P instead? 11:43
yoleaux moritz: I'll pass your message to colomon.
11:44 raz9r joined 11:45 patrickz left 11:50 kid51 joined
leont may be observing a rules engine reentrancy bug, is such an issue recently known? 11:54
11:56 kaare__ left
leont Doing a subst inside of an action gives me «Cannot assign to a readonly variable or a value» 11:57
moritz leont: the only gotcha that I'm aware of in that direction is that if you have a $/ declared (for example because you're in an action method), you cannot regex-match, because your $/ is read-only
it helps to give the parameter a different name (like $m instead of $/) 11:58
dalek kudo/nom: b06d30d | lizmat++ | src/core/ (2 files):
Refactor fail/die again

Alas, 2e92038e92c46fa7ec9e introduced quite some spectest errors. This commit goes back to using CALLER::CALLER:: PseudoStash Perl6 objects, rather than the lowlevel NQP objects that do not support things like EXISTS-KEY. But instead of before, it will only be doing a CALLER::CALLER:: once. And to further improve performance, the parameterless fail/die are not separate candidates, hopefully make it easier to inline.
Unfortunately, this causes 3 (new) spectest failures, but all failing tests seem suspect to me, so I'm putting this in rather than just revert 2e92038e92c46fa7ec9e .		 11:59
leont Yes, that does seem to help 12:00
12:02 pmurias_ joined
leont It is a rather surprising issue though 12:03
12:03 pmurias_ left
leont And it did used to work until fairly recently, it doesn't make sense for me for Str.subst to populate $/ 12:03
Might just be me 12:04
dalek p: c28a669 | (Pawel Murias)++ | / (3 files):
[js] Serialize variable length integers up to 32bit.

Add a unit test for that.
jnthn leont: I think the reason it does want to set $/ is so that .subst(/pattern/, { $0 }) works; otherwise for $0 to work you'd have to write -> $/ { $0 } 12:05
leont Right
Also, can't make sense out of an «Cannot bind attributes in a type object» error 12:06
12:07 Xor_ left
leont I've reduced the line it points to to «return $ret» and it still complains :-s 12:07
12:07 travis-ci joined
travis-ci NQP build failed. Pawel Murias '[js] Serialize variable length integers up to 32bit. 12:07
travis-ci.org/perl6/nqp/builds/83547070 github.com/perl6/nqp/compare/71dbe...8a6691a2f2
12:07 travis-ci left
leont Ah, it's make that's still assuming $/, of course 12:11
lizmat cycling& 12:12
leont doesn't like this work-around
12:14 BenGoldberg joined, mr_ron left 12:29 diana_olhovik_ joined 12:30 cognominal left, aborazmeh left 12:36 Xor_ joined
dalek kudo/nom: 53b8878 | peschwa++ | src/core/CompUnitRepo.pm:
Fix RT #126087.
 12:37
kudo/nom: d0885e5 | peschwa++ | src/ (3 files):
Implement bracket-delimited categorical sub declarations.

This fixes RT #119919. RT #117737 is related and still behaves the same, which I would argue as correct in the first place - a bare regexp that matches a single character is not a useful enough compile-time value to declare an operator with. I've added a typed exception to be thrown in that case though.
ast: 21a9ea0 | peschwa++ | S06-operator-overloading/sub.t:
Add tests for RT #119919 and RT #117737.
12:48 domidumont left
psch oh hrm 12:49
that probably should've actually been two commits each
13:00 araujo joined, araujo left, mr_ron joined 13:03 pmurias_ joined 13:05 Xor_ left
awwaiid hm. I don't think I've ever seen infix:[something], instead always infix:<something>. Is there a doc that talks about that you can point me at? 13:06
13:06 pmurias left
psch awwaiid: design.perl6.org/S02.html#Extended_identifiers 13:07
m: say &infix:['+'](2,3) 13:08
camelia rakudo-moar d0885e: OUTPUT«5␤»
13:09 Xor_ joined
psch m: constant sym = "µ"; sub infix:[sym] { $^a ~ "µ" ~ $^b }; say 5 µ 5 13:09
camelia rakudo-moar d0885e: OUTPUT«5µ5␤»
awwaiid I get the feeling sometimes that there could secretly be a much smaller "language" that is perl6 and most things be defined inside of that. And I feel like nqp isn't quite that, but I'm pretty uneducated on nqp.
thanks psch
ahhh, I see. so the "<" and ">" in that are just like other quotelike <> things. 13:11
psch m: constant sym = "µ"; sub infix:<<sym>> { $^a ~ "µ" ~ $^b }; say 5 µ 5 # this is still wrong i guess... 13:12
camelia rakudo-moar d0885e: OUTPUT«5===SORRY!5=== Error while compiling /tmp/0q5U65pxNc␤Two terms in a row␤at /tmp/0q5U65pxNc:1␤------> 3infix:<<sym>> { $^a ~ "µ" ~ $^b }; say 57⏏5 µ 5 # this is still wrong i guess...␤ expecting any of:␤ infix␤ infix…»
psch m: constant sym = "µ"; sub infix:<<sym>> { $^a ~ "µ" ~ $^b }; say 5 sym 5 # this is still wrong i guess...
camelia rakudo-moar d0885e: OUTPUT«5µ5␤»
psch m: my $a = "foo"; say <<$a>> 13:13
camelia rakudo-moar d0885e: OUTPUT«(foo)␤»
psch m: my $a = "foo"; say <<$a>>, <$a>
camelia rakudo-moar d0885e: OUTPUT«foo$a␤»
13:13 raz9r left
ShimmerFairy m: constant sym = "μ"; say «sym» 13:13
camelia rakudo-moar d0885e: OUTPUT«sym␤»
ShimmerFairy not sure why you expected <<sym>> to work there :)
awwaiid consistent at least
psch huh 13:14
m: constant $sym = "µ"; say <<$sym>>
camelia rakudo-moar d0885e: OUTPUT«(µ)␤»
psch m: constant $sym = "µ"; sub infix:<<$sym>> { $^a ~ "µ" ~ $^b }; say 5 µ 5
camelia rakudo-moar d0885e: OUTPUT«5===SORRY!5=== Error while compiling /tmp/flL4_S5FNd␤Colon pair value '$sym' too complex to use in name␤at /tmp/flL4_S5FNd:1␤------> 3constant $sym = "µ"; sub infix:<<$sym>>7⏏5 { $^a ~ "µ" ~ $^b }; say 5 µ 5␤ expecting any of:␤ …»
psch well, something is somewhat inconsistent there i think 13:15
ShimmerFairy psch: not really, how could you expect a bareword to interpolate?
