»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg camelia perl6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by sorear on 25 June 2013.
00:02 kurahaupo left, Guest33296 left 00:07 tmtwd joined 00:10 hoverboard left
tmtwd whoami 00:16
00:22 rurban joined 00:26 rurban left 00:31 takesako____ left 00:47 takesako____ joined 00:54 rurban joined, rurban left 01:05 Su-Shee_ joined 01:09 Su-Shee left 01:14 berekuk left 01:17 Sqirrel left 01:18 FROGGS left 01:20 klapperl_ joined 01:23 klapperl left 01:30 FROGGS joined 01:34 Sqirrel joined 01:43 raiph joined 01:51 kurahaupo joined
dalek p: 5f942fc | (Andrew Egeler)++ | src/vm/parrot/6model/reprs/CStruct.c:
Serialize CStruct on parrot
 01:56
p: 948999a | (Andrew Egeler)++ | src/vm/parrot/6model/reprs/CArray.c:
Fix memory corruption on CArray resize
01:57 dayangkun left 02:06 kurahaupo left 02:14 btyler joined 02:21 plobsing joined 02:34 berekuk joined
retupmoca r: say $_ for 1,2,3; say "{$_}" for 1,2,3 02:45
camelia rakudo-parrot 698d20: OUTPUT«1␤2␤3␤1␤2␤3␤»
..rakudo-{jvm,moar} 698d20: OUTPUT«1␤2␤3␤␤␤␤»
retupmoca I think parrot is correct on that one
p: say Date.new; say Date.new.succ; 02:51
camelia rakudo-parrot 698d20: OUTPUT«2014-12-24␤4015-24006-23␤»
retupmoca m: say Date.new; say Date.new.succ;
camelia rakudo-moar 698d20: OUTPUT«2014-12-24␤2014-12-25␤»
retupmoca o.O parrot obviously doesn't understand dates 02:53
03:07 takesako____ left 03:11 berekuk left 03:20 skids joined 03:23 takesako____ joined 03:24 rurban joined 03:25 Alula left 03:33 anaeem1_ joined, btyler left 03:36 BenGoldberg left, Alula joined 03:38 anaeem1_ left 03:54 rurban left 04:03 nick___ joined 04:08 btyler joined 04:13 nick___ left 04:16 btyler left 04:25 rurban joined 04:32 rurban left 04:51 raiph left 04:54 anaeem1 joined 04:58 anaeem1 left 05:00 stux left 05:01 bjz joined 05:04 stux joined 05:29 rurban joined 05:33 rurban left 05:54 anaeem1_ joined 06:03 tmtwd left 06:09 anaeem1_ left 06:18 telex left 06:20 telex joined 06:24 Sqirrel left, Sqirrel joined 06:32 rurban joined 06:36 rurban left 06:44 Sqirrel left
Woodi good morning everyone :) 06:44
r: my %a= <a 1 b 2 c 3 d 4>; my %b = <b 2 c 1>; %a += %b;
camelia rakudo-moar 698d20: OUTPUT«Odd number of elements found where hash expected␤ in method STORE at src/gen/m-CORE.setting:9340␤ in block at src/gen/m-CORE.setting:17119␤ in block at /tmp/tmpfile:1␤␤»
..rakudo-parrot 698d20: OUTPUT«Odd number of elements found where hash expected␤ in method STORE at gen/parrot/CORE.setting:9302␤ in block at gen/parrot/CORE.setting:17316␤ in block at /tmp/tmpfile:1␤␤»
..rakudo-jvm 698d20: OUTPUT«Odd number of elements found where hash expected␤ in method STORE at gen/jvm/CORE.setting:9281␤ in block at gen/jvm/CORE.setting:17059␤ in block at /tmp/tmpfile:1␤␤»
06:44 Sqirrel joined
Woodi is maybe some shortcut to sum hashes ? 06:45
06:52 virtualsue left 06:53 virtualsue joined, darutoko joined
Timbus m: my %a= <a 1 b 2 c 3 d 4>; my %b = <b 2 c 1>; say (%a (+) %b).hash 06:56
camelia rakudo-moar 698d20: OUTPUT«("a" => 1, "b" => 4, "c" => 4, "d" => 4).hash␤»
Woodi \o/ thanx Timbus++ :) 06:58
m: my %a= <a 1 b 2 c 3 d 4>; my %b = <b 2 c 1>; %a = (%a (+) %b).hash; say %a 07:01
camelia rakudo-moar 698d20: OUTPUT«("a" => 1, "b" => 4, "c" => 4, "d" => 4).hash␤»
Timbus m: my %a= <a 1 b 2 c 3 d 4>; my %b = <b 2 c 1>; for %a.kv -> $k, $v is rw { $v += %b{$k}//0 }; say %a 07:03
camelia rakudo-moar 698d20: OUTPUT«("a" => 1, "b" => 4, "c" => 4, "d" => 4).hash␤»
Woodi .kv is why I wanted "just" PerlScript :) but I needed to do such merge in loop and second loop looked ugly... 07:06
07:14 nnunley left 07:24 thou left 07:33 rurban joined 07:34 baest_ joined, bjz_ joined, takesako_____ joined 07:36 eternaleye_ joined 07:37 rurban left 07:41 takesako____ left, baest left 07:45 eternaleye left, eternaleye_ is now known as eternaleye 07:47 bjz left
FROGGS I dunno if + should ever do more that numeric addition.... if you make it sum hashes, than somebody also wants it to concat strings 07:50
07:50 berekuk joined
FROGGS so, hash + hash should only return the sum of the number of keys 07:50
Woodi I didn't thinked about it :) but short syntax for merging/sum hashes would be convenient 07:53
FROGGS m: my %a= <a 1 b 2 c 3 d 4>; my %b = <b 2 c 1>; say %(%a (+) %b) # that is short enough for me 07:55
camelia rakudo-moar 698d20: OUTPUT«("a" => 1, "b" => 4, "c" => 4, "d" => 4).hash␤»
FROGGS std: my (%a, %b); %a (+=) %b # I wonder if this should be allowed... TimToady? 07:57
camelia std ec3d853: OUTPUT«===SORRY!===␤Confused at /tmp/TXVgCyiubl line 1:␤------> my (%a, %b); %a ⏏(+=) %b # I wonder if this should be all␤ expecting any of:␤ feed_separator␤ infix or meta-infix␤ infixed function␤ statement modifie…»
Woodi it's works for me. note: it's Bag and that new types are very new features
btw PATRICIA trees are sometimes better then hashes, maybe we can have them too ? 07:59
FROGGS Woodi: that sounds like an awesome thing for modules.perl6.org :o) 08:03
even if that means that it will went into core, it is a good place to prototype it 08:04
masak good mornin', #perl6
lizmat mood gornin', #perl6!
FROGGS o/ :o) 08:05
lizmat FROGGS masak Woodi o/
Woodi: if you're using hashes like that, you probably should use Bag's :-)
Bags have the added feature that you can haz objects as keys without them being stringified 08:06
r: class A {}; (A.new => 1).Bag.say 08:07
camelia rakudo-{parrot,jvm,moar} 698d20: OUTPUT«bag(A.new())␤»
lizmat r: class A {}; (A.new => 1).hash.say # alas, A.new is gone
camelia rakudo-parrot 698d20: OUTPUT«("A<7818003069261027125>" => 1).hash␤»
..rakudo-jvm 698d20: OUTPUT«("A<496755591>" => 1).hash␤» 08:08
..rakudo-moar 698d20: OUTPUT«("A<139845920269608>" => 1).hash␤»
08:09 nnunley joined
Woodi lizmat: as I was trying to say: I have no idea what to expects from that new types :) except Sets of course :) but S32 already opened ! 08:10
I thinked hashes could have object as keys too...
ah, but with ~...
lizmat only if you specify them like : my %h{Any}
with .hash you only get stringified keys 08:11
08:13 nnunley left 08:18 Su-Shee_ is now known as Su-Shee, virtualsue left, clkaoud left, pnu left, flussence left, _sri left
Woodi is still thinking %a +=h %b is common case... 08:22
moritz m: my %a = a => 1; my %b = b => 2; %a ,= %b; say %a.perl 08:25
camelia rakudo-moar 698d20: OUTPUT«("a" => 1, "b" => 2).hash␤»
FROGGS ohh!
