»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or rakudo:, or /msg camelia p6: ... | irclog: irc.perl6.org or colabti.org/irclogger/irclogger_logs/perl6 | UTF-8 is our friend!
Set by moritz on 22 December 2015.
geraud m: say ૪ + ৪; 00:02
camelia rakudo-moar f0c546: OUTPUT«8␤»
[Coke] lizmat: I thought you just said that was problematic? 00:04
m: say 8/૪
camelia rakudo-moar f0c546: OUTPUT«2␤»
lizmat m: say (^3.5 .. ^7.5).minmax 00:05
camelia rakudo-moar f0c546: OUTPUT«Range objects are not valid endpoints for Ranges␤ in block <unit> at /tmp/MwrbQZx10K:1␤␤»
lizmat m: say (^3.5..^7.5).minmax
camelia rakudo-moar f0c546: OUTPUT«Range objects are not valid endpoints for Ranges␤ in block <unit> at /tmp/rWEJpqjRNe:1␤␤»
mscha p6: for -Inf^..^Inf { .say } 00:06
lizmat m: say (3.5^..^7.5).minmax
camelia rakudo-moar f0c546: OUTPUT«(timeout)-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-Inf␤-I…»
rakudo-moar f0c546: OUTPUT«4.5..6.5␤»
lizmat [Coke]: ^^^ not incorrect result
*note
$ 6 'say (3.5^..^7.5).minmax' 00:07
Cannot return minmax on Range with excluded ends
$ 6 'say (-Inf..Inf).minmax'
(-Inf Inf)
pochi m: class Foo { has $!secret = 42; method magic(Foo $other) { return $!secret + $other!secret } }; say Foo.new().magic( Foo.new() )
camelia rakudo-moar f0c546: OUTPUT«5===SORRY!5=== Error while compiling /tmp/85EG25zOmP␤No such private method 'secret' for invocant of type 'Foo'␤at /tmp/85EG25zOmP:1␤------> 3(Foo $other) { return $!secret + $other!7⏏5secret } }; say Foo.new().magic( Foo.new␤»
_nadim timotimo: I was serious, hapilly broken AKA, I can do something with it. Compared to unhapilly broken and I don't understand what's going on :)
mscha What's the diff between -Inf^..^Inf and -Inf..Inf? ;)
pochi how do I access the private variable secret for an object other than the invocant? 00:08
lizmat mscha: the ^ indicates that the endpoint is to be excluded
timotimo _nadim: ok, that's good!
mscha I know, but how do you exclude -Inf?
lizmat mscha: in the case of Inf, in practice it doesn't make any difference
dalek kudo/nom: 0b80308 | lizmat++ | src/core/Range.pm:
Implement basic Range.minmax, so ^Inf won't hang

Also fail if any of the endpoints are excluded *and* we don't have an integer range.		 00:09
lizmat [Coke]: ^^^
and on that note, I wish #perl6 a good night! 00:10
00:10 smls left
[Coke] lizmat++ 00:10
mscha p6: my $r1 = -Inf^..^Inf; my $r2 = -Inf..Inf; say -Inf ~~ $r1; say -Inf ~~ $r2; for |$r1 { .say; last; } 00:12
camelia rakudo-moar b69aa5: OUTPUT«False␤True␤-Inf␤»
mscha I hate math with infinites... 00:13
00:16 MadcapJake left
labster m: say Inf + 1; 00:17
camelia rakudo-moar b69aa5: OUTPUT«Inf␤»
00:22 yqt left
mscha m: say 2×Inf - Inf; 00:23
camelia rakudo-moar b69aa5: OUTPUT«NaN␤»
00:24 MadcapJake joined
mscha m: say ∞; 00:25
camelia rakudo-moar b69aa5: OUTPUT«Inf␤»
[Coke] m: say ∞**∞ 00:26
camelia rakudo-moar b69aa5: OUTPUT«Inf␤»
[Coke] anyone know how to update rakudobrew on hack? 00:27
mscha m: say 2.9999999 ≈ 3;
camelia rakudo-moar b69aa5: OUTPUT«5===SORRY!5=== Error while compiling /tmp/BEd32TN4rv␤Confused␤at /tmp/BEd32TN4rv:1␤------> 3say 2.99999997⏏5 ≈ 3;␤ expecting any of:␤ infix␤ infix stopper␤ postfix␤ statement end␤ statemen…»
mscha m: say 2.999999 ≈≈ 3;
camelia rakudo-moar b69aa5: OUTPUT«5===SORRY!5=== Error while compiling /tmp/1LihM8ZQVT␤Confused␤at /tmp/1LihM8ZQVT:1␤------> 3say 2.9999997⏏5 ≈≈ 3;␤ expecting any of:␤ infix␤ infix stopper␤ postfix␤ statement end␤ statem…»
mscha say π ≠ e; 00:28
m: say π ≠ e;