psch ShimmerFairy: no, that's not it. the sigiled constant doesn't interpolate in infix:<<>> 13:16
that is inconsistent
ShimmerFairy psch: I'm not sure that's the case; I think it's complaining about the variable itself 13:17
psch m: constant $sym = "µ"; sub infix:[$sym]($,$) { 42 }; say 5 µ 5
camelia rakudo-moar d0885e: OUTPUT«42␤»
13:18 Xor_ left
awwaiid yeah, if htat works then infix:<<$sym>> should too 13:18
13:18 Xor_ joined
ShimmerFairy It may have something to do with the fact that «» does val() processing, but I can't be sure 13:18
(since «» is a shortcut for qq:ww:v[...]) 13:19
BenGoldberg m: my $a = 'aaaa'; ($a ~~ s:1st:2nd:3rd /a/b/); $a.say;
camelia rakudo-moar d0885e: OUTPUT«abaa␤»
psch ShimmerFairy: nibble_to_str tries to handle &val, i'm checking if it maybe misses something... 13:21
awwaiid ah! ok so I feel like infix should be more like s or q to be fully consistent. like infix<foo> instead of infix<foo>, just like rx<foo> or q<foo>. I think that's why I never noticed the quotelike for infix:<foo> 13:22
ShimmerFairy psch: if I had to guess, it'd be someplace that chooses not to make use of nibble_to_str (ISTR some issues cropping up related to not piping nibbles through that) 13:24
13:28 araujo joined 13:29 araujo left 13:31 raz9r joined, Peter_R left, raiph joined 13:32 kaare__ joined
psch no, nibbling is fine 13:32
it's forbidding e.g. infx:[/./] that's a bit overzealous here
+i
13:32 araujo joined 13:33 araujo left
psch 'cause <<>> builds val(QAST::Var.Stringy.words.Slip) 13:33
and atm it's implemented to only infix:<,> and &val for the QAST::Op case, which is apparently too restrictive :) 13:34
so the question turns to "which values are obviously bogus"? 13:36
i maintain that sub infix:[/./] { ... } is not a useful declaration, and will probably only make everything horrible if it was allowed :P
ShimmerFairy I don't quite follow what the "it" is in "it's implemented to only..."
and yeah, I feel a regex-y operator declaration is really begging for a slang more than a simple sub :) 13:37
psch "it" is nibble_to_str
github.com/rakudo/rakudo/blob/nom/....nqp#L2385
ShimmerFairy psch: ah, as I recall the &val handling was something I had to add, so it used to be just infix:<,> :P
psch this else forbids infix:[/./]
but it also forbids everything that's not infix:<,>... 13:38
checking for specific WVals is probably overly bloat-y
13:39 khw joined, raz9r left
ShimmerFairy psch: just to point out that the infix:<,> check has to do with how a :w-adverbed quote is constructed, which seems separate from your /./ issue 13:40
awwaiid re-reads about adverbial syntax and is slightly re-convinced about the : in infix:<foo>
psch ShimmerFairy: gist.github.com/peschwa/d8bd05d9ea4f3f461d73 is $ast for infix:[/./] 13:41
ShimmerFairy psch: ah, I think in that case going through nibble_to_str for the regex is wrong in the first place, unless you know how to magically convert a regex to the "correct" string :P 13:42
psch ShimmerFairy: i actually want to die for things like a regex that don't have a meaningful enough compile time value... :)
ShimmerFairy m: sub infix:[/./] { ... } 13:43
camelia rakudo-moar d0885e: OUTPUT«5===SORRY!5=== Error while compiling /tmp/hKbOKNtWLn␤Colon pair value '/./' too complex to use in name␤at /tmp/hKbOKNtWLn:1␤------> 3sub infix:[/./]7⏏5 { ... }␤»
ShimmerFairy psch: there you go, it dies already :P
psch ShimmerFairy: yes, but for the wrong reason... 13:44
m: constant $sym = "µ"; sub infix:<<$sym>> { $^a ~ "µ" ~ $^b }; say 5 µ 5 # this dies for the same reason
camelia rakudo-moar d0885e: OUTPUT«5===SORRY!5=== Error while compiling /tmp/fxaM5lC3Y1␤Colon pair value '$sym' too complex to use in name␤at /tmp/fxaM5lC3Y1:1␤------> 3constant $sym = "µ"; sub infix:<<$sym>>7⏏5 { $^a ~ "µ" ~ $^b }; say 5 µ 5 # this d␤ expecting any …»
awwaiid "Only identifiers that produce a list of one or more values (preferably strings) are allowed as name extensions" where it then goes on to mentioned not allowing {} blocks. So seems reasonable to not allow regex
psch and that *shouldn't* die
ShimmerFairy psch: I think your best bet is to stop bad values before they reach nibble_to_str, not within it (in other words, the else branch you pointed out isn't the problem directly)
psch ShimmerFairy: that'd just mean looking for compile time values before nibble_to_str instead of inside it? 13:46
ShimmerFairy psch: well, unless a regex object counts as a nibble, I don't think it should hit nibble_to_str in the first place 13:47
psch: maybe nibble_to_str could use a "is this really a nibble?" check, though, if that were feasible
13:47 vvz left 13:48 mr_ron left
timotimo oh, sorry for the trouble, everyone! :S 13:48
13:50 TEttinger left
timotimo lizmat: and thanks for building something better 13:51
lizmat: though perhaps instead of .EXISTS-KEY we could use the nqp op for that 13:52
psch ShimmerFairy: afaiu, every colonpair has to be nibbled 13:53
ShimmerFairy: so, yeah, the user is probably silly to put a regex in there
ShimmerFairy: but that's not really how we can implement it, with all the torturing around... :) 13:54
ShimmerFairy: the problem doesn't really come from nibbling in the first place either, it's just that we want a compile time value for what we nibbled in some cases
13:56 Peter_R joined 14:01 diana_olhovik_ left
colomon o/ 14:09
yoleaux 11:43Z <moritz> colomon: smoke.perl6.org/report shows a T for IO::Notification::Recursive, even though that repo has no tests; should that be a P instead?
colomon moritz: it’s entirely possible Test failures in pre-reqs count as Test failures in the base module. Unfortunately, IO::Notification::Recursive is working fine now, and I’m having trouble finding something to test it on. 14:10
14:11 FROGGS_ joined 14:15 FROGGS left 14:16 aborazmeh joined, aborazmeh left, aborazmeh joined
masak afternoon, #perl6 14:18
14:18 shlomif joined, rindolf left, shlomif is now known as rindolf 14:19 telex left
timotimo lizmat: could it be that for some reason we only need one less CALLER:: now? 14:21
14:22 mr_ron joined, telex joined, telex left 14:23 aborazmeh left 14:24 telex joined
masak huh, 007 made Hacker News. did not expect that. 14:25
HN seems to like the readme :> 14:26
14:30 laouji left 14:31 raiph left, raiph joined 14:35 laouji joined
masak added a comment: news.ycombinator.com/item?id=10327631 14:39
timotimo you're breaking the fourth wall??