moritz++
Woodi r: my Sethash $a = <a 1 b 2 c 3 d 4>; my SetHash $b = <b 2 c 1>; $a += $b; 08:26
camelia rakudo-{parrot,jvm,moar} 698d20: OUTPUT«===SORRY!===␤Type 'Sethash' is not declared. Did you mean any of these?␤ SetHash␤ Stash␤␤at /tmp/tmpfile:1␤------> my Sethash⏏ $a = <a 1 b 2 c 3 d 4>; my SetHash $b =␤Malformed my␤at…»
Woodi r: my %a= <a 1 b 2 c 3 d 4>; my %b = <b 2 c 1>; %a ,= %b;
camelia ( no output )
Woodi r: my %a= <a 1 b 2 c 3 d 4>; my %b = <b 2 c 1>; %a ,= %b; say %a
camelia rakudo-{parrot,jvm,moar} 698d20: OUTPUT«("a" => "1", "b" => "2", "c" => "1", "d" => "4").hash␤»
FROGGS m: my SetHash $a = <a 1 b 2 c 3 d 4>; my SetHash $b = <b 2 c 1>; $a += $b;
camelia rakudo-moar 698d20: OUTPUT«Type check failed in assignment to '$a'; expected 'SetHash' but got 'Parcel'␤ in block at /tmp/I7KIPB8S8A:1␤␤»
lizmat FROGGS, Woodi : you probably want BagHash instead of SetHash 08:27
Woodi a bit a mess we have with that types :) 08:28
lizmat m: my $a = <a a b b c>.Bag; $a (+)= <b b c c>.Bag; say $a # would have expected this to work
camelia rakudo-moar 698d20: OUTPUT«===SORRY!=== Error while compiling /tmp/E2ZKwnJFmx␤Two terms in a row␤at /tmp/E2ZKwnJFmx:1␤------> my $a = <a a b b c>.Bag; $a ⏏(+)= <b b c c>.Bag; say $a # would have␤ expecting any of:␤ postfix…»
08:29 virtualsue joined, clkaoud joined, pnu joined, flussence joined, _sri joined
lizmat m: my $a = <a a b b c>.Bag; $a = $a (+) <b b c c>.Bag; say $a 08:29
camelia rakudo-moar 698d20: OUTPUT«bag(a(2), b(4), c(3))␤»
lizmat guess we need to add set operators to some list that allow = postfix? 08:30
Woodi r: my BagHash $a = <a 1 b 2 c 3 d 4>; my BagHash $b = <b 2 c 1>; BagHash $c = $a + $b; say $c
camelia rakudo-{parrot,jvm,moar} 698d20: OUTPUT«===SORRY!=== Error while compiling /tmp/tmpfile␤Two terms in a row␤at /tmp/tmpfile:1␤------> 4>; my BagHash $b = <b 2 c 1>; BagHash ⏏$c = $a + $b; say $c␤ expecting any of:␤ postfix…»
Woodi r: my BagHash $a = <a 1 b 2 c 3 d 4>; my BagHash $b = <b 2 c 1>; BagHash $c = $a (+) $b; say $c 08:31
camelia rakudo-{parrot,jvm,moar} 698d20: OUTPUT«===SORRY!=== Error while compiling /tmp/tmpfile␤Two terms in a row␤at /tmp/tmpfile:1␤------> 4>; my BagHash $b = <b 2 c 1>; BagHash ⏏$c = $a (+) $b; say $c␤ expecting any of:␤ postfix…»
lizmat Woodi: my missing with $c
Woodi r: my BagHash $a = <a 1 b 2 c 3 d 4>; my BagHash $b = <b 2 c 1>; my BagHash $c = $a (+) $b; say $c 08:32
camelia rakudo-{parrot,jvm,moar} 698d20: OUTPUT«Type check failed in assignment to '$a'; expected 'BagHash' but got 'Parcel'␤ in block at /tmp/tmpfile:1␤␤»
Woodi r: my BagHash $a = <a 1 b 2 c 3 d 4>; my BagHash $b = <b 2 c 1>; my BagHash $c = $a + $b; say $c
camelia rakudo-{parrot,jvm,moar} 698d20: OUTPUT«Type check failed in assignment to '$a'; expected 'BagHash' but got 'Parcel'␤ in block at /tmp/tmpfile:1␤␤»
Woodi anyone blogged about XxYy types maybe ? :) 08:33
lizmat I did, for the advent, if I remember correcly 08:35
Woodi ook
lizmat hmmm... one could argue that the result of (+) of two BagHashes should also be a BagHash, and not a Bag 08:36
08:38 skids left 08:44 skids joined 08:57 bjz_ left, bjz joined 08:58 plobsing left
moritz then you depart from p6's principle that operators should coerce to the type they operate on 09:01
though there's precedence
+ on two Ints vs on two Rats
09:09 kivutar joined 09:25 virtualsue left, clkaoud left, pnu left, flussence left, _sri left 09:31 daniel-s left, rindolf joined 09:32 daniel-s joined 09:33 molaf joined 09:35 rurban joined 09:36 virtualsue joined, clkaoud joined, pnu joined, flussence joined, _sri joined
dalek kudo/nom: 71411c9 | (Elizabeth Mattijsen)++ | src/core/signals.pm:
Make sure we send Signal to executed code

Rather than just the int that we got from the OS		 09:38
09:39 rurban left
FROGGS jnthn: I need your support :/ gist.github.com/FROGGS/a80aabefe5e39fb553b4 09:46
jnthn: I am rethrowing wrongish it seems, but I don't know why because I rethrow like I create the ex in the first place in "last FOO"
09:53 dmol joined
FROGGS hmmm, I think I should have a close look ad handlerDynamic() 09:55
09:58 nnunley joined
jnthn FROGGS: Just underneath where you added the label checkis: 10:00
# Wraps a handler with code to set/clear the current handler.
I wonder if the clear needs to happen first
nwc10 good almost-UGT-violation, jnthn
jnthn But not convinced it's that...
o/ nwc10 10:01
dalek ecs: f47db42 | (Elizabeth Mattijsen)++ | S17-concurrency.pod:
Spec signal() and some POSIX signals cleanup
10:02 nnunley left
lizmat jnthn: is there an equivalent of "kill -l" under Win ? 10:03
10:04 baest joined
FROGGS hmmmm 10:05
jnthn lizmat: Not afaik
I mean, there's a taskkill program but it's not really about signals in any way...Windows doesn't do signals really. libuv emulates a few.
10:06 baest_ left
jnthn (As in, turns other Windows things into signal-y things) 10:06
lizmat ah, ok... hmmm... so qx/kill -1 $*PID/ to test signalI() functionality, is not going to work 10:07
reliably anyway
would libuv support kill() ?
FROGGS jnthn: the unwind_check happens within the catch of the delemit_handler 10:08
jnthn Not sure it does
Woodi lizmat: msdn.microsoft.com/en-us/library/xdkz3x12.aspx
lizmat can we fudge tests depending on OS ? 10:09
jnthn Not sure about fudge, but you check OS in a test. 10:10
And you've got a plan *; then you don't even need to emit skips I guess.
lizmat I assume you mean $*OS right>
jnthn Right
lizmat hmmm... that appears to be a Str atm 10:11
jnthn FROGGS: if (vmex.payload.hashCode() == where) { 10:12
FROGGS yes?
jnthn I'm not sure if hashCode is stable over an object's lifetime... 10:13
vendethiel lizmat: taskkill /F /PID N ?
FROGGS jnthn: but that is what nqp::where does... and I was thinking it will never move... :/ 10:14
okay, I'll keep in mind that I have to keep the label object, and not its address
lizmat vendethiel++ will put that in the MSWin32 case
$*OS doesn't seem specced 10:15
jnthn FROGGS: Was gonna say..
$il.append(JAST::PushIVal.new( :value($label ?? nqp::where($label.value) !! 0) ));
That could potentially we doing a wval thingy instead
FROGGS: What is making me curious, though, is we never seem to event hit the outer exception ahdnler at all 10:16
That is, I'd expect we'd see another call to _is_same_label 10:17
FROGGS jnthn: me too, that is why I look closer at handlerDynamic
jnthn yeah. Get it to dump out what it actually puts into the unwindexception, maybe 10:18
lizmat perl6: say $*OS # not a lot of consistency :-( 10:19
jnthn FROGGS: oooh
camelia rakudo-{parrot,moar} 698d20: OUTPUT«linux␤»
..niecza v24-109-g48a8de3: OUTPUT«Unix␤»
..rakudo-jvm 698d20: OUTPUT«Linux␤»
jnthn FROGGS: One important thing you might want to know...
FROGGS: There's a single unwindException per thread
public UnwindException unwinder;
in ThreadContext
Which is re-used
FROGGS ohhh, I am setting the result in one case.... so I need to clear it in other cases I guess 10:20
jnthn Well, also that clearup code I mentioned earlier
oh, maybe not... 10:21
But anyway, that could be your problem
FROGGS but as I said, the unwind_check is being put into the catch of the delimit handler, so this happens first, no? 10:26
10:31 denis_boyun_ joined 10:35 itz2 joined
itz2 twitter.com/bin3ry/status/46507198...04/photo/1 10:35
FROGGS nice :o) 10:37
dalek kudo/nom: c5ac807 | (Elizabeth Mattijsen)++ | src/core/signals.pm:
Make sure we do each given signal once
 10:40
10:47 Rotwang joined
FROGGS jnthn: in invokes the same handler again 10:49
jnthn FROGGS: So, that bit of logic that places us outside of the handled region I mentioend might fix that... 10:50
10:52 itz2 left
jnthn Wow, Perl ukulele :) 10:52
10:54 virtualsue left, clkaoud left, pnu left, flussence left, _sri left 10:59 nnunley joined 11:05 virtualsue joined, clkaoud joined, pnu joined, flussence joined, _sri joined
lizmat I'm seeing all sorts of OS related information spread out in several IO modules, and test-files 11:22
and $*OS just being s atring 11:23
would it be a plan to make $*OS a bit smarter and e.g. add a method "is-win" , "FSTYPE" etc 11:24
.name -> .gist / .Str so the current behaviour would still work
does that make sense? Or am I too soon? or do I need to spec first?