camelia rakudo-moar b69aa5: OUTPUT«5===SORRY!5=== Error while compiling /tmp/eiVAUWkbap␤Confused␤at /tmp/eiVAUWkbap:1␤------> 3say π7⏏5 ≠ e;␤ expecting any of:␤ infix␤ infix stopper␤ postfix␤ statement end␤ statement modif…»
mscha m: sub infix<≠> { $^a != $^b }; say π ≠ e; 00:29
camelia rakudo-moar b69aa5: OUTPUT«5===SORRY!5=== Error while compiling /tmp/oUNB9M29XK␤Missing block␤at /tmp/oUNB9M29XK:1␤------> 3sub infix7⏏5<≠> { $^a != $^b }; say π ≠ e;␤ expecting any of:␤ new name to be defined␤»
mscha m: sub infix<≠>($a, $b) { $a != $b }; say π ≠ e;
camelia rakudo-moar b69aa5: OUTPUT«5===SORRY!5=== Error while compiling /tmp/s107E8hJhy␤Missing block␤at /tmp/s107E8hJhy:1␤------> 3sub infix7⏏5<≠>($a, $b) { $a != $b }; say π ≠ e;␤ expecting any of:␤ new name to be defined␤»
mscha m: sub infix:<≠>($a, $b) { $a != $b }; say π ≠ e;
camelia rakudo-moar b69aa5: OUTPUT«True␤»
mscha m: sub infix:<≠> { $^a != $^b }; say π ≠ e; 00:30
camelia rakudo-moar b69aa5: OUTPUT«True␤»
00:30 mohae_ joined 00:31 mohae left
AlexDaniel mscha: actually, I've always wondered if these should be built in 00:36
m: say 2.999999 ≅ 3 00:37
camelia rakudo-moar 0b8030: OUTPUT«False␤»
AlexDaniel m: say 2.999999e0 ≅ 3
camelia rakudo-moar 0b8030: OUTPUT«False␤»
AlexDaniel m: say 2.99999999999999e0 ≅ 3
camelia rakudo-moar 0b8030: OUTPUT«False␤»
AlexDaniel m: say 2.9999999999999999999e0 ≅ 3
camelia rakudo-moar 0b8030: OUTPUT«True␤»
AlexDaniel m: say 2.9999999999999999999 ≅ 3
camelia rakudo-moar 0b8030: OUTPUT«True␤»
timotimo [Coke]: it seems like every user has their own .rakudobrew
AlexDaniel m: say 2.999999999999999 ≅ 3
camelia rakudo-moar 0b8030: OUTPUT«True␤»
AlexDaniel m: say 2.999999999999 ≅ 3
camelia rakudo-moar 0b8030: OUTPUT«False␤»
AlexDaniel mscha: approx is there though 00:38
mscha Cool
AlexDaniel ≤ ≥ ≠ hmm
00:39 skids joined
mscha m: say 3 ≤ π; 00:39
camelia rakudo-moar 0b8030: OUTPUT«5===SORRY!5=== Error while compiling /tmp/mD0rz8isCF␤Confused␤at /tmp/mD0rz8isCF:1␤------> 3say 37⏏5 ≤ π;␤ expecting any of:␤ infix␤ infix stopper␤ postfix␤ statement end␤ statement modif…»
[Coke] timotimo: there is a /home/rakudobrew though
AlexDaniel .u ≛
yoleaux U+225B STAR EQUALS [Sm] (≛)
AlexDaniel what a neat symbol :D
00:39 _nadim left, kanishka left
AlexDaniel .u ⩲ 00:40
yoleaux U+2A72 PLUS SIGN ABOVE EQUALS SIGN [Sm] (⩲)
mscha .u 🕴
yoleaux No characters found
timotimo oh?
AlexDaniel too bad there's no MINUS SIGN ABOVE EQUALS SIGN
mscha: yoleax is a bit old in that sense 00:41
timotimo huh. i can't seem to sudo on hack
AlexDaniel mscha: there are also ⩵ and ⩶ but nobody knows how these are supposed to look like with fixed-width fonts :) 00:42
timotimo must have changed my password and forgotten to update it in the password database?
mscha .u 💩 00:43
yoleaux U+1F4A9 PILE OF POO [So] (💩)
[Coke] timotimo: I can.
timotimo i wouldn't just go into /home/rakudobrew and rakudobrew self-upgrade or something 00:44
since i don't know if it's part of any kind of setup 00:45
like a cron job or something
[Coke] m: say $*PERL; 00:50
camelia rakudo-moar 0b8030: OUTPUT«Perl 6 (6.b)␤»
[Coke] . ask TimToady - you want to change our version to be 6.c ? I'd like to have it running for a day at least before release.
00:52 mohae_ left
[Coke] .ask TimToady - you want to change our version to be 6.c ? I'd like to have it running for a day at least before release. 00:53
yoleaux [Coke]: I'll pass your message to TimToady.
00:53 mohae joined 00:54 mohae left
TimToady sure, we can do that 00:54
yoleaux 00:53Z <[Coke]> TimToady: - you want to change our version to be 6.c ? I'd like to have it running for a day at least before release.