14:40 laouji left
masak timotimo: you mean, completely unlike the README? :P 14:41
timotimo %) 14:42
14:44 Xor_ left
timotimo re-read the ufo readme, too 14:45
14:45 skids joined
timotimo i hadn't remembered it to be so good :3 14:45
14:46 zakharyas joined, liztormato joined
masak thanks :) 14:47
liztormato timotimo: could well be
masak heh, ufo is so old it predates me (and Github) doing Markdown ;)
timotimo yeah 14:48
liztormato: huh, removing one of the CALLER:: doesn't fix it; but maybe i was working on the wrong one
oh, wait
we generate the stack trace inside that one function where we try { die() }
14:51 laouji joined 14:52 tokuhiro_ joined 14:53 Xor_ joined
timotimo liztormato: in Backtrace.pm i was able to replace "2 + $offset" with "1 + $offset" and it fixed fail.t, probably breaks everything else, so i expect some caller should supply one less to the $offset parameter 14:53
liztormato It's all relatiy :-) 14:54
*relative 14:55
14:56 laouji left
timotimo related, too 14:56
14:56 tokuhiro_ left 14:58 diana_olhovik_ joined 15:00 patrickz joined 15:02 Xor_ left
timotimo # Failed test 'When $! is set, die() is die($!).' 15:03
grondilu masak++ 007 story 15:05
I cracked at ""Did you know that macros allow introspection and modification of program elements from within the program itself?",
timotimo liztormato: that failure is fixed by removing one CALLER:: from sub die 15:06
masak grondilu: :)
timotimo liztormato: maybe we accidentally made all of this a whole bunch better? we need one less lookup now :)
liztormato Perhaps. :-) 15:08
15:09 diana_olhovik_ left
jnthn If you're going to use nqp::ctxcaller and so on, you really need to nqp::existskey on the result. 15:11
liztormato Ah, duh! 15:12
jnthn ;)
jnthn is relaxing a good bit this weekend after last week's teaching (nice course, great students, but exhausting), but will be back to Perl 6 hacking tomorrow :) 15:13
(For today, just dropping by the keyboard once in a while :)) 15:14
15:14 ely-se left, Xor_ joined
liztormato is also relaxing between cycling  15:14
skids timotimo: FWIW to the original patch, Failures should not say "Died" they should say "Failed" 15:17
masak .u 
yoleaux No characters found
masak jnthn++ # relaxing 15:18
15:22 Xor_ left 15:24 Xor_ joined
leont Great, yet another issue with :temp 15:27
jnthn ...:temp?
leont In rules
timotimo skids: did i break that? :o 15:28
oh
skids Yeah just a pasto
timotimo i suppose i copy-pasted it over withotu fixing it
leont I'm doing <directive>\n:temp $foo = $<directive>.ast\n<bar>, but it seems the temp is executed before <directive> is matched, so it's not getting the data it needs
Putting a code block in between seems to force things to work 15:29
jnthn There's a bit in S05 about when $/ is updated and...yes, you may need to do that.
leont Seems even an empty code-block fixes it 15:30
jnthn Aye, that's considered a sequence point 15:31
design.perl6.org/S05.html#When_%24%2F_is_valid
colomon masak++ 15:36
15:36 patrickz left 15:37 liztormato left
masak HN wanted unquotes in 007. anyone want to send me a patch? ;) 15:39
timotimo HNDD? 15:40
15:41 domidumont joined 15:42 ely-se joined 15:43 Xor_ left, Suman joined
Suman I want perl 6 to be more focussed on scientific computing, data analysis, data visualisation and machine learning. 15:44
15:45 jasanj``` joined, mr_ron left
Suman Python has grown up from a general purpose language to a language suitable for scientific computing. 15:45
This age is all about big data
15:45 mr_ron joined
arnsholt It works the other way, I think 15:46
Having good ML and such is basically a question of having a module ecosystem large enough to support bindings to BLAS and LAPACK and friends
Suman We should be able to handle big big data with perl 6 (like exabytes of data)
arnsholt And the stuff that goes on top of those to make working them non-annoying 15:47
Suman And Apache spark is famous not because of anything but its fast distributed scalable computing
so fast scalable distributed computing should be our aim too 15:48
leont Suman: so what are you going to do to make that happen?
15:49 jasanj`` left
Suman leont Personally I am a perl fan. Currently I use R and python for my work. In fact I am consumer of such softwares. I am afraid I can't do much in writin gsoftware. However I can spread the word and support via donation. Thanks 15:50
timotimo for scientific computing we're getting shaped arrays of native integers and doubles - the support is in, only the syntax for definitions isn't sugared yet 15:52
that's a big part of what makes scientific computing dobale, because vectors and matrices are basically all over the place
we already have arbitrary precision rational numbers on top of our regular rational numbers that overflow into doubles when it's no longer possible to hold the denominator in 64bit 15:53
that ought to be a good thing for numerics, i suppose?
jnthn Suman: It also can help to identify concrete things that are missing and give an idea of how much they matter. I mean, sure, we're probably missing plenty of stuff for people to be able to use Perl 6 in that area. But as somebody who works on compiler/VM level stuff for Perl 6, I just don't have the knowledge of what scientific computing folks tend to need. timotimo seems to hae more clue than me, maybe... :)
leont A lot of scientific computing in higher level languages are really abstractions around highly optimized pieces of C and Fortran code 15:54
timotimo handling exabytes of data is much more about what facilities we offer to directly work with operation system APIs directly, and since those are basically always doable from NativeCall, we already have that, so to speak
leont R, PDL and NumPy/SciPy all do that
timotimo that's right
leont Science is actually keeping fortran alive that way :-o 15:55
Suman timotimo true agreed but perl 6 should natively support data import (in csv, excel,hdfs,tsv,spss,sas formats), analysis(descriptive and inferential statistics)
leont Suman: none of those should be particularly hard to implement, really
Suman For the data visualisation part we can use python matplotlib API
timotimo i don't think perl6 should support csv "natively" 15:56
or excel, hdfs, tsv, spss, sas
that's for modules to do
just like your processor can't handle xml directly ... oh, shit, wait ... :P 15:57
Suman Without having good packages for data analysis, perl 6 won't go anywhere
timotimo agreed
leont Suman: yes, and those will be written.
Suman Rightly said data analysis is the sexiest job of 20th century
leont Now is the time to start writing them, really
15:57 spider-mario left
timotimo also, i totally agree with what arnsholt said about BLAS and LAPACK 15:57
15:58 tokuhiro_ joined
timotimo also, we'll want someone to write modules that make OpenCL and friends easy to use 15:58
15:58 Peter_R left
Suman I am afraid that even perl 5 doesn't have good easy to use packages for data analysis like pandas, numpy in python 15:58
skids Data anaysis is the sexy job that will make security continue to be the sexiest job. :-)
Suman We may be able to learn something from them
timotimo … can we not use the word "sexy" for this? :\
Suman we may be abe to build more powerful
15:59 Xor_ joined
timotimo gist.github.com/timo/b5e0c8c167a6079cbac6 - what controls whether or not a stack trace would show up? 16:00
Suman leont "Science is actually keeping fortran alive that way :-o" Agreed fully we want perl 6 to thrive that way for next century 16:02
16:02 cognominal joined, tokuhiro_ left, Peter_R joined 16:04 vvz joined
skids timotimo: around Exception.pm line 260 is where to start figuring that out. 16:05
timotimo ah 16:06
16:06 domidumont1 joined
timotimo no, i don't think so 16:06
skids The notable exception (no pun intended) to that is if you are dealing with a Failure and its double backtrace.
timotimo that seems to be what's responsible for when an exception hits the top level
look at the $comp.^add_method calls below that 16:07
AFK for a little bit
16:08 araujo joined 16:09 domidumont left
skids timotimo: at the top level is when the backtrace actually happens -- it has to reach there first to ensure nothing tried to handle it. IIRC. 16:11
leont What is the state of not using strings as keys for hashes? 16:13
skids leont: you have to declare specially, is all.
m: :{ 6 => 2 }.keys[0].WHAT.say 16:14
camelia rakudo-moar d0885e: OUTPUT«(Int)␤»
psch m: my %h{Any}; %h{5} = 5; say %h.keys[0].WHAT
camelia rakudo-moar d0885e: OUTPUT«(Int)␤»
masak m: say grep { /\.pl6/ }, <a.pl6 b.txt> 16:17
camelia rakudo-moar d0885e: OUTPUT«Method 'match' not found for invocant of class 'Any'␤ in block <unit> at /tmp/XGpCAMzLWP:1␤␤»
masak m: say grep /\.pl6/, <a.pl6 b.txt>
camelia rakudo-moar d0885e: OUTPUT«(a.pl6)␤»
leont Ah! the colon is the trick is seems
masak that's from p6u
surely the block form used to work, too?