11:25 kivutar left
jnthn lizmat: I think S02 does specify various other special vars like $*DISTRO and $*KERNEL that might be worth considering into this 11:27
colomon wow, Perl 6 stickers?!!
lizmat colomon: we'll be taking them to YAPC::NA 11:28
otherwise, if you want any, please contact woolfy
colomon will do. ;) 11:29
lizmat perl6: say $VM<name> 11:30
camelia rakudo-jvm 71411c: OUTPUT«(timeout)»
..niecza v24-109-g48a8de3: OUTPUT«===SORRY!===␤␤Variable $VM is not predeclared at /tmp/tmpfile line 1:␤------> say ⏏$VM<name>␤␤Unhandled exception: Check failed␤␤ at /home/p6eval/niecza/boot/lib/CORE.setting line 1502 (die @ 5) …»
..rakudo-{parrot,moar} 71411c: OUTPUT«===SORRY!=== Error while compiling /tmp/tmpfile␤Variable '$VM' is not declared␤at /tmp/tmpfile:1␤------> say $VM<name>⏏<EOL>␤ expecting any of:␤ postfix␤»
lizmat perl6: say $*VM<name> # sigh 11:31
camelia rakudo-moar 71411c: OUTPUT«moar␤»
..rakudo-jvm 71411c: OUTPUT«jvm␤»
..rakudo-parrot 71411c: OUTPUT«parrot␤»
..niecza v24-109-g48a8de3: OUTPUT«(Any)␤»
dalek ecs: 1cba1e7 | (Elizabeth Mattijsen)++ | S02-bits.pod:
Fix straggling META.info reference
 11:38
FROGGS jnthn: I think I understand now 11:39
lizmat afk for a few hours& 11:42
11:44 kaare_ joined 11:51 kaare_ left 12:04 virtualsue left
colomon m: my @a = 1..10; say @a.pairs 12:05
camelia rakudo-moar c5ac80: OUTPUT«0 => 1 1 => 2 2 => 3 3 => 4 4 => 5 5 => 6 6 => 7 7 => 8 8 => 9 9 => 10␤»
12:10 virtualsue joined
FROGGS nice... all nqp's pass the labels test... 12:20
jnthn \o/
FROGGS++
FROGGS in theory perl6-j should already be fine, and perl6-p needs the nqp::handle adjustments in the pir block in MapIter.reify 12:22
but spectests first :o)
ewww 12:23
12:23 rurban joined
FROGGS its lets an exception slip when building Perl6::Pod 12:23
it*
12:24 new6er joined
dalek p/loop_labels: f44aaab | (Tobias Leich)++ | src/vm/jvm/ (3 files):
add loop labels support for jvm backend
 12:34
p/loop_labels: 18d11f7 | (Tobias Leich)++ | t/nqp/84-loop-labels.t:
add test for labeled for loop
p/loop_labels: 9dd4604 | (Andrew Egeler)++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/Ops.java:
Add write barrier to JVM push ops

This was done on moarvm a while back, and fixes a precompilation issue.
p/loop_labels: 5f942fc | (Andrew Egeler)++ | src/vm/parrot/6model/reprs/CStruct.c:
Serialize CStruct on parrot
p/loop_labels: 948999a | (Andrew Egeler)++ | src/vm/parrot/6model/reprs/CArray.c:
Fix memory corruption on CArray resize
p/loop_labels: 63910f6 | (Tobias Leich)++ | src/vm/ (3 files):
Merge branch 'master' of github.com:perl6/nqp into loop_labels
Heuristic branch merge: pushed 18 commits to rakudo/loop_labels_test by FROGGS
masak FROGGS++ 12:38
FROGGS we're not quite there yet 12:41
12:45 new6er left, anaeem1_ joined
masak m: class B { ... }; class A { method foo(B) { "A.foo(B)" } }; class B is A { method foo(A) { "B.foo(A)" } }; my $b = B.new; say $b.foo(A) 13:16
camelia rakudo-moar c5ac80: OUTPUT«B.foo(A)␤»
masak why is it again that this favors B.foo(A), and doesn't result in a dispatch ambiguity? 13:17
oh, I didn't declare them proto, for one thing. 13:18
jnthn masak: Did you mean to write "multi"?
masak um, multi.
masak tries again
m: class B { ... }; class A { multi method foo(B) { "A.foo(B)" } }; class B is A { multi method foo(A) { "B.foo(A)" } }; my $b = B.new; say $b.foo(A)
jnthn You're just getting normal method dispatch behavior there :)
camelia rakudo-moar c5ac80: OUTPUT«B.foo(A)␤»
masak ok, question repeated. :)
FROGGS you are calling foo(A), so the foo(A) is narrower than foo(B) 13:19
m: class B { ... }; class A { multi method foo(B) { "A.foo(B)" } }; class B is A { multi method foo(A) { "B.foo(A)" } }; my $b = B.new; say $b.foo(B)
camelia rakudo-moar c5ac80: OUTPUT«Ambiguous call to 'foo'; these signatures all match:␤:(A: B, *%_)␤:(B: A, *%_)␤ in block at /tmp/pkM9YolUOD:1␤␤»
jnthn Yeah, but the invocant type should play into it also... 13:20
masak FROGGS++
jnthn ah, right
masak right.
jnthn Yeah, the other candidate doesn't even count. :)
masak ok, everything's fine.
I think I'm in love with Perl 6's multi dispatch. 13:21
FROGGS ohh yes, I love it too 13:23
it is totally DWIM and very flexible
even moar with the power of the signatures 13:24
13:25 anaeem1_ left
jnthn m: say 5.20 / 5.84 13:25
camelia rakudo-moar c5ac80: OUTPUT«0.890411␤»
dalek p/loop_labels: 5e3e216 | (Tobias Leich)++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/ExceptionHandling.java:
add accidently removed comments
 13:26
masak I do wish I could use signatures for destructuring, though.
hoping to write a language extension for that sometime.
jnthn You...can't?
r: class A { has $.b; has $.c }; my (:$b, :$c) := A.new(b => 1, c => 2); say $b; say $c; 13:27
camelia rakudo-jvm c5ac80: OUTPUT«(timeout)»
..rakudo-{parrot,moar} c5ac80: OUTPUT«1␤2␤»
jnthn bah, shoulda m: :)
Granted, you should not do that on any hot path as it's heavily unoptimized. 13:28
masak m: my %struct = name => "Gabriel", addr => ["Pearly Gates", "1", "Heaven"]; my (:$name, :addr([$l1, $l2, $l3])) = %struct; say $name; .say for $l1, $l2, $l3 13:29
camelia rakudo-moar c5ac80: OUTPUT«"name" => "Gabriel"␤(Mu)␤(Mu)␤(Mu)␤»
masak :/
jnthn: like that.
jnthn: but (more crucially), if done in a 'when' expression, the variables should be available inside the block. 13:30
jnthn m: my %struct = name => "Gabriel", addr => ["Pearly Gates", "1", "Heaven"]; my (:$name, :addr($) [$l1, $l2, $l3]) := %struct; say $name; .say for $l1, $l2, $l3 # maybe 13:31
camelia rakudo-moar c5ac80: OUTPUT«Gabriel␤Pearly Gates␤1␤Heaven␤»
jnthn Just gotta use sub-sig syntax correct, and remember to bind :P 13:32
masak ooh
\o/
well, that takes care of use case 1.
the syntax is kinda wonky, but fine.
jnthn About when, I think something is spec'd there wrt sig matching, but not sure I remember implementing it.
masak oki 13:33
anyway, jnthn++
13:39 araujo left 14:09 benabik joined 14:19 skids left, skids joined 14:20 virtualsue left 14:22 virtualsue joined 14:23 daniel-s left, daniel-s_ joined
dalek p/loop_labels: 50cae66 | (Tobias Leich)++ | src/vm/jvm/ (3 files):
fix regressions by adding exception category for labels
 14:31
14:35 skids left, skids joined
masak r: package Foo; say "alive" 14:42
camelia rakudo-{parrot,jvm,moar} c5ac80: OUTPUT«===SORRY!=== Error while compiling /tmp/tmpfile␤This appears to be Perl 5 code. If you intended it to be Perl 6 code, please use a Perl 6 style package block like "package Foo { ... }", or "module Foo; ...".␤at /tmp/tmpfi…»
skids Darn. Actions.nqp line 5543 looked like the perfect explanation for the 0x0123456789abcdef bug, but I changed that to ".0" and no dice.
Still looks fishy, maybe there is another copy of that lying around. 14:43
r: 0x123456789abcdef.base(16).say
camelia rakudo-moar c5ac80: OUTPUT«123456789ABCDF0␤» 14:44
..rakudo-{parrot,jvm} c5ac80: OUTPUT«123456789ABCDEF␤»
skids see it is adding a zero.
r: Int("0x0123456789abcdef").base(16).say # but not here.
camelia rakudo-{parrot,jvm,moar} c5ac80: OUTPUT«123456789ABCDEF␤»
skids r: 0x00020000000000001.base(16).say; 0x00020000000000000.base(16).say; # and only above this boundary. 14:46
camelia rakudo-{parrot,jvm} c5ac80: OUTPUT«20000000000001␤20000000000000␤»
..rakudo-moar c5ac80: OUTPUT«20000000000000␤20000000000000␤»
skids oh.
hmm.