00:54 mohae joined
TimToady maybe call it 6.c.RC1 first? 00:54
00:55 BenGoldberg joined
mscha sub infix:<≤> { $^a <= $^b }; 3 ≤ 7 ≤ 5; # oops 00:55
p6: sub infix:<≤> { $^a <= $^b }; 3 ≤ 7 ≤ 5; # oops
camelia ( no output )
mscha p6: sub infix:<≤> { $^a <= $^b }; 3 ≤ 7 ≤ 5;
camelia ( no output ) 00:56
TimToady say?
geekosaur need a say in there
mscha p6: sub infix:<≤> { $^a <= $^b }; say 3 ≤ 7 ≤ 5;
camelia rakudo-moar 0b8030: OUTPUT«True␤»
TEttinger p6: sub infix:<≤> { $^a <= $^b }; say 3 ≤ 7 ≤ 5;
camelia rakudo-moar 0b8030: OUTPUT«True␤»
TimToady and you probably want 'is equiv(&[<=]) or so
TEttinger neat
TimToady otherwise it has no hope of chaining
BenGoldberg p6: sub infix:<≤> is assoc('List') { $^a <= $^b }; say 3 ≤ 7 ≤ 5;
camelia rakudo-moar 0b8030: OUTPUT«False␤»
TimToady it's not doing what you think
nor is that
all you're testing is that True <= 5 00:57
BenGoldberg p6: say True <= 5
camelia rakudo-moar 0b8030: OUTPUT«True␤»
mscha p6: sub infix:<≤> is equiv(&[<=]) { $^a <= $^b }; say 3 ≤ 7 ≤ 5;
camelia rakudo-moar 0b8030: OUTPUT«5===SORRY!5=== Error while compiling /tmp/tmpfile␤Routine given to equiv does not appear to be an operator␤at /tmp/tmpfile:1␤»
TimToady or some such
BenGoldberg p6: say True <= -3
camelia rakudo-moar 0b8030: OUTPUT«False␤»
BenGoldberg p6: sub infix:<≤> { $^a <= $^b }; say -3 ≤ -2 ≤ -1; 00:58
camelia rakudo-moar 0b8030: OUTPUT«False␤»
BenGoldberg p6: say -3 <= -2 <= -1;
camelia rakudo-moar 0b8030: OUTPUT«True␤»
BenGoldberg p6: say [<=] -3, -2, -1; 00:59
camelia rakudo-moar 0b8030: OUTPUT«True␤»
BenGoldberg p6: sub infix:<≤> { $^a <= $^b }; say [≤] -3, -2, -1;
camelia rakudo-moar 0b8030: OUTPUT«False␤»
00:59 adu left
TimToady m: sub infix:<≤> is equiv(&infix:«<=») { $^a <= $^b }; say 3 ≤ 7 ≤ 5; 00:59
camelia rakudo-moar 0b8030: OUTPUT«5===SORRY!5=== Error while compiling /tmp/RcXZiuBhDP␤Routine given to equiv does not appear to be an operator␤at /tmp/RcXZiuBhDP:1␤»
TimToady hmm, bug
m: say &[<=>].prec 01:00
camelia rakudo-moar 0b8030: OUTPUT«Method 'prec' not found for invocant of class 'Sub+{<anon|63995968>}'␤ in block <unit> at /tmp/nMzURGtrjB:1␤␤»
TimToady m: say &[<=].prec
camelia rakudo-moar 0b8030: OUTPUT«Method 'prec' not found for invocant of class 'Sub+{<anon|63995968>}'␤ in block <unit> at /tmp/PNAJEGHx6u:1␤␤»
TimToady that's also buggy 01:01
lichtkind perlhist is down, you know wo runs it?
TimToady m: sub infix:<≤> is prec('m=') is assoc('chaining') { $^a <= $^b }; say 3 ≤ 7 ≤ 5; 01:02
camelia rakudo-moar 0b8030: OUTPUT«5===SORRY!5=== Error while compiling /tmp/vt7gOhwIZ1␤Can't use unknown trait 'is prec' in a sub+{precedence} declaration.␤at /tmp/vt7gOhwIZ1:1␤ expecting any of:␤ rw raw hidden-from-backtrace hidden-from-USAGE␤ pure default D…»
mscha p6: sub infix:<≤> is assoc:<chain> { $^a <= $^b }; say 3 ≤ 7 ≤ 5;
camelia rakudo-moar 0b8030: OUTPUT«5===SORRY!5=== Error while compiling /tmp/tmpfile␤Can't use unknown trait 'is assoc:<chain>' in a sub+{precedence} declaration.␤at /tmp/tmpfile:1␤ expecting any of:␤ rw raw hidden-from-backtrace hidden-from-USAGE␤ pure defaul…»
TimToady m: sub infix:<≤>($a,$b) is prec('m=') is assoc('chaining') { $a <= $b }; say 3 ≤ 7 ≤ 5;
camelia rakudo-moar 0b8030: OUTPUT«5===SORRY!5=== Error while compiling /tmp/19Orkmcxvm␤Can't use unknown trait 'is prec' in a sub+{precedence} declaration.␤at /tmp/19Orkmcxvm:1␤ expecting any of:␤ rw raw hidden-from-backtrace hidden-from-USAGE␤ pure default D…»
ShimmerFairy TimToady: any particular reason we don't have ≤ and ≥ in core? We've got lots of other nifty Unicode things stuffed in there already :P 01:03
mscha p6: proto sub infix:<≤>($, $) returns Bool:D is assoc:<chain>; sub infix:<≤> { $^a <= $^b }; say 3 ≤ 7 ≤ 5;
camelia rakudo-moar 0b8030: OUTPUT«5===SORRY!5=== Error while compiling /tmp/tmpfile␤A unit-scoped sub definition is not allowed except on a MAIN sub;␤Please use the block form.␤at /tmp/tmpfile:1␤------> 3>($, $) returns Bool:D is assoc:<chain>;7⏏5 sub infix:<≤> { $^a <= …»
TimToady looks like whoever implemented it did so with a :pasttype<chain> to hardwire it, hmmm... 01:04
mscha p6: proto sub infix:<≤>($, $) returns Bool:D is assoc:<chain>; sub infix:<≤>($a, $b) { $a <= $b }; say 3 ≤ 7 ≤ 5;
camelia rakudo-moar 0b8030: OUTPUT«5===SORRY!5=== Error while compiling /tmp/tmpfile␤A unit-scoped sub definition is not allowed except on a MAIN sub;␤Please use the block form.␤at /tmp/tmpfile:1␤------> 3>($, $) returns Bool:D is assoc:<chain>;7⏏5 sub infix:<≤>($a, $b) {…»
TimToady ShimmerFairy: well, especially since it seems impossible to add them at the moment after the fact...