m: say grep { $_ ~~ /\.pl6/ }, <a.pl6 b.txt>
camelia rakudo-moar d0885e: OUTPUT«(a.pl6)␤»
masak is this a "masak submits rakudobug" situation?
mr_ron m: '123' ~~ /(123)/; say ~ $0, ' ', $0.WHAT, ' ', so $0.isa(Cool); # mighti mean documentation update 16:19
camelia rakudo-moar d0885e: OUTPUT«123 (Match) True␤»
mr_ron I want to add Match to the list of Cool descendants here: github.com/perl6/doc/edit/master/l...ol.pod#L20 16:20
any objections?
skids m: sub s1 { sub s2 { fail("foo"); }; s2()(); }; s1();
camelia rakudo-moar d0885e: OUTPUT«foo␤ in sub s1 at /tmp/tJMRwF6eCx:1␤ in block <unit> at /tmp/tJMRwF6eCx:1␤␤Actually thrown at:␤ in sub s1 at /tmp/tJMRwF6eCx:1␤ in block <unit> at /tmp/tJMRwF6eCx:1␤␤»
16:21 xtreak joined
skids Yeah that definitely got broken. Should report s2 in the first backtrace. 16:21
16:21 xtreak left
AlexDaniel NativeCall was already mentioned, but it should also be noted that all these possibilities to extend perl6 syntax should also help a lot 16:22
e.g. unicode operators and stuff like that 16:23
so, modules will happen, I don't think that there is too much stuff that is missing from perl6 core.
psch masak: irclog.perlgeek.de/perl6/2015-09-23#i_11262003 16:24
leont m: my %foo = 1 => 2; %foo.perl.WHO.say
camelia rakudo-moar d0885e: OUTPUT«Method 'pull-one' must be implemented by ProcessStr because it is required by a role␤ in any compose_method_table at gen/moar/m-Metamodel.nqp:2666␤ in any apply at gen/moar/m-Metamodel.nqp:2676␤ in any compose at gen/moar/m-Metamodel.nqp:2812␤ …» 16:25
leont Looks like a bug to me
skids: how do I create that without the :{} syntax?
Or to un-XY, I have a list of Pairs, I want to make such a hash, how I do that? 16:26
dalek c: 4055f38 | (Ronald Schmidt)++ | lib/Type/Cool.pod:
Update Cool.pod

Add Match to list of descendants of Cool		 16:28
skids Hrm good question -- does pairup stringify?
masak psch: thank you. 16:29
psch m: ( 5 => 5 ).key.WHAT.say
camelia rakudo-moar d0885e: OUTPUT«(Int)␤»
skids oh no pairup is for lists --> pairs m.
m: (5 => 6, 7 => 8).hash.perl.say 16:30
camelia rakudo-moar d0885e: OUTPUT«{"5" => 6, "7" => 8}␤»
psch m: my @pairs = 1 => 2, 3 => 4; my %h{Any} = @pairs; say %h.perl
camelia rakudo-moar d0885e: OUTPUT«(my Any %{Any} = 1 => 2, 3 => 4)␤»
psch m: my @pairs = 1 => 2, 3 => 4; my %h{Any} = @pairs; say %h.keys[0].WHAT 16:31
camelia rakudo-moar d0885e: OUTPUT«(Int)␤»
masak m: my $_ = "foo"; /foo/; say so $/ 16:32
camelia rakudo-moar d0885e: OUTPUT«Potential difficulties:␤ Redeclaration of symbol $_␤ at /tmp/5NnB0cdYhg:1␤ ------> 3my $_7⏏5 = "foo"; /foo/; say so $/␤True␤»
masak m: $_ = "foo"; /foo/; say so $/
camelia rakudo-moar d0885e: OUTPUT«True␤»
masak if that works, then why doesn't the .grep({ /foo/ }) thing work? 16:33
or, more to the point -- if that works, then why *shouldn't* the grep thing work?
I think .grep({ /foo/ }) is how many p5ers would write it, and expect it to work 16:34
jdv79 RabidGravy: what did you do?
masak and if we break that expectation, I think there should be a good reason, more than just "it was an accident that it ever worked" or "there's an easier way, learn that"
16:34 raiph left
RabidGravy jdv79, I appllied your patch, it worked I pushed it :) 16:35
16:35 dhj joined
jdv79 you also typo'ed my email addr but ok 16:36
how would you test that?
are we able to redefine recv?
16:44 Suman left
skids m: <foo bar foog>.grep({$_.WHAT.say; /foo/}).say # masak: A good reason would be "because there's no $_ in a parameterless block" but that does not seem to be the happening reason 16:44
camelia rakudo-moar d0885e: OUTPUT«(Str)␤Method 'match' not found for invocant of class 'Any'␤ in block <unit> at /tmp/1wn0Il839j:1␤␤»
16:45 dhj left 16:51 diana_olhovik joined 16:54 zakharyas left 16:59 yqt joined
moritz skids: no, the reason is that the grep block returns the regex 17:01
skids: and grep itself boolifies it
skids: but at that time, the block's $_ has ceased to exist
timotimo leont: in case it got lost in the shuffle, the most basic thing you'll want to know is you can define a hash that won't stringify keys by default with the "my %h{Any}" syntax; the :{ ... } thing is sugar for creating such a parameterized hash literal 17:02
masak moritz: outside of a .grep block, a regex in an empty statement still seems to match against $_. but maybe the difference is that that regex is in void^Wsink context, and the one in the .grep block isn't? 17:04
dalek href="https://perl6.org:">perl6.org: 2772cd4 | (Zoffix Znet)++ | / (2478 files):
Get archive out of the way
perl6.org: 244774c | (Zoffix Znet)++ | / (22 files):
perl6.org: First styled home page
17:04 dalek left
moritz masak: aye 17:04
17:05 dalek joined, ChanServ sets mode: +v dalek
skids m: <foo bar foog>.grep({/foo/; $/;}).say # moritz is right 17:05
camelia rakudo-moar d0885e: OUTPUT«(foo foog)␤»
cognominal m: say sort <== < 1 2 3 > # <== in term position does not work. Am I doing it wrong, or is this a known problem?