and under this boundary: r: 0x10020000000000001.perl.say; 14:48
well, anyway there sure are a lot of different codepaths for parsing code/text into Numeric. 14:49
14:49 logie_ joined 14:55 logie_ left 15:01 SamuraiJack_ joined
dalek kudo/loop_labels_test: 96e3bdc | (Tobias Leich)++ | src/core/Any.pm:
unbreak build caused by a partial fine branch merge
 15:06
p/loop_labels: 748ef15 | (Tobias Leich)++ | src/vm/jvm/QAST/Compiler.nqp:
add nqp constant for CONTROL_LABELED
 15:15
timotimo could more people try building their moarvm with --optimize=3? 15:16
retupmoca timotimo: it worked fine for me last week or so 15:17
timotimo i wonder if it's the gcc version? 15:18
flussence r: printf('%x', 0x123456789abcdef +& 0xFFFF_FFFF) 15:30
camelia rakudo-moar c5ac80: OUTPUT«89abcdf0»
..rakudo-{parrot,jvm} c5ac80: OUTPUT«89abcdef»
flussence r: printf('%x', 0x123456789abcdef +& 0xFFFF_FFFF_0000_0000) 15:31
camelia rakudo-{parrot,jvm,moar} c5ac80: OUTPUT«123456700000000»
skids r: say nqp::atpos(nqp::radix_I(16, "0123456789abcdef", 0, 2, Int),0).base(16); # this codepath seems to be clean 15:41
camelia rakudo-{parrot,jvm,moar} c5ac80: OUTPUT«123456789ABCDEF␤»
skids Oh, it isn't adding a zero. It is clobbering a nybble. 15:43
flussence++
flussence seems like a weird range to have an off-by-one in... 15:45
r-m: say 0x00020000000000000..0x00020000000000005 X+& 0xFFFF_FFFF 15:46
r: say 0x00020000000000000..0x00020000000000005 X+& 0xFFFF_FFFF
camelia rakudo-moar c5ac80: OUTPUT«(timeout)» 15:47
rakudo-{parrot,moar} c5ac80: OUTPUT«(timeout)»
..rakudo-jvm c5ac80: OUTPUT«0 1 2 3 4 5␤»
flussence r-m: say 0x00020000000000000..0x00020000000000005
camelia rakudo-moar c5ac80: OUTPUT«9007199254740992..9007199254740996␤»
flussence r-m: say 0x00020000000000000...0x00020000000000005
camelia rakudo-moar c5ac80: OUTPUT«9007199254740992 9007199254740993 9007199254740994 9007199254740995 9007199254740996␤» 15:48
flussence r-m: say 0x00020000000000000...0x00020000000000005 X+& 0xFFFF_FFFF
camelia rakudo-moar c5ac80: OUTPUT«(timeout)»
flussence r-m: say $_ +& 0xFFFF_FFFF for 0x00020000000000000...0x00020000000000005
camelia rakudo-moar c5ac80: OUTPUT«0␤1␤2␤3␤4␤»
15:49 guru joined, guru is now known as ajr_
flussence r-m: for 0x00020000000000000..0x00020000000000005 { say $_; say $_ +& 0xFFFF_FFFF } 15:49
camelia rakudo-moar c5ac80: OUTPUT«(timeout)9007199254740992␤0␤9007199254740992␤0␤9007199254740992␤0␤9007199254740992␤0␤9007199254740992␤0␤9007199254740992␤0␤9007199254740992␤0␤9007199254740992␤0␤9007199254740992␤0␤9007199254740992␤0␤9007199254740992…»
flussence oh I get it... that number problem is making the range iterator horribly confused... 15:50
r-m: for 0x00020000000000000...0x00020000000000005 { say $_; say $_ +& 0xFFFF_FFFF }
camelia rakudo-moar c5ac80: OUTPUT«9007199254740992␤0␤9007199254740993␤1␤9007199254740994␤2␤9007199254740995␤3␤9007199254740996␤4␤»
flussence r-m: for 0x00020000000000000...0x00020000000000005 { say ($_, $_ +& 0xFFFF_FFFF)».fmt('%x') } 15:51
camelia rakudo-moar c5ac80: OUTPUT«20000000000000 0␤20000000000001 1␤20000000000002 2␤20000000000003 3␤20000000000004 4␤»
flussence ...but not the sequence op, for some reason. 15:52
15:52 spider-mario joined
flussence r-m: for 0x123456789abcdef...0x123456789abcdff { say ($_, $_ +& 0xFFFF_FFFF)».fmt('%x') } 15:53
camelia rakudo-moar c5ac80: OUTPUT«123456789abcdf0 89abcdf0␤123456789abcdf1 89abcdf1␤123456789abcdf2 89abcdf2␤123456789abcdf3 89abcdf3␤123456789abcdf4 89abcdf4␤123456789abcdf5 89abcdf5␤123456789abcdf6 89abcdf6␤123456789abcdf7 89abcdf7␤123456789abcdf8 89abcdf8␤123456789abcd…»
flussence r-m: for 0x123456789abcdef..0x123456789abcdff { say ($_, $_ +& 0xFFFF_FFFF)».fmt('%x') }
r-p: for 0x123456789abcdef..0x123456789abcdff { say ($_, $_ +& 0xFFFF_FFFF)».fmt('%x') }
camelia rakudo-moar c5ac80: OUTPUT«(timeout)123456789abcdf0 89abcdf0␤123456789abcdf0 89abcdf0␤123456789abcdf0 89abcdf0␤123456789abcdf0 89abcdf0␤123456789abcdf0 89abcdf0␤123456789abcdf0 89abcdf0␤123456789abcdf0 89abcdf0␤123456789abcdf0 89abcdf0␤123456789abcdf0 89abcdf0␤1234…» 15:54
rakudo-parrot c5ac80: OUTPUT«(timeout)123456789abcdf0 89abcdf0␤123456789abcdf0 89abcdf0␤123456789abcdf0 89abcdf0␤123456789abcdf0 89abcdf0␤123456789abcdf0 89abcdf0␤123456789abcdf0 89abcdf0␤123456789abcdf0 89abcdf0␤123456789abcdf0 89abcdf0␤123456789abcdf0 89abcdf0␤12…»
flussence r-j: for 0x123456789abcdef..0x123456789abcdff { say ($_, $_ +& 0xFFFF_FFFF)».fmt('%x') }
camelia rakudo-jvm c5ac80: OUTPUT«123456789abcdef 89abcdef␤123456789abcdf0 89abcdf0␤123456789abcdf1 89abcdf1␤123456789abcdf2 89abcdf2␤123456789abcdf3 89abcdf3␤123456789abcdf4 89abcdf4␤123456789abcdf5 89abcdf5␤123456789abcdf6 89abcdf6␤123456789abcdf7 89abcdf7␤123456789abcdf…»
flussence r: say 0x123456789abcdef.succ 15:55
camelia rakudo-{parrot,jvm} c5ac80: OUTPUT«81985529216486896␤»
..rakudo-moar c5ac80: OUTPUT«81985529216486897␤»
flussence r: say 0x123456789abcdef.succ.succ
camelia rakudo-moar c5ac80: OUTPUT«81985529216486898␤»
..rakudo-{parrot,jvm} c5ac80: OUTPUT«81985529216486897␤»
flussence r: say 0x123456789abcdf0.succ 15:56
camelia rakudo-{parrot,jvm,moar} c5ac80: OUTPUT«81985529216486897␤»
flussence bignum scary.
flussence goes back to his nice sane html regexing... 15:57
skids Yeah who needs numbers anymore :-)
16:16 araujo joined, araujo left, araujo joined 16:26 clkaoud left, pnu left, flussence left, _sri left
TimToady m: my %a = <a 1 b 2 c 3 d 4>; my %b = <b 2 c 1>; say %a »+« %b 16:32
camelia rakudo-moar c5ac80: OUTPUT«use of uninitialized value of type Any in numeric context␤use of uninitialized value of type Any in numeric context␤("a" => 1, "b" => 4, "c" => 4, "d" => 4).hash␤»
TimToady that should likely suppress those warnings
it should probably just copy over values that are in only one hash or the other 16:35
16:38 clkaoud joined, pnu joined, flussence joined, _sri joined
colomon +1 16:41
TimToady m: my %a = <a 1 b 2 c 3 d 4>; my %b = <b 2 c 1>; %a »+=« %b; say %a 16:42
camelia rakudo-moar c5ac80: OUTPUT«use of uninitialized value of type Any in numeric context␤use of uninitialized value of type Any in numeric context␤("a" => 1, "b" => 4, "c" => 4, "d" => 4).hash␤»
16:42 thou joined
TimToady m: my %a = <a 1 b 2 c 3 d 4>; my %b = <b 2 c 1>; %a »+«= %b; say %a 16:42
camelia rakudo-moar c5ac80: OUTPUT«===SORRY!=== Error while compiling /tmp/OvaiRAfr97␤Preceding context expects a term, but found infix = instead␤at /tmp/OvaiRAfr97:1␤------> b 2 c 3 d 4>; my %b = <b 2 c 1>; %a »+«=⏏ %b; say %a␤»
TimToady ah, yes, rakudo don't parse assignops correctly yet 16:43
n: my %a = <a 1 b 2 c 3 d 4>; my %b = <b 2 c 1>; %a »+«= %b; say %a
camelia niecza v24-109-g48a8de3: OUTPUT«Use of uninitialized value in numeric context␤ at /home/p6eval/niecza/lib/CORE.setting line 1389 (warn @ 5) ␤ at /home/p6eval/niecza/lib/CORE.setting line 301 (Any.Numeric @ 8) ␤ at <unknown> line 0 (ExitRunloop @ 0) ␤ at /home/p6eval/niec…»
colomon which of the two forms is correct? »+«= vs »+=« 16:44
TimToady niecza++ does but has even more verbose warnings
they are both correct :)
16:49 daniel-s_ left 16:50 daniel-s_ joined 16:51 zakharyas joined
colomon There's more than one way, eh? 16:57
16:59 xinming joined
TimToady they mean different things, but achieve the same result, assuming standard hash semantics 17:02
well, maybe not, now that I think about it more 17:05
17:05 anaeem1 joined
TimToady there's nothing in »+=« that will make sure the keys of the right side will end up in the left side's hash 17:06
unless the hyper somehow notices it's a mutating operator
I guess if it's careful to iterate the right hash'es keys, autoviv will work on the left side appropriately 17:10
m: my %a = <a 1 b 2 c 3 d 4>; my %b = <b 2 c 1 e 5>; %a »+=« %b; say %a 17:12
camelia rakudo-moar c5ac80: OUTPUT«use of uninitialized value of type Any in numeric context␤use of uninitialized value of type Any in numeric context␤("a" => 1, "b" => 4, "c" => 4, "d" => 4, "e" => 5).hash␤»
TimToady ha, works already, somehow 17:13
m: my %a = <a 1 b 2 c 3 d 4>; my %b = <b 2 c 1 e 5>; %a »R[+=]« %b; say %b 17:14
camelia rakudo-moar c5ac80: OUTPUT«use of uninitialized value of type Any in numeric context␤("a" => 1, "b" => 4, "c" => 4, "d" => 4, "e" => 5).hash␤»
TimToady and works with a reversed op even, but that's why we get the warning, because it's always calling the 2-arg form (which is what allows autoviv, I guess) 17:15
so the correct thing is to suppress the warning, not introspect the args to try to figure out which ones are there vs never there vs maybe there through autoviv 17:16
17:16 denis_boyun_ left, denis_boyun_ joined
TimToady introspecting is once again a code smell 17:16
17:18 firnsy left
lizmat wishes warnings worked more like "is DEPRECATED" 17:18
instead of just mumbling, keeping the into until the end and then show where they happened (and how many times) 17:19
*info
17:27 btyler joined
dalek p/lexopts: c9e9e90 | jnthn++ | src/vm/moar/stage0/ (11 files):
Update stage0 for MoarVM.