which is a bug, but I don't know if we'll get to it before 6.c 01:05
ShimmerFairy true, it's proving a nice test of what's possible with custom ops at the moment :)
TimToady: Other operator things that bug me is how unary precedence stuff is faked by NQP (last I checked), and how, at least in rakudo core, it's hard to set up "aliases" like constant &infix:<texas> := &infix:<☃> or however 01:06
TimToady m: constant &infix:<≤> = &infix:«<=»; say 3 ≤ 7 ≤ 5;
camelia rakudo-moar 0b8030: OUTPUT«True␤»
TimToady well, at least that works 01:07
TimToady fondly remembers the days when the saying was: "It's good that there's more than one way to do it, because most of them don't work."
01:07 yeahnoob joined
mscha TimToady: it works ... incorrectly. 01:07
timotimo that's true, hah 01:08
01:08 raiph joined
BenGoldberg m: constant &infix:<≤> = &infix:«<=»; say -3, -2, -1; 01:08
camelia rakudo-moar 0b8030: OUTPUT«-3-2-1␤»
mscha m: constant &infix:<≤> = &infix:«<=»; say 3 ≤ 7 ≤ 5; say 3 <= 7 <= 5;
camelia rakudo-moar 0b8030: OUTPUT«True␤False␤»
timotimo it probably ends up giving 3 <= 7 -> True, then True <= 5
BenGoldberg m: constant &infix:<≤> = &infix:«<=»; say -3 ≤ -2 ≤ -1;
camelia rakudo-moar 0b8030: OUTPUT«False␤»
TimToady yeah, it's applying it as a binary, i guess
ShimmerFairy something like that, even if it needs a special construct, would be the best way to handle texas/unicode variant stuff, I feel. 01:10
BenGoldberg m: sub infix:<≤> is assoc<list> { [<=] @_ }; say -3 ≤ -2 ≤ -1;
camelia rakudo-moar 0b8030: OUTPUT«True␤»
timotimo i'd almost say get a different prefix (like "operator") in there so the assignment knows to also move stuff like precedence and such over into the grammar 01:11
BenGoldberg m: sub infix:<≤> is assoc<list> { $^a <=> $^b }; say -3 ≤ -2 ≤ -1;
camelia rakudo-moar 0b8030: OUTPUT«5===SORRY!5=== Error while compiling /tmp/RiYkPYdzLQ␤Calling infix:<≤>(Int, Int, Int) will never work with declared signature ($a, $b)␤at /tmp/RiYkPYdzLQ:1␤------> 3ssoc<list> { $^a <=> $^b }; say -3 ≤ -2 7⏏5≤ -1;␤»
TimToady we already know it's an operator
that's not the problem
timotimo ah, ok
BenGoldberg m: sub infix:<≤> is assoc<list> { [<=] @_ }; say [≤] -3, -2, -1; 01:12
camelia rakudo-moar 0b8030: OUTPUT«True␤»
BenGoldberg I think that works
TimToady it only works for cascaded <=, but won't interact well with chaining ops at the right precedence
mscha m: sub infix:<≤> is assoc<list> { [<=] @_ }; say 3 ≤ 7 ≤ 5;
camelia rakudo-moar 0b8030: OUTPUT«False␤»
timotimo m: say 1 < 10 > 5
camelia rakudo-moar 0b8030: OUTPUT«True␤»
timotimo ^- that, for example
mscha m: sub infix:<≤> is assoc<list> { [<=] @_ }; say 3 ≤ 7 < 5; 01:13
camelia rakudo-moar 0b8030: OUTPUT«True␤»
BenGoldberg m: sub infix:<≤> is assoc<list> { [<=] @_ }; say 1 ≤ 10 > 5;
camelia rakudo-moar 0b8030: OUTPUT«False␤»
mscha m: sub infix:<≤> is assoc<list> { [<=] @_ }; say 3 ≤ 7 < 5; say 3 <= 7 < 5;
camelia rakudo-moar 0b8030: OUTPUT«True␤False␤»
01:14 mohae left
TimToady it has to be at chain precedence with chain associativity for it to work right, and there's just no way to get there from here at the moment 01:15
BenGoldberg m: say 1 R, 2 R, 3
camelia rakudo-moar 0b8030: OUTPUT«(2 1)␤»
TimToady that looks like a buglet of some sort 01:16
m: say [R,] 1,2,3
camelia rakudo-moar 0b8030: OUTPUT«(3 2 1)␤»
stmuk whats happening with git post Xmas? will there be two branches?