camelia rakudo-moar d0885e: OUTPUT«5===SORRY!5=== Error while compiling /tmp/7QJ5Qm3IKm␤Unable to parse expression in quote words; couldn't find final '>'␤ ␤at /tmp/7QJ5Qm3IKm:1␤------> 3ng it wrong, or is this a known problem?7⏏5<EOL>␤ expecting any of:␤ …»
masak moritz: the { /foo/ } device stopped working because we got better at detecting sink context? 17:06
17:06 leont left
El_Che something you've heard a million time: For Hash types isn't confusing to have adverbs and methods doing the same thing? Specially :delete is pretty exotic(that performs an action, something I would expect to be done by a method). Or is there something I haven't understand yet? 17:06
17:07 khw left
El_Che (Looking at perl6advent.wordpress.com/2013/12/...y-adverbs/ , perl6advent.wordpress.com/2013/12/...-only-way/ and doc.perl6.org/type/Hash ) 17:07
skids std: say sort <== < 1 2 3 >
camelia std 28329a7: OUTPUT«ok 00:00 135m␤»
skids Well, feeds might be in flux on the backend but they should probably parse better than that. 17:08
17:08 jsimonet left
masak El_Che: sounds like the missing bit is that (a) it's nice/powerful to use the same .{} hash lookup even when we're deleting or checking for existence 17:09
timotimo El_Che: we use an adverb for that because postcircumfix:<{ }> and postcircumfix:<[ ]> are for selecting a bunch of things, then you put the delete adverb in there to specify what happens with it
masak (b) and in order to do that, we put the adverbs on the lookup itself 17:10
timotimo El_Che: the other thing is we have some things that support both positional and associative access, so it would have to be .delete_key and .delete_pos methods, rather than just a .delete method
moritz also, this way you can combine the auto-quoting of <...> with :delete
which a mere method wouldn't provide
masak El_Che: maybe it's easier to think of %h{'foo'}:delete as a "destructive read"? :) 17:11
right, in some sense we're retaining all the power of .{} and .<>
skids m: {:a(1),:b(2)}<a>:kv:delete.say # and combining adverbs
camelia rakudo-moar d0885e: OUTPUT«(a 1)␤»
El_Che Last sentence of masak made my head explode :)
masak El_Che: well, :delete *does* still return the value.
El_Che that's true 17:12
masak which means that *technically*, it's a kind of lookup :)
that's all I meant
my point is that :exists and :delete don't feel all that misplaced as adverbs. because (no matter what data structure you have) if you want to check for existence or remove something from a collection, first you have to *find* it. i.e. lookup 17:13
having :exists and :delete as adverbs on ordinary lookup means that those two related operations benefit from all the (syntactic) power and flexibility that we already have with .[], .{} and .<> 17:14
which doesn't mean so much when you're just deleting individual items, sure. but it pays off whenever you want to do something more than that. 17:15
m: my %letters = "a".."z" X=> 1; my @vowels = <a e i o u>; %letters{@vowels} :delete; say %letters.keys.sort 17:16
camelia rakudo-moar d0885e: OUTPUT«(b c d f g h j k l m n p q r s t v w x y z)␤»
17:18 adhoc left 17:20 adhoc joined
masak m: my @deck = <♠ ♥ ♦ ♣> X~ flat(1..10, <j q k>); my @hand = @deck.pick(5); say @hand; @deck = (@deck (-) @hand).list; say @deck.elems 17:23
camelia rakudo-moar d0885e: OUTPUT«[♣6 ♦k ♠4 ♥7 ♥10]␤47␤»
masak that last one didn't lead me where I planned :) could someone re-write it to use .{} :delete, maybe?
17:25 Xor_ left 17:26 araujo left 17:27 araujo joined 17:28 Xor_ joined 17:30 araujo left
BenGoldberg m: my %deck = SetHash.new((<♠ ♥ ♦ ♣>) X~ flat(1..10, <j q k>)); my @hand = %deck.keys.pick(5); @hand.say; %deck{@hand}:delete; %deck.elems.say; 17:32
camelia rakudo-moar d0885e: OUTPUT«[♦j ♦6 ♦4 ♦7 ♠2]␤47␤»
skids m: my %deck; %deck{<♠ ♥ ♦ ♣> X~ flat(1..10, <j q k>)} = (); my %hand = %deck{%deck.keys.pick(5)}:delete:kv; say %hand.keys; say +%deck.keys
camelia rakudo-moar d0885e: OUTPUT«(♥q ♣8 ♠3 ♦9 ♣7)␤47␤»
masak cute. skids++ 17:33
also, BenGoldberg++
skids: I don't think I've seen the `= ()` idiom before. 17:34
BenGoldberg perl5, of course
skids Oh that goes way back to P5
masak oh, ok.
I'm so used to just filling 1s in there. I'd be more likely to write `= 1 xx *` 17:35
17:35 laouji joined 17:37 araujo joined, araujo left, araujo joined 17:40 laouji left
skids m: my @deck = <♠ ♥ ♦ ♣> X~ flat(1..10, <j q k>); my @hand = @deck[@deck.keys.pick(5)]:delete; say @hand; say +@deck # 52? 17:40
camelia rakudo-moar d0885e: OUTPUT«[♠j ♣3 ♣6 ♥j ♠5]␤52␤»
skids oh right. dead slots. 17:41
17:45 vendethiel joined 17:46 pmurias_ is now known as pmurias
pmurias are lines over 80 characters a serious problem for anybody in (the js runtime part) of nqp-js? 17:46
El_Che masak: I get the reasoning (it makes sense). But you'll get a lot of confused people :) 17:47
pmurias I'm making an exception to the google coding standard for the line length but if somebody is hindered by overly long lines I can shorten them
dalek p: 5ca7473 | (Pawel Murias)++ | src/vm/js/nqp-runtime/core.js:
[js] remove dead commented out code
 17:48
p: 33c9123 | (Pawel Murias)++ | src/vm/js/nqp-runtime/ (12 files):
[js] Run fixjsstyle src/vm/js/nqp-runtime/*.js
p: 5522fb4 | (Pawel Murias)++ | src/vm/js/nqp-runtime/ (12 files):
[js] Run fixjsstyle --strict src/vm/js/nqp-runtime/*.js
p: 779dc95 | (Pawel Murias)++ | tools/build/ (2 files):
[js] set the maximum line length while linting to 200
p: c67d5bc | (Pawel Murias)++ | src/vm/js/nqp-runtime/ (5 files):
[js] Fix stuff that js-lint complains about. Comply to the google js coding (except allow longer lines).
17:52 travis-ci joined
travis-ci NQP build failed. Pawel Murias '[js] Fix stuff that js-lint complains about. Comply to the google js coding (except allow longer lines).' 17:52
travis-ci.org/perl6/nqp/builds/83577902 github.com/perl6/nqp/compare/c28a6...7d5bc312fe
17:52 travis-ci left, yqt left 17:58 diana_olhovik left 18:03 Xor_ left 18:07 rivarun left
dalek href="https://perl6.org:">perl6.org: 0d56442 | moritz++ | includes/footer:
Remove Niecza from footer

it is sufficiently outdated that we do not need to advertise it on every single page		 18:09
href="https://perl6.org:">perl6.org: 471fc02 | moritz++ | includes/footer:
Remove wiki link from footer
 18:11
href="https://perl6.org:">perl6.org: e7ab176 | moritz++ | source/specification/index.html:
Use macros instead of concurrency as ane example of stuff in flux

I think concurrency is not that much in flux anymore, in large parts thanks to jnthn++		 18:12
moritz Zoffix: it seems faq.perl6.org uses the perl6.org style sheet, and now has horrible contrast. Would be nice if you could fix it 18:13
(if not, I'll simply copy the old perl6.org style sheet to faq.perl6.org, but that would be a less-than-awesome solution) 18:14
news.ycombinator.com/item?id=10325670 a hn discussiono without any hate 18:17
18:18 Xor_ joined
El_Che I heard today out of the blue how someone was impressed by Perl6 during Ovid's intro talk. Good work, p6 people! 18:25
18:26 FROGGS[mobile] left
jdv79 how to do something like (in p5 parlance) *IO::Socket::INET::recv = sub {}? 18:29
18:29 FROGGS_ is now known as FROGGS, diana_olhovik joined
FROGGS jnthn: what do you mean, "like what happens when assigning through a pseudo-package" ? 18:29
jnthn: can you give me an example?
masak El_Che: confusing people (in the short run) if the reasoning makes sense (in the long run) sounds like a fair deal to me. :)
18:30 lichtkind joined
El_Che masak: expect some good trolling on that :) 18:32
moritz expects much more trolling in other areas 18:34
18:35 BenGoldberg left, BenGoldberg joined 18:36 BenGoldberg left, BenGoldberg joined
carlin m: my $fh = '/dev/urandom'.IO.open; Thread.start({ $fh.read(16) }); 18:37
camelia rakudo-moar d0885e: OUTPUT«(signal SEGV)»
vendethiel carlin: I'm pretty sure you can't do that in the "safemode" 18:38
18:38 diana_olhovik left
carlin I get a segv doing that locally too 18:40
FROGGS jnthn: if you are talking about 'my Int:D $a; MY::<$a> = Int' that works, I mean fails 18:53
jdv79 what's a good way to get the passed args?