The previous bootstrap contained mis-generated code that upset a more aggressive spesh. Furthermore, the generated code is far better these days. Between spesh and the better stage0 (meaning we build stage 1 faster), the NQP build is now down to 41s (from around 50s) for me.		 17:28
lizmat jnthn++ 17:29
jnthn Faster NQP builds for all. Once that merges... :) 17:31
masak the above discussion about »+=« reminds me of rt.perl.org/Ticket/Display.html?id=118223
r: my %l = foo => 1, bar => 2; my %r = bar => 3, baz => 4; say %l >>+<< %r
camelia rakudo-{jvm,moar} c5ac80: OUTPUT«use of uninitialized value of type Any in numeric context␤use of uninitialized value of type Any in numeric context␤("bar" => 5, "baz" => 4, "foo" => 1).hash␤»
..rakudo-parrot c5ac80: OUTPUT«use of uninitialized value of type Any in numeric context in block at /tmp/tmpfile:1␤␤use of uninitialized value of type Any in numeric context in block at /tmp/tmpfile:1␤␤("bar" => 5, "baz" => 4, "foo" => 1).hash␤»
masak shouldn't warn. 17:32
timotimo jnthn: ŷay! \o/ 17:33
lizmat masak jnthn : looking into it now
but short of implementing "no warnings" I don't really see an easy fix yet 17:34
I see some optimizations
17:35 daniel-s_ left 17:36 daniel-s_ joined 17:38 hoverboard joined 17:41 firnsy joined, firnsy left, firnsy joined, thou left 17:42 btyler_ joined
tadzik aww yiss 17:43
17:45 btyler left
TimToady lizmat: well, you need something like a CONTROL that traps warnings dynamically, since the code in question is handed to you by reference 17:46
alternately, we need some way for warn to know its context and refrain from warning in the first place 17:48
17:49 anaeem1 left
lizmat tries to grok CONTROL better 17:50
TimToady but HOP basically requires dynamic scoping for indirect function control, since you don't have textual control of the, er, text 17:51
the only other alternative is to parameterize everything
and when you need to parameterize everything, that usually is a design smell that you need dynamic scoping instead 17:52
17:53 Zaegnair left
FROGGS ohh 17:56
jnthn++
masak m: say lines>>.comb.grep("a".."z").Bag.format("%s: %s") 17:57
17:57 MikeFair_ joined
camelia rakudo-moar c5ac80: OUTPUT«No such method 'format' for invocant of type 'Bag'␤ in block at /tmp/1CVFQCnVST:1␤␤» 17:57
masak requested: .format on Bag objects.
m: say lines>>.comb.grep("a".."z").Bag.Hash.format("%s: %s") 17:58
camelia rakudo-moar c5ac80: OUTPUT«No such method 'Hash' for invocant of type 'Bag'␤ in block at /tmp/4S2S9dbKOr:1␤␤»
masak hm.
TimToady lizmat: otoh, I can argue myself out of doing by warning supression, since that doesn't fix the case of »*=« wanting to default to multiplying by 1 rather than 0
so some degree of introspection is probably required in hypers 17:59
we really need to be able to ask an operator which argument or arguments it expects to be rw 18:00
colomon wonders if there is overlap with the Bag operators here...
TimToady yes, but a Bag only does +=, not *= 18:01
masak m: .say for lines>>.comb.grep("a".."z").Bag.pairs
camelia rakudo-moar c5ac80: OUTPUT«"h" => 103␤"e" => 150␤"r" => 71␤"w" => 23␤"t" => 95␤"m" => 44␤"n" => 84␤"c" => 18␤"a" => 67␤"o" => 81␤"u" => 23␤"f" => 12␤"s" => 51␤"i" => 55␤"y" => 25␤"d" => 56␤"l" => 55␤"v" => 18␤"p" => 11␤"g" => 10␤"z" => 1␤"k"…»
masak ok, that works.
m: .say for lines>>.comb.grep("a".."z").Bag.pairs.sort
camelia rakudo-moar c5ac80: OUTPUT«"a" => 67␤"b" => 17␤"c" => 18␤"d" => 56␤"e" => 150␤"f" => 12␤"g" => 10␤"h" => 103␤"i" => 55␤"k" => 8␤"l" => 55␤"m" => 44␤"n" => 84␤"o" => 81␤"p" => 11␤"r" => 71␤"s" => 51␤"t" => 95␤"u" => 23␤"v" => 18␤"w" => 23␤"x"…»
masak \o/ 18:02
TimToady m: my %a = <a 1 b 2 c 3 d 4>; my %b = <b 2 c 1 e 5>; %a »*=« %b; say %a 18:03
camelia rakudo-moar c5ac80: OUTPUT«use of uninitialized value of type Any in numeric context␤use of uninitialized value of type Any in numeric context␤("a" => 0, "b" => 4, "c" => 3, "d" => 0, "e" => 5).hash␤»
18:03 daniel-s_ left
TimToady see, those 0's are probably wrong 18:04
timotimo retupmoca: thanks :)
masak timotimo: yes, those 0s are wrong. 18:05
er, TimToady*
TimToady but we can't just default undefineds to the identity element, or be break autoviv 18:06
s/be/we/
so we really need to know which args are being modified 18:07
masak oh, I meant they're wrong because »*=« should only act on keys that are in both hashes, not at-least-one.
«*=» should act on all possible keys. 18:08
TimToady no, that would be «*=» instead, according to spec
masak oh, ok.
timotimo does/should that factor in "is default"? 18:09
TimToady that should be invisible to the hyper
you fetch from the hash, and it gives you a value
you don't care if it was the default
lizmat but the * is the problem there, not the *=, right?
m: my %a = <a 1>; my %b = <b 2>; say %a »*« %b 18:10
camelia rakudo-moar c5ac80: OUTPUT«use of uninitialized value of type Any in numeric context␤use of uninitialized value of type Any in numeric context␤("a" => 0, "b" => 0).hash␤»
TimToady which is why that case should probably just copy over such values
much as reduce on a single arg just copies it, on the assumpion that applying the identity would end up with result 18:11
lizmat would that then be special cased for &infix:<*> ?
how would that work for a custom op "foo"
TimToady it's a special case for hypers
the op doesn't have to care
lizmat but suppose an op "foo" has a multiplcative component 18:12
then %a >>foo<< %b would do the wrong thing
TimToady m: my %a = <a 1 b 2 c 3 d 4>; my %b = <b 2 c 1 e 5>; %a «*=« %b; say %a 18:13
camelia rakudo-moar c5ac80: OUTPUT«("a" => "1", "b" => 4, "c" => 3, "d" => "4", "e" => 5).hash␤»
lizmat because hyper wouldn't know to copy the value if one of the sides is missing
TimToady I guess using «« is a workaround
lizmat: I don't understand your objection, can you give an example 18:14
note that «« doesn't warn either 18:15
lizmat m: sub infix:<foo> { $^a * $^b }; my %a=<a 1>; my %b = <b 2>; say %a >>foo<< %b 18:16
camelia rakudo-moar c5ac80: OUTPUT«use of uninitialized value of type Any in numeric context␤use of uninitialized value of type Any in numeric context␤("a" => 0, "b" => 0).hash␤»
18:16 rurban left, molaf left
lizmat if we special case infix:<*> to copy if one side is missing, it wouldn't know how to do that with "foo" as an operator 18:17
TimToady I never said to special-case *
I said to special-case hyper
if one side is missing, it would just copy the other side
lizmat ah, in that way, regardless of op
TimToady (except that doesn't work well with autoviv)
since it wouldn't trigger the = part of the op 18:18
lizmat hmmm.....
TimToady m: sub infix:<foo> { $^a * $^b }; my %a=<a 1>; my %b = <b 2>; %a <<foo=<< %b; say %a
camelia rakudo-moar c5ac80: OUTPUT«Not enough positional parameters passed; got 0 but expected 2␤ in sub infix:<foo> at /tmp/36S8cT_EtP:1␤ in block at src/gen/m-CORE.setting:17119␤ in sub hyper at src/gen/m-CORE.setting:17365␤ in sub hyper at src/gen/m-CORE.setting:17474␤ in…»
TimToady hmm, that's maybe another bug 18:19
lizmat possibly :-) looks like
TimToady m: sub infix:<foo> { $^a * $^b }; my %a=<a 1>; my %b = <b 2>; %a «foo=« %b; say %a 18:20
camelia rakudo-moar c5ac80: OUTPUT«Not enough positional parameters passed; got 0 but expected 2␤ in sub infix:<foo> at /tmp/sbwdG2o9ck:1␤ in block at src/gen/m-CORE.setting:17119␤ in sub hyper at src/gen/m-CORE.setting:17365␤ in sub hyper at src/gen/m-CORE.setting:17474␤ in…»
TimToady something is trying to look up the identity value, I think
m: multi infix:<foo> () { 1 }; multi infix:<foo> { $^a * $^b }; my %a=<a 1>; my %b = <b 2>; %a «foo=« %b; say %a
camelia rakudo-moar c5ac80: OUTPUT«("a" => "1", "b" => 2).hash␤»
TimToady yup 18:21
masak m: sub MAIN($year = Date.today.year) { sub last-day-of-month($month) { my $d = Date.new($year, $month, 1); $d.clone(:day($d.days-in-month)) }; say last-day-of-month($_) for 1..12 }
camelia rakudo-moar c5ac80: OUTPUT«2014-01-31␤2014-02-28␤2014-03-31␤2014-04-30␤2014-05-31␤2014-06-30␤2014-07-31␤2014-08-31␤2014-09-30␤2014-10-31␤2014-11-30␤2014-12-31␤»
masak can anyone think of a nicer, more idiomatic way to implement last-day-of-month?
specifically, does it really have to be two statements like that?
lizmat well, the last day of the month is one day before the first day of the next month 18:22
TimToady add one month, change to 1, back up one day
masak not sure we have "add one month" :)
18:22 anaeem1 joined
masak oh wait, we have .delta :P 18:22
TimToady yeah, somebody implemented that once upon a time, I FORGET WHO!!! 18:23
masak :P
m: sub MAIN($year = Date.today.year) { sub last-day-of-month($month) { Date.new($year, $month, 1).delta(1, month).delta(-1, day) }; say last-day-of-month($_) for 1..12 }
camelia rakudo-moar c5ac80: OUTPUT«2014-01-31␤2014-02-28␤2014-03-31␤2014-04-30␤2014-05-31␤2014-06-30␤2014-07-31␤2014-08-31␤2014-09-30␤2014-10-31␤2014-11-30␤2014-12-31␤»
masak delta++
dalek kudo/nom: aff2a66 | (Elizabeth Mattijsen)++ | src/core/metaops.pm:
Move tests outside of loops for better performance
TimToady still kinda hates that API 18:24
18:24 bjz left
lizmat jnthn: ^^^ does it make sense to make this type of optimization 18:24
or is that something that will be optimized automagically in the future ?