TimToady presumably
or at least, once we want to start making language changes for 6.d.alpha 01:17
01:18 mohae joined
TimToady also, rakudo doesn't have to branch the same as roast 01:18
there might well be buglets we could patch in a 6.c roast without compromising backward compat
01:19 yqt joined
TimToady and if we're forced to issue a maintenance release based on 6.c due to a security flaw, then we'd certainly need to branch 01:20
so we might as well plan for it at some point
but we can probably decide that rather more post-xmas than we are now 01:21
we can branch lazily when we need to
stmuk is it Xmas UTC or PST? :)
timotimo timotired
TimToady that will depend on who does the release when, I imagine
01:22 adu joined
timotimo good night perl6 people! 01:22
01:23 cognominal left, mscha left
timotimo i'd really be interested to see a good example where the line between "this is a buglet. we'll fix it in roast and in rakudo" and "this is a feature. it'll go into the next language version." is 01:24
skids Well one criterion would probably be "is it reasonable for someone to have relied on behavior we are about to change?" 01:26
BenGoldberg R, losing arguments is surely a buglet. 01:27
AlexDaniel skids: what if it is a new feature that nobody could have possibly relied on in the past 01:28
TimToady indeed, nobody in their right mind, with the possible exception of BenGoldberg++, would even try to do that :)
BenGoldberg On the other hand, the specific order of the results of 1 R, 2 R, 3 is probably something which shouldn't be relied on. I recall an earlier version of rakudo producing truly scrambled results ;)
skids AlexDaniel: then we'd have too rely on other criteria.
timotimo oh, and: how legit is it to push a few tests just before 6.c into roast for bugs we've discovered in the last day? 01:29
TimToady in new features, it tends to come down to whether we're going to pollute a namespace that someone has already put something else into
timotimo like my $foo = Failure.new; $foo += 1; being supposed to throw the failure
TimToady fortunately, our lexical setting allows us to add things that will automatically be shadowed in user code
yeah, that's more something we have t ofix 01:30
*to fix, even
timotimo er. if i keep being awake, i'm not going to be sleeping any time soon.
ShimmerFairy btw, how do feel about pack/unpack ? Should we hide it behind that same "experimental" thing that stuff like macros are under, or move it to a separate module entirely? I really don't want to find ourselves maintaining backward-compat with a weird subset of P5-era functionality.
TimToady good night, timotimo, we promise not to talk about anything interesting now :)
flussence IMO, pack/unpack = ewww.
timotimo no. talk about interesting things! it's far too close to christmas to look out for little me's backlog size 01:31
flussence but there isn't really a good substitute right now, so I'm not sure about hiding it
01:31 thomax left
TimToady we'll just have to play the bits of pack/unpack by ear 01:31
skids Well, how easy is it to use P5 pack/unpack via Inline::Perl5 from perl6, maybe we don't need a P6 version of it quite yet?
ShimmerFairy flussence: my concern is having to do backward-compat for something that's not designed for P6's strengths and needs :)
there was talk before of moving it out to a module, but I guess it hasn't happened yet. 01:32
01:32 thomax joined
TimToady we can always name them something else 01:32
I mean, the new shiny ones
flussence tried writing binary-network-IO stuff a week or two ago, and it ended up being an absolute trainwreck
TimToady I don't think we can fix that trainwreck before xmas 01:33
flussence thing is, I'm not sure *what* I want for that type of code. :(
ShimmerFairy TimToady: yeah, I had a binary grammar idea where I referred to them as "formats" (cf. file formats), so they could theoretically be .format/.deformat, or something like that.
BenGoldberg Maybe instead of pack/unpack we should have something resembling javascript's DataView... (just an idea).
01:33 Skarsnik left
stmuk pack/unpack sounds like ecosystem to me and I would prefer a nicer API to it 01:33
skids flussence: you mean packet dissection/construction?
TimToady we can talk about it...after xmas...