FROGGS m: sub foo(Int:D $a?) { say $a }; foo() 18:57
camelia rakudo-moar d0885e: OUTPUT«Parameter '$a' requires an instance of type Int, but a type object was passed. Did you forget a .new?␤ in sub foo at /tmp/9XvkFVL6AM:1␤ in block <unit> at /tmp/9XvkFVL6AM:1␤␤»
FROGGS jdv79: what args?
jdv79 passed to a routine 18:58
i got it - @_ and %_ seem ok
now wondering wny a wrap && callsame don't work
shouldn't that?
FROGGS dunno offhand 18:59
FROGGS is sleepy
jdv79 gist.github.com/anonymous/5fdaad8a37bae71abe8b 19:01
vendethiel oooh, that new landing page is great.
jdv79 with the bile colored header? 19:02
FROGGS jdv79: what are you trying to do? maybe augmenting IO::Socket::INET helps? or subclassing it... 19:03
jdv79 i want to log recev calls
i can't. the thing i'm trying to test doesn't do DI 19:04
is wrapping a method not that simple ^H?
FROGGS I thought it was
vendethiel m: 5->6 19:05
camelia rakudo-moar d0885e: OUTPUT«5===SORRY!5=== Error while compiling /tmp/lOOZ4nw9yV␤Unsupported use of -> as postfix; in Perl 6 please use either . to call a method, or whitespace to delimit a pointy block␤at /tmp/lOOZ4nw9yV:1␤------> 035->7⏏056␤ expecting any of:…»
FROGGS I also cant remember when to use callsame vs. nextsame
jdv79 oh, i'll try that one
timotimo callsame when you have code after it and want to do something with return value 19:06
nextsame is more like a goto
jdv79 that's what i thought. so callsame is correct here. 19:07
timotimo going to see inside out now :) 19:08
jdv79 huh?
19:09 zakharyas joined
masak jdv79: www.imdb.com/title/tt2096673/ 19:12
jdv79 why do i have to give a sig on the wrapper routine? 19:15
19:26 domidumont joined 19:27 krunen_ joined, ChoHag_ joined, maddingu1 joined, pyrimidi_ joined, apotheon_ joined 19:28 pochi_ joined, diego_k joined, obra_ joined, pink_mis1 joined, carlin_ joined 19:29 leedo_ joined, domm_ joined, ashleyde1 joined, carlin_ is now known as Guest49363, domidumont1 left 19:30 pyrimidine left, ChoHag left, krunen left, pochi left, pink_mist left, takadonet left, pink_mis1 is now known as pink_mist, leedo left, FROGGS left, prammer left 19:31 domidumont left, stux|RC-only joined, pink_mist left, pink_mist joined, takadonet joined 19:32 charsbar joined, prammer joined, carlin left, maddingue left, apotheon left, domm left, Debolaz left, obra left, FROGGS joined, Debolaz joined 19:33 LittleC47 joined, lestrrat joined, LittleC47 left 19:37 mr_ron left, yqt joined 19:38 mr_ron joined 19:55 zakharyas left
jdv79 m: my @foo = (1,2,3); @foo.perl.say; (1,2,3).perl.say; 19:57
camelia rakudo-moar d0885e: OUTPUT«[1, 2, 3]␤(1, 2, 3)␤»
jdv79 why the diff?
oh
FROGGS m: my @foo = (1,2,3); @foo.WHAT.say; (1,2,3).WHAT.say; 19:58
camelia rakudo-moar d0885e: OUTPUT«(Array)␤(List)␤»
jdv79 aw, subbuf doesn't do *-4 type stuff 19:59
20:02 kaare__ left
masak jdv79: sounds like a bug to me 20:07
20:08 xfix left
FROGGS m: class Foo { has Int:D $.bar }; say Foo.new.bar 20:18
camelia rakudo-moar d0885e: OUTPUT«(Int)␤»
lichtkind i had made a token but <.tokenname> didnt matched?? 20:21
yoleaux 3 Oct 2015 16:40Z <Ven> lichtkind: shouldn't that be a regexp? github.com/perl6/problem_solver_tu...n4.pl6#L25
20:21 Psyche^ joined
FROGGS m: my ($a, Int:D $b); say $b 20:21
camelia rakudo-moar d0885e: OUTPUT«(Int)␤»
lizmat masak: re .grep: { /foo/ } last I checked with jnthn he said it was never supposed to work
lichtkind yoleaux exactly i had pronlem there but solven jsut not comited now 20:22
lizmat and yes, I broke it when I iterified .grep
20:22 darutoko left
lichtkind yoleaux but why isnt that a regrx in your opinion? 20:23
lizmat .tell jnthn should VM supported object hashes be added to nope.md ? 20:24
yoleaux lizmat: I'll pass your message to jnthn.
dalek href="https://perl6.org:">perl6.org: dd78e8a | (Zoffix Znet)++ | source/index.html:
Fix overflow on download button (Closes #23)
FROGGS lichtkind: yoleaux is a message bot
lichtkind FROGGS nobody told me :)
20:24 Psyche^_ left
FROGGS .botsnack 20:25
yoleaux :D
FROGGS :o)
masak lizmat: I understand and I agree that with jnthn it's weird that it ever worked.
lizmat nevertheless, it appears to be a common mistake to make
masak lizmat: the question is more, what do we do now? do we make it work, or add a FAQ somewhere explaining to people why they are thinking wrong?
lizmat so maybe we should have a less LTA error message ?
masak specifically, this seems like something a p5er would be likely to run into 20:26
lichtkind FROGGS but dont know why <.rulename> doesnt match automatically
20:26 mr_ron left
dalek kudo-star-daily: 363cbfd | coke++ | log/ (8 files):
today (automated commit)
 20:26
lizmat m: <a b c>.grep: { /foo/ }
camelia rakudo-moar d0885e: OUTPUT«Method 'match' not found for invocant of class 'Any'␤ in block <unit> at /tmp/5yqI5GtvwG:1␤␤»
dalek rl6-roast-data: 7db26a5 | coke++ | / (9 files):
today (automated commit)
masak just as p5ers might do `$s1 . $s2`
lizmat if we add a method Match to Any, we could have it suggest something sensible ?