FROGGS TimToady: is there a DateTime API out there taht you don't hate?
DateTime always sucks 18:25
lizmat fwiw, I agree DateTime sucks :-)
TimToady I'd rather have N different deltas than to have weird constants that use up namespace without provided any other value 18:26
FROGGS I mean, drolsky++ did a great job, but there are so many absurdities you must take are of that make it hard to create a nice API
TimToady .add-month(1) is better than .delta(1,month)
18:27 anaeem1 left
FROGGS yeah, days => 1 as in P5 is slightly nicer I think 18:27
masak still not too late to change. 18:28
TimToady well, .delta(month => 1) would be better than the current form, I think
lizmat .delta( :day ) # one day
FROGGS True
TimToady smelly
FROGGS :P
masak yeah.
TimToady .add-day
masak .add-days ?
jnthn lizmat: Probably worth it for the time being
TimToady I'd make it singular so that the default of 1 makes more sense
jnthn lizmat: Maybe one day we'll be able to auto-opt it, but it's not trivial and won't happen soon 18:29
lizmat jnthn; I also timed %k{a.keys} = ()
but that was twice as slow (at least)
I guess because it's building the result list
%keys{a.keys} = () 18:30
wouldn't need the result list
masak I suggest .add and .subtract (just like CPAN's DateTime), and then day => 1 and days => 1 both work.
lizmat since this is the core setting, I couldn't use Set :-)
masak all unknown keys give an error.
TimToady not sure add/subtract is the best refacter of delta 18:31
18:31 SamuraiJack_ left
TimToady it's like reinventing signs 18:31
but if we did, .later/.earlier might make more sense 18:32
or some other more time-oriented metaphor
FROGGS I like later/earlier somehow
TimToady .forward/.back 18:33
but that's backwardly huffmanized, perhaps
TimToady wonders what's the opposte of "ago"...
lizmat If we're talking about dates, wouldn't a "string" like object like "2014-05-10" become a datetime object automatically?
18:34 hoverboard left
lizmat or something like "2014y05m10d" ? 18:34
TimToady not by itself it won't
you need some context to indicate that the string is meant as a date. Just looking like a date is insufficient 18:35
it's fine if a function wants to interpret a string that way, but a bare string shouldn't
FROGGS we have Str.path... so Str.date feels right, but I don't like ending up with a thousand methods on Cool 18:36
lizmat how about "0D20140510"
FROGGS but it would be nice if it was short
lizmat: that is still a string
TimToady Introspection of Strings Considered Harmful
lizmat m: say 0d12345.WHAT,
18:36 daniel-s_ joined
camelia rakudo-moar c5ac80: OUTPUT«===SORRY!=== Error while compiling /tmp/MWrCtt_BTf␤Two terms in a row␤at /tmp/MWrCtt_BTf:1␤------> say 0d12345.WHAT,⏏<EOL>␤ expecting any of:␤ method arguments␤ postfix␤ infix sto…» 18:36
FROGGS ahh 18:37
hmmm
lizmat 0D prefix indicates a Date
FROGGS I dunno
weird :o)
masak sounds like something a module would do, not core. 18:38
FROGGS a datetime literal... I mean, the idea is nice
dalek ecs: 337f433 | masak++ | S32-setting-library/Temporal.pod:
[S32/Temporal] retire TimeUnit enum and .delta

Use strings in .truncate-to instead.
  .delta now becomes .later and .earlier, and they expect string keys as named
arguments.
masak kudos to whoever updates Rakudo to follow the above change :)
hint: not so hard.
FROGGS we have version literals that are like \d.\d.\d, we could also parse \d-\d-\d[T\d:\d:\d] 18:39
lizmat nonono
TimToady m: say 1.2.3
camelia rakudo-moar c5ac80: OUTPUT«===SORRY!=== Error while compiling /tmp/KO5aMMqosE␤Confused␤at /tmp/KO5aMMqosE:1␤------> say 1.2.⏏3␤ expecting any of:␤ postfix␤ dotty method or postfix␤»
FROGGS leading v, yeah
TimToady std: say 1.2.3
camelia std ec3d853: OUTPUT«===SORRY!===␤Number contains two decimal points (missing 'v' for version number?) at /tmp/T101xqeoFz line 1:␤------> say 1.2⏏.3␤Check failed␤FAILED 00:01 122m␤»
lizmat I think version literals are not a good model
TimToady the rakudo message is LTA
FROGGS lizmat: but that is the most similar thing we have 18:40
lizmat true
FROGGS and I like consistency
in P5 I have a dt() sub that understands some datetime strings and epoch and returns a DateTime object 18:41
I use that everywhere and it is enough...
TimToady std: say d2014-05-10
18:41 Crazylemon64 joined
camelia std ec3d853: OUTPUT«===SORRY!===␤Undeclared routine:␤ 'd2014' used at line 1␤Potential difficulties:␤ Leading 0 does not indicate octal in Perl 6; please use 0o5 if you mean that at /tmp/7AFDeyre_M line 1:␤------> say d2014-05⏏-10␤…» 18:41
18:42 bjz joined
TimToady well, specific dates seem like an unlikely thing to want to hardcode into a program 18:42
FROGGS problem is that you sometimes want/need to specify the timezone as well
not at all
I calculate fees at $work, and there are a lot of hardcoded datetimes in there
like where the bosses decide that something gets handled differently from a given date 18:43
TimToady seems like something that would make a program go obsolete rather quickly
lizmat like 20% VAT before d2014-01-01 and 21% after
TimToady not sure it's worth the huffman cost
lizmat well, in a business like Booking.com, pretty much everything revolves around dates 18:44
FROGGS sure, I dunno if it is important enough to have that, or just nice now
TimToady well, modules will happen, once we actually have a compiler that can derive slangs
lizmat anybody doing any logging: dates
TimToady or it'd be a pretty trivial macro once we have is-parsed 18:45
btyler_ date literals seem like an invitation to put something in code that should probably live in a database of some sort
not that perl has ever been in the business of laying down laws like that :) just a feeling
lizmat btyler_: you underestmate the about of ad-hoc processing of logs that occur in companies
*amount 18:46
btyler_ that's fair
was thinking about it from a business logic point of view, log processing is a good counterexample
lizmat and we don't want Hadoop to take that all :-)
masak I agree that literal dates in core don't carry their weight enough to bother.
TimToady a date comparison routine could certainly upgrade strings to date
FROGGS especially in one-liners a date(time) literal would make sense, aye
masak even one-liners could load a module. 18:47
FROGGS but but... it is extra typing :P
not saying it has to be in core though
TimToady
.oO(autoloading any language tweaks that recognize the current syntax error...)
 18:48
"Oh, hey, you just wrote some COBOL, I'll switch to that for you."
lizmat TimToady: regarding $*OS, is that a keeper or should that be deprecated in favour of $*DISTRO ? 18:49
dalek kudo-star-daily: dc31b7b | coke++ | log/ (5 files):
today (automated commit)
rl6-roast-data: 5b09b57 | coke++ | / (6 files):
today (automated commit)
rl6-roast-data: 3bf8de5 | coke++ | / (6 files):
today (automated commit)
TimToady "This looks like both Fortran and Python. I'll just run it both ways for you."
FROGGS *g*
lizmat the only mention in the spec is in S32/IO:836 18:50
vendethiel was it VB that had a syntax for date S?
TimToady $*OS is a zombie, it just keeps coming back after we kill it
FROGGS lizmat: ethercalc.org/wrwsbhvuim
vendethiel #01/01/2014#
18:50 Sqirrel left
TimToady either you want $*KERNEL or you want $*DISTRO, generally 18:50
FROGGS lizmat: last time TimToady admitted that we might are allowed to keep $*OS for some reason... but $*KERNEL was preferred I think 18:51
lizmat vendethiel: if we do that, please let us stick to YYYY-MM-DD 18:52
vendethiel please don't, though :P
lizmat because your format is ambiguous as to MM-DD-YYYY or DD-MM-YYYY
btyler_ "day month or month day? arrrgh!"