ShimmerFairy Of course, then the question is if (un)pack really deserve to be in core, if they're just going to forever be a P5-ish thing :)
TimToady is gonna go fix bugs... 01:34
AlexDaniel skids: hah, though hiding qqx in Inline::Perl5 sounds like a great idea :D
flussence skids: yeah, along with some variable-width integer parsing to get blob lengths. 01:35
ShimmerFairy Like I said, I'm only concerned about the "having to support a subset of a P5-era design in core as backwards compatibility" aspect. If that's not much of an issue, then I'm OK :)
stmuk "use JUNKIE-EVIL" for qqx injections!
skids flussence: I've run into that as well, and there's a gist out there with some vague ideas. 01:36
Juerd In my opinion, current Perl 6 eval is barely useful, and it will be impossible to maintain backwards compatibility with it.
s/eval/pack and unpack/
ShimmerFairy flussence: in case you're interested, here's my incomplete idea on binary grammars I pasted a while back: gist.github.com/ShimmerFairy/72a25...9aab077158
(could be done as a slang, perhaps)
flussence
.oO( pack is just a roundabout way of eval'ing machine code into boxed datatypes... )
 01:37
dalek kudo/nom: 9a01b4b | TimToady++ | src/Perl6/Actions.nqp:
make sure >>[=]>> and X= want their base ops
01:37 mohae left
Juerd I actively used pack and unpack, wrote a wrapper to deal with *some* of the missing functionality, got a great feel for how hard it is to parse those templates. github.com/Juerd/p6-mqtt/blob/mast.../MyPack.pm 01:37
It would be good if this were explicitly marked as experimental 01:38
01:38 mohae joined
Juerd Note that currently, they simply *ignore* unsupported things in templates 01:38
If they're not marked as experimental, it should at least croak on unsupported stuff so there's some unicode left for future implementations. 01:39
ShimmerFairy Juerd: yeah, which is why I'd like to at least hide it behind "experimental", so you know that it's fragile to be using it.
Juerd Could not agree more strongly.
TimToady please rakudobug the chaining issue; I think I should work on making 'while' work better instead when a return value is wanted 01:40
that is, I should work on #126005 now 01:41
and if anyone here feels like trying to narrow down #123272, or install some instrumentation by which we might narrow it down, by all means feel free 01:43
otherwise we're likely to have to distribute with that unfixed
another approach would be to figure out how to detect when it goes wrong, and compensate somehow, but that's if we can't find the actual bug soon 01:44
skids Uhm isn't that one fixed?
TimToady not resolved in RT 01:45
skids Well, I added tests, and if run_alt has decided to shift to other places, they might be passing now.
TimToady but anyway, please be aware that last-minute bikeshedding is a little counterproductive to getting the release out 01:46
skids I'll run a spectest and see.
01:46 uruwi left
TimToady the RT seems to indicate the test for it is fudged? 01:46
01:46 administ1ator joined
TimToady anyway, Not My Bug Right Now 01:47
Juerd Camelia will be your bug right now :)
grondilu J-2 huh? is everything going well for the Xmas release?
cxreg seems sometimes a panda install fails with "Cannot unbox a type object" and then works on retry. known behavior?
01:47 administ1ator left 01:48 raiph left 01:50 uruwi joined, muraiki joined 01:52 mohae left 01:54 mohae joined
flussence ShimmerFairy: that text sounds interesting. It doesn't seem to have any way of handling variable things besides simple null-terminated strings (e.g. UTF-8, p5's «pack 'w*'»), but I'm not going to suggest any syntax for that cause I'm no designer :) 01:54
ShimmerFairy flussence: yeah, it ends right where I was going to write about dealing with specific byte positions (checking them and such), but back when I posted the thing as-is. So it's certainly not a complete, much less final, design :) 01:55
skids RT #123272 tests are currently not fudged and passing, adding note to that effect to RT. 01:56
flussence (now that I've looked at what w* does, it sounds like I really wanted exactly that for writing that net code the other day...)
01:58 mohae left 02:00 raiph joined, pnu left 02:04 pnu joined
flussence is reading masak++'s advent post and sees WP mutilated the HTML *again*... (infix:{missing-thing}) 02:06
02:06 goblinJRT joined, goblinJRT left
skids is happy to see the loop/while return value thing being worked on pre-xmas, btw 02:07
It was perhaps my biggest remaining concern.
ShimmerFairy flussence: and just so you know, my idea was pretty regex-like, so e.g. <u8>+ would theoretically capture variable amounts of unsigned bytes. Problem is, in binary grammars a lot of things are pretty much different kinds of /./, at least how I've thought of it so far :) 02:10
flussence well you've put more thought into it than I have, so I'll let you be the expert :)
Juerd ShimmerFairy: I'm thinking along the lines of signatures, much like NCI 02:13
ShimmerFairy: They can be nested, and I can imagine that with some creativity a syntax could be invented for repetition 02:14
ShimmerFairy sure, that's why I went along the regex route when I wrote that, since those do well with repetition and nesting :) 02:15
Juerd Yes, but regexes don't do well with value types, and that's where signatures shine
Regexes already have repetition, signatures already have types. Argh :)
02:15 vendethiel left
Juerd Note that (un)pack also needs typed repetition. It's not enough to know that something has a length of *, it also matters that the * itself is encoded as e.g. a uint32 02:16
02:17 raiph left
ShimmerFairy Juerd: yeah, that's how I handled the idea of strings with a length beforehand; specify the type of the length prefix, and then the type of the string's characters. 02:18
Juerd Well, strings are relatively easy
But you may need to encode the length of a group of things, e.g. n(n/a* n n x z*) 02:19
er, with a / after the first n
ShimmerFairy there's also the <,u8 a b c> syntax I dreamt up to handle a field of three bytes (for example), though it didn't come designed to handle different types in one construct. 02:21
02:24 psy_ left
Juerd I have a feeling that much of the repetition syntax isn't needed, by the way 02:25
02:25 lichtkind left
Juerd There are length prefixes and there's repetition. But would it be bad to lose the n4 syntax as a shortcut for nnnn? 02:25
02:26 FROGGS_ joined
Juerd Length suffixes are usually *, and used for strings 02:26
ShimmerFairy Juerd: yeah, <,u8 ...> for example is a case of me going "hey, I get annoyed with writing info1 = stuff[0x06]; info2 = stuff[0x07]; in C++, so how's about shortening that?"