FROGGS lichtkind: because of a bug... use <&space>
lichtkind so its the name , if i named it other thatn space it would work? 20:27
FROGGS lichtkind: no, it would look up a method Cursor.theregexname 20:28
lichtkind so just waite till fix? 20:30
20:30 araujo left
dalek kudo/nom: b6247a8 | lizmat++ | src/core/Map.pm:
Expose MapIterator role to the world

We can use it in other places in the core		 20:32
kudo/nom: d2da732 | lizmat++ | src/core/Baggy.pm:
Iterify Baggy.pairs/keys/kv/values/antipairs/kxxv

I have two reasons for doing this:
  - to make Bag/BagHash/Mix/MixHash faster
  - *if* we don't get VM support for object hashes, this code can be
   put in a role and used for a single hash refactoring of object
   hashes in Perl 6
This refactor also exposed two faulty tests for Baggy.kxxv
20:32 mr_ron joined
lizmat hmm.. actually faulty tests for Baggy.kv 20:33
well, actually tests for Baggy.kv that depended on faulty Baggy.kv behaviour 20:35
dalek ast: 778d230 | lizmat++ | S02-types/ (2 files):
Fix faulty BagHash/MixHash.kv test
 20:38
20:45 zjmarlow joined
dalek q: ca5b7ed | (Zoffix Znet)++ | web/index.tmpl:
Make FAQ work with new Perl6.org design
 20:47
Zoffix moritz, done.
I think this one might be outdated? github.com/perl6/faq/blob/master/a...iple-lines 20:48
20:57 Xor_ left, Xor_ joined 21:06 laouji joined 21:11 laouji left
lizmat m: my %h = a => 42; %h.values[0]++; dd %h # apart from the LTA error message, we deem this to be acceptable, right ? 21:11
camelia rakudo-moar d2da73: OUTPUT«Parameter '$a' expected a writable container, but got Int value␤ in block <unit> at /tmp/t9L58DdiJG:1␤␤»
lizmat m: my %h = a => 42; %h.pairs[0].value++; dd %h # but we allow this pairs to have a lvaluable Pair 21:12
camelia rakudo-moar d2da73: OUTPUT«Hash %h = {:a(43)}␤»
lizmat masak jnthn TimToady ^^^ 21:13
21:20 mr_ron left 21:21 pmurias_ joined
masak lizmat: let's see. S02 says that Pair is mutable in value but not in key. so your second eval seems kosher. 21:22
lizmat m: my %h = a => 42; $_++ for %h.values; dd %h #hmmmm
camelia rakudo-moar d2da73: OUTPUT«Hash %h = {:a(43)}␤»
lizmat so maybe subscripting .values is at fault ?
masak yes, mebbe.
maybe it should yield an array of aliased values. 21:23
lizmat: want me to rakudobug it?
lizmat masak: not yet
masak hokay
lizmat masak: ok, Pair.value is a left value, ok I get that
but does that also mean it changes the hash ?
21:24 pmurias left
lizmat m: my %h = a => 42; %h.values.AT-POS(0)++; dd %h # just testing 21:25
camelia rakudo-moar d2da73: OUTPUT«Parameter '$a' expected a writable container, but got Int value␤ in block <unit> at /tmp/GaDEu6f8dz:1␤␤»
masak I don't see why not. the Pair is conceptually in the hash
making it a Pair with copied values feels counterintuitive
lizmat Bag.pairs should have ro values on the Pairs 21:26
21:26 rindolf left
lizmat otherwise you could alter the Bag 21:26
masak +1 21:27
Bags are immutable. Hashes are mutable.
lizmat so I was wondering maybe we should have a values-ro and pairs-ro method on Map 21:28
masak but the choice is not the user's, it's the type's
so a Bag would just return Enums, not Pairs 21:29
no need for separate methods
21:30 Xor_ left
masak works out Liskovly and everything 21:30
lizmat m: my $b = <a b c>.BagHash; $b.pairs[0].value = "foo"; dd $b # another thing to fix :-(
camelia rakudo-moar d2da73: OUTPUT«BagHash $b = ("a"=>foo,"c"=>1,"b"=>1).BagHash␤»
lizmat m: my $b = <a b c>.Bag; $_++ for $b.values; dd $b # another one 21:32
camelia rakudo-moar d2da73: OUTPUT«Bag $b = ("a"=>2,"c"=>2,"b"=>2).Bag␤»
masak aye 21:34
immutability is hard :/
I guess this is an instance of the "don't leak access rights through your getter methods" best practice 21:35
BenGoldberg Just deep clone everything we return ;) 21:40
masak aye
I first learned about this in www.amazon.com/Effective-Java-2nd-J...0321356683 -- item #39 -- "make defensive copies when needed"
that's for the immutable types
for the mutable types, we actually want the aliasing behavior, and so we don't copy
lizmat yeah, gotcha 21:41
AlexDaniel what shell does qqx// use on windows? For example, the spec says this about “shell()”: “shell goes through the system shell (cmd on windows, /bin/sh on Unixish systems), thus interpreting all the usual shell meta characters.” 21:42
masak lizmat: not a criticism of you or anyone in particular. I just find this immutability stuff fascinating.
AlexDaniel I can't really find any meaningful description of QX though
lizmat masak :-)
feels like torture to me :-)
21:43 zjmarlow left
masak I sincerely believe it's one of the places in software engineering where we're "not there yet". and we'll look back in 2035 and go "how the heck did we get anything done with those paltry/wrong primitives?" 21:43
I see early attempts at getting it right, with linear types, immutable object graphs, etc. none of the solutions are in that sweet spot of highly useful and easy yet, though.
21:44 Xor_ joined
BenGoldberg ponders whether we should "merely" steal the 'const' keyword from C / C++ 21:45
AlexDaniel so, in other words, qx// is not in any way portable? (although making assumptions about the underlying shell is stupid anyway)
masak in a sense, I think we need better metaphors. for example, the Pair we extract from a Hash is like an "alias" or a "remote control" of the contents of the Hash. it's like a little portable one-entry hash slice.
BenGoldberg: the 'const' keyword in C/C++ does solve some issues, but it introduces others. 21:46
21:47 alpha- joined
masak BenGoldberg: for example, programmers have a hard time reasoning about the immutability of the *variable*, the *container* it refers to, and its *value* which may be an entity with properties, etc 21:47
21:47 ely-se left, ^elyse^ joined
masak BenGoldberg: this is the problem I foresee with ES6's new `constant` keyword. people think it does one (fairly strong) thing, but it actually does another (fairly weak) thing 21:48
BenGoldberg Also, if you want a container which is const, and you want to make it cache/memoize computations, it's tricky.