FROGGS vendethiel: we will stick to YYYY-MM-DD! :P
vendethiel I'm french, so it's always dd/mm/yyyy here.
not the bad one with day in the middle 18:53
TimToady the only thing $*OS has going for it is its brevity
lizmat well, that me be logical in some way, but no wrt to e.g. sorting :-)
well, $*OS could be short for $*KERNEL.name ?
tadzik new game ready to test at github.com/tadzik/steroids
this time it needs SDL2_ttf too 18:54
18:54 rurban joined
btyler_ tadzik: somehow your Makefile was categorized as R, heh 18:54
FROGGS tadzik: sounds like you need Alien::SDL2 :o)
tadzik heheh
oh, btw, makefiles are a mess
btyler_ I really wish github would let authors just correct the language detection. I have several repos that github thinks are 2/3rds COBOL because some files have a .cob extension 18:55
tadzik since you need to run both
and I forgot to update the firsst one
ok all good now
TimToady lizmat: it's still not what people usually thing of when you say "OS", so it'll confuse people that think it should mean $*DISTRO.name, or something nebulous between distro and kernel
lizmat so, deprecate $*OS ? 18:56
18:56 clkaoud left, pnu left, flussence left, _sri left
TimToady like that portion of your distro that came from GNU, which is the part properly called GNU/Linux, but that's neither the distro nor the kernel 18:56
FROGGS TimToady: it would be the kernal.auth in my book: ethercalc.org/wrwsbhvuim 18:57
TimToady gah, that's a commodore spelling
avoiding that misspelling is a good reason to shorten it to $*KERN, while teaching people the correct spelling is a good reason to keep it $*KERNEL :) 18:58
18:58 rurban left
FROGGS this list was intended to be a start to sort things out... but I was the only editor, so.... :/ 18:58
ohh, that is what you mean :o) 18:59
masak m: say Date("2014-05-10") 19:00
camelia rakudo-moar c5ac80: OUTPUT«Cannot find method 'Date'␤ in block at /tmp/Higo_XFrZW:1␤␤»
TimToady I'm sure that Linus would pitch a complete fit if you told him that GNU was the auth of his kernel
masak m: say Date.new("2014-05-10")
camelia rakudo-moar c5ac80: OUTPUT«2014-05-10␤»
lizmat afk for a few hours for some mindless entertainment: en.wikipedia.org/wiki/Eurovision_So...ntest_2014
masak oh, that's today? 19:01
TimToady the GNU part of GNU/Linux is only the add-ons outside the kernel
the part that $*KERNEL is specifically meant to exclude, because those things are added by the $*DISTRO 19:02
tadzik sees the polish song participating and cringes a little
TimToady or perhaps by the $*VM
19:03 darutoko left 19:04 Khisanth left
TimToady $*ARCH < $*KERNEL < $*DISTRO < $*VM # more or less 19:06
assuming we can include $*VM as part of a $*DISTRO, even if it wasn't originally supplied on the .iso :) 19:07
presumably the $*DISTRO gave you the means of installing the $*VM
19:07 clkaoud joined, pnu joined, flussence joined, _sri joined
TimToady we may get other layers in that stack eventually, given virtualization of kernels 19:07
though anything called "virtual" is introducing a feedback loop of some sort 19:08
so a VM is trying to pretend to be the root of a new $*ARCH, in a way 19:09
and the VM will have its own kernel and utilities built on its architecture, so maybe we need something a bit more recursive here 19:10
19:11 rurban joined
btyler_ tadzik: trying to get space invaders going...makefile needed tweaking to include SDL2_ttf so sdlwrapper would compile; now Steroids::Colour seems to be missing? 19:12
TimToady so $*ARCH on a vm might say "JVM" but there's some way to get to $*JVM.native-arch
TimToady suspects we will only get more layers of machine emulation in the future 19:13
tadzik btyler_: hmm 19:14
btyler_: makefile-sdlwrapper should be fine
the other makefile was just updated 20 minutes aog
after I said "ready for testing", sorry :)
19:14 plobsing joined
btyler_ tadzik: Configure.pl seems to want to work on Makefile, rather than makefile.sdlwrapper (which contains some paths from your machine). anyways, that's a super easy tweak. where can I grab steroids::colour? 19:18
19:18 Khisanth joined
tadzik oh, I didn't commit it! Stupid me 19:18
Configure.pl may be outdated, yeah :/
btyler_: fixed, thansk :) 19:19
just pull
rindolf Hi all.
Happy Saturday.
btyler_ tadzik: is there also an assets/font.ttf waiting to be pushed? :) 19:21
19:21 clkaoud left, pnu left, flussence left, _sri left, molaf joined
TimToady so, obviously openbsd, netbsd, freebsd, etc are diffierent distros, but are they different kernels? 19:23
at what point does the species known as 'bsd kernel' turn into multiple species?
or maybe those are different kernels now, with multiple distros on top of each of those? 19:24
19:24 rurban left
TimToady Linnean hierarchies don't deal all that well with evolutionary branching... 19:25
one might posit that the kernel is most like the actual genome, but distros are in the business horizontal gene transfer 19:26
but all these analogies break down sooner or later in the face of reality
the taxonomy of execution contexts is not a solved problem, sigh... 19:27
FROGGS I dunno, that is why I tried to make up that list and put in real values... 19:28
tadzik btyler_: oh, of course :O 19:29
btyler_: it should all be there now :o
nwc10 TimToady: I don't know either. and Debian/kfreebsd messes it all up further 19:30
I tend to think of the *BSDs as rabbits, because they seem to be breedng like ...
19:32 clkaoud joined, pnu joined, flussence joined, _sri joined
btyler_ tadzik: yay! got it working :) good stuff 19:32
explosions look excellent, and it runs very smoothly 19:33
tadzik smoothly? Neat :D 19:35
btyler_: if you have an xbox controller, it should work too
btyler_ oh! sadly no, but that's awesome
tadzik not on the menu though :D I need to update that
19:36 anaeem1 joined
colomon url? 19:36
colomon has been playing a good bit of Galaga '88 lately.... 19:37
btyler_ a few things I had to tweak to get it going: configure.pl needed -lSDL2_ttf added to line 31; $*VM<config><dll> for OSX includes "lib" in the name, so after the %s subst the makefile produces sdlwrapperlib.dylib (and, of course, lib/Steroids/SDL.pm is looking for sdlwrapper.dylib)
tadzik colomon: github.com/tadzik/steroids 19:38
btyler_: yeah, Configure.pl is a bit rusty, I need to update it
thanks for playtesting :)
btyler_ and (this is just me being a newb, I'm sure), SpaceInvaders sets blib/lib as the lib location, but the repo just has a /lib; I just tweaked the 'use lib' line in spaceInvaders, but I suspect there's a way I'm supposed to turn lib into blib/lib :) 19:39
I've been really enjoying the various steroids-based games, really exciting to see how concise and straightforward the game logic is 19:40
19:41 molaf_ joined 19:43 clkaoud left, pnu left, flussence left, _sri left
TimToady btyler_: perhaps there are some graphical RosettaCode entries that you're ready to write now :) 19:43
19:44 molaf left
tadzik btyler_: makefile should generate blib/lib for you 19:45
precompile all the things
btyler_ TimToady: what kind of entries? I always thought the idea with RC was to avoid problems that likely required substantial external libs 19:46
TimToady well, most language require external libs for a lot more than Perl does, so external libs are fine 19:47
rosettacode.org/wiki/Image_noise is an example of something that's really an animation sort of task 19:48
colomon tadzik: SDL_image.h ? 19:50
btyler_ hm, neat. heh, looks like some of the entries are pretty thin wrappers around openGL
TimToady most of the languages pull in libraries for that tsk 19:51
task too
only languages that try to enforce a particular graphical context are exempt, like basic, icon, and such 19:52
most language know better than to try to do thta
tadzik colomon: what abou tit?
TimToady that too
tadzik oh, may need sdl2-image installed
colomon tadzik: that would probably be it
19:54 clkaoud joined, pnu joined, flussence joined, _sri joined 19:56 daniel-s_ left
nwc10 where should I look to find out which dynamic vars are implemented, such as $*OS ? 19:56
colomon tadzik: SDL_ttf?
tadzik colomon: yeah, that too :) 19:57
19:57 daniel-s_ joined
moritz m: say PROCESS::.keys 19:58
camelia rakudo-moar aff2a6: OUTPUT«$FATAL &cwd &chdir $IN $OUT $ERR $INITTIME $TZ @ARGS $ARGFILES %ENV $VM $PERL %CUSTOM_LIB @INC $CWD $OS $OSVER $PID $EXECUTABLE $EXECUTABLE_NAME $PROGRAM_NAME $TMPDIR $THREAD $SCHEDULER␤»
moritz nwc10: ^^
nwc10 mmm, bother. that's just going to say "linux" 19:59
and "moarvm"
I was rather hoping to show the architecture
colomon tadzik: library not found for -lrt? 20:00
nwc10 OK. This will do: 20:01
$ ./perl6-m -e 'shell("uname -a")'
Linux gcc1-power7.osuosl.org 3.8.8-202.fc18.ppc64p7 #1 SMP Thu Apr 18 14:11:12 MST 2013 ppc64 ppc64 ppc64 GNU/Linux
running spectests
tadzik colomon: lrt? That doesn't ring a bell. What happens if you remove it? 20:02
colomon it's librt, realtime stuff?
tadzik I don't recall using it :)
ah
it's probably from Linux makefile 20:03
I should really fix Configure.pl :<
colomon looks like it builds with it removed.
nwc10 jnthn: patches for dancing bears coming soon.
20:03 pecastro joined
colomon argh, Cannot locate native library './sdlwrapper.dylib' 20:04
because the makefile generates a .so instead
btyler_ colomon: look for "sdlwrapperlib.dylib" in that dir 20:05
rename it sdlwrapper.dylib --> success
colomon no such file
nwc10 logistics/location fail. There is no tonic here, and whilst at home the whisky isn't fantastic, it is a better choice than here 20:06
btyler_ colomon: oh...did 'make' finish successfully? 20:07
jnthn nwc10: Oh, I've got a nice big bottle of Yeti to drink here :)
colomon yes?
or no? 20:08
btyler_ colomon: bizarre. osx 10.9.2 here on latest moar/nqp/rakudo, makefile spat out sdlwrapperlib.dylib, changing makefile or renaming (to ditch the 'lib' suffix) got lib/Steroids/SDL to find it 20:09
ls
erg, ww
20:09 clkaoud left, pnu left, flussence left, _sri left
colomon I can see -o sdlwrapperlib.dylib in the Makefile 20:10
okay, I touched sdlwrapper.c and now I get the error load and clear.