Juerd: I'm not sure what you're referring to with n4 and stuff, though
Juerd C4 in Perl 5 pack is C C C C 02:27
In most cases that I've been using pack or unpack, I've never needed those at all
And I have the feeling that string encoding like hex and binary aren't in the right place in pack anyway
You tend not to use those when you're dealing with actual binary data 02:28
ShimmerFairy yeah, I don't see the worth in that, if needed it should be something like C**4 (that is, like in regexes)
Juerd When dealing with binary data, you usually have a template that describes a protocol or file format
And that's typically struct-like, except it has length prefixes
02:29 Sqirrel left
ShimmerFairy A lot of my rough ideas were designed precisely for the fact that binary, unlike textual, data is more position- and size-oriented than it is "syntax"-oriented :) 02:29
02:29 FROGGS left
Juerd Signature literals could really do this, e.g. :(uint32 = :(int16, uint32 = Blob, uint32 = Str)) 02:30
Where I'm stealing default assignments to indicate length prefix types
So uint32 = Blob is like n/a* 02:31
Er, n/a
Missing from signatures (and types) are things like endianness, but I have the feeling that most of what's needed is already there
02:33 cognominal joined
ShimmerFairy Yeah, the endianness is interesting. My thoughts are to go with written order (i.e. big endian, if you wrote it down that way), and require special constructs to "mangle" written order as needed. 02:33
Juerd Maybe it can even do Str is encoded("latin1") and uint32 is long-endian 02:34
I wouldn't mind a verbose template language, because I keep ending up using perldoc -f pack now anyway
ShimmerFairy yes, I definitely don't want the cryptic mess that is pack's syntax :) 02:35
Juerd Building on the type system may make it easier to introduce your own (number) encodings
For MQTT I needed to support a variable width length prefix type 02:36
It's just a number, though, and if I had my own type I could write my own blobification routine
m: my uint32 $foo = 42; say $foo.Blob
camelia rakudo-moar 9a01b4: OUTPUT«Method 'Blob' not found for invocant of class 'Int'␤ in block <unit> at /tmp/mUL10SCY5U:1␤␤»
ShimmerFairy true, though for certain number (and otherwise) encodings they could possibly want to handle the data in a more regex-y than data-type-y way.
Certainly something to work out in an external module, for sure :) 02:37
Juerd I can imagine that .Blob should just do a single-item pack.
Or maybe it's .encode instead of .Blob 02:39
After all, "32 bit signed long endian" is just an encoding. Not for strings, but for numbers. 02:40
02:40 labster left, raiph joined
Juerd Maybe even, .unpack should instead be part of Blob.decode 02:40
Or maybe I shouldn't write braindumps on IRC so close to Christmas. Not productive towards a release. 02:41
In any case It's almost 4am here and I'm going to bed. Good night!
02:42 vendethiel joined 02:44 ilbot3 left 02:47 ilbot3 joined
Juerd And then suddenly I realise I said "long-endian". Hah. Should have been "big-endian" of course :) 02:53
BenGoldberg . o O (my $u = Unpacker.new($some-blob, $starting-offset, $auto-advance, $endianess); my $i = $u.uint32; my $l = $u.sint64; ... ) 02:55
02:59 kaare_ joined 03:05 vendethiel left, rickbike joined 03:06 labster joined 03:09 Ben_Goldberg joined 03:11 kanishka joined 03:12 BenGoldberg left 03:14 mohae__ joined 03:15 Ben_Goldberg left 03:16 Ben_Goldberg joined 03:17 mohae__ is now known as mohae 03:19 mohae left 03:20 noganex joined 03:21 mohae joined
AlexDaniel “When selecting a text with your mouse, double-click on the first word, hold down the mouse on the second click and then select your text. It will now select text by words, not characters.” 03:23
03:24 noganex_ left
TimToady what mouse? 03:24
03:24 BenGoldberg_ joined
AlexDaniel oh wow… how come I did not know this. Not that it is extremely useful since most of the stuff I do is keyboard-only, but hey! 03:24
TimToady just has to figure out now how to transform 'while ++$ < 3 { dostuff }' into 'for Nil xx * -> $ { last unless ++$ < 3; dostuff }' using nothing but QAST trees... 03:27
03:27 Ben_Goldberg left
TimToady just a Small Matter Of Cargoculting 03:27
diakopter TimToady: why do you need the for Nil 03:30
TimToady either that, or figure out exactly where 'while' discards its return values...
which I've tried to do before
skids Are while/loop results going to be eager?