skids suspects eventual hardware ref-counting/following assist will play a role in what such things look like in a decade or two. 21:49
When "stop the world" GC starts to chafe. 21:50
21:50 bjz left, ^elyse^ left
BenGoldberg On a completely unrelated note... 21:51
m: gist.github.com/BenGoldberg1/bc39a...aeb733012a 21:52
camelia rakudo-moar d2da73: OUTPUT«Cannot send a message on a closed channel␤ in block <unit> at /tmp/JFN8o88RyH:31␤␤»
BenGoldberg m: gist.github.com/BenGoldberg1/bc39a...aeb733012a
camelia rakudo-moar d2da73: OUTPUT«Cannot send a message on a closed channel␤ in block <unit> at /tmp/qiBNJLKQw7:31␤␤»
BenGoldberg m: gist.github.com/BenGoldberg1/bc39a...aeb733012a
camelia rakudo-moar d2da73: OUTPUT«Attempt to return outside of any Routine␤ in block <unit> at /tmp/O3b50H1yjz:31␤␤»
BenGoldberg Could someone tell me what's happening here? 21:53
lizmat BenGoldberg: not sure, but if I understand jnthn's ideas correctly, earliest is not going to be much longer in this world 21:54
react { } would be a better approach, I think
although I must admit not seeing how you would convert your code to use react instead offhand
masak yes, `earliest` is going away; what lizmat said 21:56
21:56 mr_ron joined
dalek kudo/nom: efbd9ca | lizmat++ | src/core/ (2 files):
No need to special case Bag/Mix.antipairs
 22:01
kudo/nom: b14723d | lizmat++ | src/core/Exception.pm:
Don't mention nameless scalars
kudo/nom: 2765b6a | lizmat++ | src/core/ (4 files):
BagHash/MixHash accept values of correct type only
22:03 tokuhirom joined
dalek href="https://perl6.org:">perl6.org: 7973405 | (Matt Oates)++ | source/index.html:
.IO.words is a little more idiomatic
 22:04
skids m: role Foo { }; class A does Foo { }; class B is A { }; my B $b.= new; ("roles:", $b.^roles).say; $b.^role_typecheck_list.say; say B ~~ Foo; # headscratch -- what magic makes B ~~ Foo find roles from A 22:05
camelia rakudo-moar d2da73: OUTPUT«(roles: ((Foo)))␤()␤True␤»
22:07 tokuhirom left 22:08 pmurias_ left 22:10 Xor_ left
Ulti Zoffix you might want to use the updated VIM markup on the P6 source on the front page it's not rendered that well atm especially the custom operators bit, then you get all the types highlighted etc. mattoates.co.uk/files/vim_p6_markup.html 22:12
BenGoldberg lizmat, Reading about supplies hurts my head -- I'll need to see more examples before it makes sense to me. 22:13
Zoffix Ulti, how did you generate that? I've used pygments.org/ and yeah, I noticed it didn't do it 100% correct.
22:13 bjz joined
lizmat BenGoldberg: have you looked at jnthn's slides of his last concurrency talk? 22:13
Ulti you can use vim from the command line with the TOhtml command
Zoffix Ulti, ah. Thanks. I'll look into that tomorrow. 22:14
_itz Ulti: or use the perl6 module which does that
Ulti or that ^
Zoffix k :)
Ulti need to configure vim a bit though
Zoffix For the module?
BenGoldberg no, googling for them now... 22:15
_itz github.com/stmuk/p6-Text-VimColour/
Zoffix Ah, I've actually tried that last night: github.com/stmuk/p6-Text-VimColour/issues/6
Ulti for the up to date highlighting and for recognising Perl 6
also not really convinced that custom operators should be one of three head line features....
or infinite series 22:16
lizmat BenGoldberg: jnthn.net/papers/2015-spw-concurrency.pdf
_itz "didn't find a recent vim/ex"
Ulti these are cool but not really practical every day power tools that Perl 6 brings
_itz do you have vim/ex 7.4 in your path?
Zoffix shrugs
I don't use vim, so probably not
Ulti, custom operators is definitely something that excited me about Perl 6 :) 22:17
22:17 bjz left
Zoffix But I don't know Perl 6, so... yeah, my examples likely suck :) 22:17
Ulti its the kind of thing that scares anyone who works in languages where normal operator overloading is abused though 22:18
though I guess given slangs exist people better get used to it :) 22:19
22:19 bjz joined
masak 'night, #perl6 22:19
Zoffix _itz, I have vim version 7.3.429, so that's likely it. I'll close that Issue
Bye
_itz it probably should give a better error message
22:27 vvz left 22:29 bjz left 22:38 TEttinger joined
timotimo m: say try EVAL('my $a; $a ++'); 22:43
camelia rakudo-moar 2765b6: OUTPUT«Nil␤»
timotimo m: say try EVAL('my $a; $a ++') orelse $!; 22:44
camelia rakudo-moar 2765b6: OUTPUT«Nil␤»
timotimo i think the "do_safely" thing in the FAQ is quite verbose and weird
skids m: try EVAL("my \$a; \$a ++") orelse $!.say; 22:47
camelia ( no output )
skids m: try EVAL("my \$a; \$a ++"); $!.say;
camelia rakudo-moar 2765b6: OUTPUT«5===SORRY!5=== Error while compiling EVAL_0␤Prefix + requires an argument, but no valid term found␤at EVAL_0:1␤------> 3my $a; $a ++7⏏5<EOL>␤ expecting any of:␤ prefix␤»
skids m: (try EVAL("my \$a; \$a ++")) orelse $!.say; 22:48
camelia rakudo-moar 2765b6: OUTPUT«5===SORRY!5=== Error while compiling EVAL_0␤Prefix + requires an argument, but no valid term found␤at EVAL_0:1␤------> 3my $a; $a ++7⏏5<EOL>␤ expecting any of:␤ prefix␤»
dalek ast: 877e74c | lizmat++ | S02-types/ (2 files):
Add immutability tests for Bag/Mix
ast: 44feca7 | lizmat++ | S02-types/ (2 files):
Add typecheck tests for BagHash/MixHash
lizmat I will make the Mix/Bag immutability tests pass tomorrow, or mark them as todo 22:49
but first I'm going to get some sleep
so good night, #perl6!
skids o/ lizmat
22:51 laouji joined, Xor_ joined 22:55 laouji left
BenGoldberg m: gist.github.com/BenGoldberg1/47b3a...bdd630cbbc 22:58
camelia rakudo-moar 2765b6: OUTPUT«True␤0.213231999251397␤True␤0.353279550610509␤True␤0.604705458162459␤True␤0.627016712158831␤True␤0.809176592597244␤True␤0.856785885671298␤True␤0.984910784517846␤True␤»
BenGoldberg I'm getting it closer to working ;) 22:59
m: gist.github.com/BenGoldberg1/47b3a...bdd630cbbc
camelia rakudo-moar 2765b6: OUTPUT«0.0764646994813678␤0.0877851502498979␤0.205447866786057␤0.327032577604799␤0.418614171629946␤0.499468599774313␤0.572545911835536␤»
BenGoldberg Yay!
23:02 rurban left, leont joined 23:05 mattprelude left 23:06 laouji joined
BenGoldberg m: Supply.new.close; 23:07
camelia ( no output )
BenGoldberg STD: Supply.new.close;
j: Supply.new.close;
camelia ( no output ) 23:08
BenGoldberg According to doc.perl6.org/type/Supply, the only method of supply called .close must take a Tap object. Why doesn't rakudo complain when I do that? 23:09
skids rakudo has a 'multi method close(Supply:D:) { self.close($_) for self.tappers }' 23:14
BenGoldberg It ought to be documented, then :P Or perhaps, renamed to closeall, so someone rewriting code from using a Channel to using a Supply doesn't get bitten by accidentally doing .close instead of .done 23:18
Or renamed and documented ;)
23:20 RabidGravy left
skids I'm not sure what's on the way in and what's on the way out WRT concurrency so I don't know what to doc, personally. 23:22
23:26 mr_ron left 23:30 Xor_ left 23:34 yqt left 23:35 Xor_ joined 23:40 Peter_R left 23:52 tokuhirom joined 23:56 tokuhirom left

Powered by Cro and the Raku® Programming Language.

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

Thank you!