20:11 daniel-s__ joined 20:12 daniel-s_ left
colomon I had to add -lSDL2_ttf to the Makefile 20:13
Cannot locate native library './sdlwrapper.dylib' 20:15
grrrr
20:16 daniel-s__ left 20:18 zakharyas left
tadzik :o 20:18
I'll goan fix that Configure.pl
20:19 pippo joined 20:21 ajr_ left
colomon it's right there, I cannot imagine why it cannot find it. 20:22
20:22 daniel-s__ joined 20:24 guru joined, guru is now known as Guest87084, rurban joined, colomon left 20:25 hoverboard joined, Guest87084 is now known as ajr_ 20:26 _sri joined, flussence joined, clkaoud joined, pnu joined 20:27 rindolf left 20:30 rurban left 20:32 btyler_ left 20:33 daniel-s__ left, dotslashb joined 20:36 dwarring joined
jnthn just did a patch to optimize out unused %_ using the var stuff in the optimizer 20:38
nwc10 faster dancing bears? 20:39
20:39 daniel-s__ joined
jnthn Saves the Hash allocation 20:39
And in theory the low level hash alloc and op to fetch it too
Still need to check that spesh's unused detection spots that and kills it
20:41 SevenWolf joined 20:43 Crazylemon64 left 20:44 daniel-s__ left 20:46 daniel-s__ joined 20:52 ajr_ left 20:53 Crazylemon64 joined 20:57 hoverboard left
jnthn hah, with adding some missing :pure it does :) 21:01
21:10 colomon joined
FROGGS jnthn: another percent faster? :o) 21:19
jnthn Takes off some more of the GC pressure. 21:20
It's not actually a huge win, though...
If the GC sucked more it'd be easier to get wins by reducing allocation... :P
FROGGS hehe 21:21
jnthn sees what the latest profile tells him
darn, p6decontrv is costly... 21:24
21:25 rurban joined 21:29 rurban left
timotimo aye, the low-level allocation is a pointer bump and a little thing 21:31
decont return value?
jnthn Yeah, but really it's "if it has an rw container take it away and put an ro one there instead"
Meaning it allocates that ro container 21:32
Which is actually down to flattening.
I think I once only bothered to do it on non-Iterable things 21:33
But then itemized Parcel got sad
'cus Parcel ain't Iterable
dalek kudo/nom: b2b6b19 | jnthn++ | src/Perl6/ (2 files):
Optimize out %_ when unused.
 21:36
jnthn I should probably try it again including Parcel in the check 21:37
Or even better
Should do away with the C op entirely
So that spesh can make the checks cheaper.
timotimo that sounds like a good idea 21:44
21:45 daniel-s__ left
FROGGS java.lang.ClassCastException: __P6opaque__72 cannot be cast to org.perl6.nqp.sixmodel.reprs.VMExceptionInstance 21:46
I am throwing from rakudo... do I need to decont there?
jnthn Uh..hm 21:47
I doubt it
Where are you getting the thing you're trying to cast?
FROGGS well, I am creating it this way: my Mu $ex := nqp::newexception();
so no decont needed
in a handler made by nqp::handle 21:48
ahh, perhaps it is the payload / the label
hmmm, that does not make much sense either
jnthn yeah...the payload will typically be a P6opaque
FROGGS but it is unlikely that I want to cast that to a VMEx 21:49
jnthn Yeah
21:50 daniel-s__ joined 21:54 rurban joined 21:56 xenoterracide__ joined 21:58 rurban left 22:05 rurban joined
FROGGS gnight all 22:05
22:06 xenoterracide__ left 22:09 rurban left, SHODAN left
jnthn 'night, FROGGS++ 22:10
22:10 SHODAN joined 22:12 xenoterracide__ joined
dalek p/lexopts: 03349b2 | (Stéphane Payrard)++ | src/QRegex/Cursor.nqp:
Fixing --target=parse on nqp

Was working fine on rakudo		 22:17
22:17 dalek left, rurban joined 22:18 dalek joined, ChanServ sets mode: +v dalek
masak m: (1, 2, 3).permutations 22:20
camelia ( no output )
masak m: say (1, 2, 3).permutations.perl
camelia rakudo-moar aff2a6: OUTPUT«([1, 2, 3], [1, 3, 2], [2, 1, 3], [2, 3, 1], [3, 1, 2], [3, 2, 1]).list␤»
22:21 rurban left 22:28 spider-mario left 22:31 daniel-s__ left 22:32 daniel-s__ joined 22:38 denis_boyun_ left
dalek kudo/nom: ccddd17 | jnthn++ | src/vm/moar/ops/perl6_ops.c:
Avoid unrequired container allocs in p6decontrv.
 22:41
masak 'night, #perl6 22:43
lizmat well, that was close :-) 22:52
22:52 anaeem1 left
jnthn didn't watch it, but heard Austria won and Poland resorted to striptease... 22:55
22:56 hoverboard joined 22:57 xenoterracide__ is now known as xenoterracide
lizmat yes, Poland was, what shall I say, blond airheads? 22:58
jnthn: I just pulled and ran a spectest 22:59
so far, no immediately gains to be seen
one failure, and 2 todo's passing
jnthn Yeah, you probably need something a bit longer running than spectests for that.
lizmat not ok 6 - :(:a($x), :b($y)) ~~ :(:a(:b(:c($x)))) 23:00
in S03-smartmatch/signature-signature.t
jnthn hmm 23:01
I didn't see that fail here...
(As in, just re-checked my last spectest output with latest matches and that file passed)
lizmat going back 1 commit, see the result then
seems to be the %_ commit 23:04
double checking
yup, reverting to aff2a6618ee3d746a4d4f5bdfa91854f54f0f91d makes the pb go away 23:06
jnthn *sigh*
I really don't fancy losing that opt. 23:07
lizmat well, maybe the test is bogus (it feels pretty weird to me, anyway)
jnthn wonders how on earth it realtes to that test
No matter how many times I run it here, I can't get it to fail 23:08
lizmat odd 23:11
well, I guess we have to wait for a roast from [Coke] then
jnthn *nod*
lizmat m: say :(:a($x), :b($y)) ~~ :(:a(:b(:c($x))))
camelia rakudo-moar b2b6b1: OUTPUT«True␤»
jnthn I'm also running with spesh_trace, but that's not meant to make behavior differences...
lizmat $ perl6 -e 'say :(:a($x), :b($y)) ~~ :(:a(:b(:c($x))))'
False
jnthn uh 23:12
I get it to say False too
But the tests passes?!
lizmat hmmm.. it is supposed to be False , I see now in the test 23:13
jnthn oh...
uh, so why does evalbot get it here as true...
m: say :(:a($x), :b($y)) ~~ :(:a(:b(:c($x))))
camelia rakudo-moar b2b6b1: OUTPUT«True␤»
jnthn m: say :(:a($x), :b($y)) ~~ :(:a(:b(:c($x)))) for ^20
camelia rakudo-moar b2b6b1: OUTPUT«False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤True␤False␤False␤False␤False␤False␤False␤False␤»
jnthn o.O
lizmat seems like an artefact in the test 23:15
if I make it the *only* test in the list, it passes
ok 1 - :(:a($x), :b($y)) ~~ :(:a(:b(:c($x)))) 23:16
jnthn OK. I think that given we're getting unstable behavior from it here on channel too, we might conclude it's not that particular patch...
23:18 rurban joined
lizmat If I leave in the last 4 tests, it fails 23:19
there *is* one test in there for *%_
:(:$x, Int :$y, Cool :$z, *%_), :(*%_), False, 23:20
m: say :(*%_) ~~ :(:x($x), Int :y($y), Cool :z($z), *%_)
camelia rakudo-moar b2b6b1: OUTPUT«False␤»
23:21 Rotwang left
lizmat gist.github.com/lizmat/f8f7377b36849a473013 23:21
is the diff of the test file that is needed to make it fail 23:22
23:22 rurban left
lizmat taking out the line with ":(:$x, Int :$y, Cool :$z, *%_), :(*%_)," makes it pass again 23:23
so it seems to be related to :(*%_)
jnthn I'm unconvinced.
The patch you reverted was the one about p6decontrv, not *%_? 23:24
Also, the nature of the *%_ thing I did wouldn't give unstable behavior, more likely explosive behavior 'cus it'd removed something.
lizmat I first reverted ccddd17f8538343247623fd076882ca3643c0995 and the pb was still there
I then reverted b2b6b19ac74509ff30aee7ef74a1ca142bfe5a81 and the problem went away 23:25
jnthn Hmm
But then, why do I not see the issue here... 23:26
lizmat all the tests around the *%_ test are supposed to be False
so if they are false for the *wrong* reason, we wouldn't see it
I just noticed another test with *%_ 23:27
if I activate that one, and disable the other one, it also fails the test
jnthn Weird 23:28
lizmat hmmm.... other combinations also fail
seems I'm too tired for this now
jnthn Yeah. Thing is, I think I've seen that test fail randomly in the past too...
*nod*
We had one other sparodic failure in first.t for a while under spesh
During the last days work I've fixed the odd thing that could potentially have caused metastable behavior. I'm running on spesh_trace here, which has those fixes. So maybe it's something around that... 23:29
Hard to know.
23:29 dmol left
lizmat ok, let's wait for those changes to merge and see what gives afterwards, ok? 23:29
jnthn Works for me 23:30
We've got a spesh slot corruption issue to resolve before it can merge, it seems.
lizmat for me as well... :-) so gnight jnthn and #perl6!
jnthn 'night, lizmat
23:31 BenGoldberg joined 23:48 xenoterracide left 23:51 berekuk left 23:59 xragnar is now known as Guest66156, xragnar_ joined, Guest66156 left, xragnar_ is now known as xragnar

Powered by Cro and the Raku® Programming Language.

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

Thank you!