TimToady depends on context, but statementlist-level loops are always in sink context 03:31
diakopter is the for Nil edition a lot faster?
TimToady probably not
getting while to keep its return values long enough to test UNDO, or even longer, to work in an I-want-the-list context, is tricksy 03:32
but if you have a good idea how to do that, feel free 03:33
grondilu while ++$ < 3 {...} is a weird idiom. Isn't it simpler to do for ^3 {...}?
TimToady was looking to transform it to a for loop just to get the semantics right
Juerd Okay, I couldn't sleep. Here's my proposal for an unpack template language: gist.github.com/Juerd/ae574b87d40a66649692
TimToady the point is that while is misbehaving wrt for
Juerd ShimmerFairy, BenGoldberg_: ^^
TimToady so just telling people to write 'for' instead doesn't fix 'while' 03:34
grondilu ok, I must have missed the context
Juerd Is perl6-language still active or does everything go on IRC now?
TimToady this is about fixing rt.perl.org/Ticket/Display.html?id=126005
but if we fix that right then probably while loops will return values in a context where the values are wanted too, without much extra effort 03:35
grondilu did not even know UNDO was a thing 03:36
diakopter did not even
ShimmerFairy Juerd: looks good at first glance; I'll give it a closer read later :)
TimToady and I'm leary of futzing with the implementation of while, which is probably pretty well tuned for working in contexts where you merely want the side effects
Juerd ShimmerFairy: Thanks. Please let me know whatever feedback you may have :) 03:37
TimToady hence, the remap to 'for' approach
grondilu guess he'll have to read S04 again someday
Juerd For now, good night//* #perl6
grondilu *guesses
TimToady o/
the for Nil xx * {} is just a proxy for loop {}, but that's just a while in disguise, which doesn't help 03:39
m: while ++$ < 3 { UNDO say "undo"; say "done" } 03:40
camelia rakudo-moar 9a01b4: OUTPUT«done␤undo␤done␤undo␤»
03:40 cpage_ left
TimToady m: for Nil xx * -> $ { last unless ++$ < 3; { UNDO say "undo"; say "done" } } 03:41
camelia rakudo-moar 9a01b4: OUTPUT«done␤done␤»
TimToady you see that map (well, really the p6for variant) is giving the desired semantics, but it also allows us to return stuff from the loop if that is desired
03:42 secwang joined
TimToady and I already know exactly where we decide whether the loops values are wanted, so it's just a matter of doing the substitution 03:42
easy, peasy, except cargo culting QAST is never quite that...
if it were just the return values that were wanted, a gather/take would work, but that wouldn't help with KEEP/UNDO, since take bypasses the return 03:44
diakopter TimToady: well, you can create a QAST dump of the structure template you want
TimToady so it has to be some sort of map variant, methinketh
sure, already have that, but there are...indirections...
you have to generate new blocks and hide them from the mainline 03:45
so it needs to perform the same sequence of closure-making calls that the 'for' loop does
it's not all just QAST::Mumble.new 03:46
that's really the only hard part...
(I hope)
diakopter if only you had macros 03:47
TimToady I had hoped to be able to just write a macro, but that didn't pan out
heh
great minds...
the parser got very confused for some reason 03:48
well, i was trying it with a map, maybe it wouldn't get so confused for a 'for', but I think I'm not going to try to depend on macros
after all, they're experimental :) 03:49
and I'm pretty sure that if I just mutter the right incantations, I can do the tranformation
03:49 BenGoldberg_ left 03:50 BenGoldberg joined, yqt left
TimToady m: my $x; say (while ++$x < 3 { UNDO say "undo"; say "done"; $x }) 03:50
camelia rakudo-moar 9a01b4: OUTPUT«done␤undo␤done␤undo␤3␤»
TimToady just to be clear, trying to make this work too
m: my $x; say (for Nil xx * -> $ { last unless ++$x < 3; { UNDO say "undo"; say "done"; $x } }) 03:51
camelia rakudo-moar 9a01b4: OUTPUT«done␤done␤(3 3)␤»
TimToady so you see it fixes this too
well, I suppose I could also fake a macro by calling the grammar incestuously 03:54
that would be...wicked...but might be the easiest way to get the semantics right before xmas
03:55 BenGoldberg left 03:56 BenGoldberg joined 03:57 Calibellus joined
TimToady testing...I see at least that nothing in the setting calls a while in non-sink context, which is relieving 03:58
skids A quick look around some ecosystem showed fewer while/loop than I expected. I only found one as last statement of a sub. 04:02
04:03 molaf joined
skids But there could be authors with styles that are exceptionally affected. 04:03
04:06 CQ2 joined 04:08 CQ left, CQ2 is now known as CQ
TimToady nobody will be using something that doesn't work yet, I suspect 04:08
skids wonders how "gather { while { ... } }" pans out 04:09
04:11 MadcapJake left
TimToady doesn't work for UNDO 04:13
skids Oh I meant whether the while would be sunk, not as a solution.
TimToady gather always sinks 04:14
m: gather 43
camelia rakudo-moar 9a01b4: OUTPUT«WARNINGS for /tmp/awrJQ2sr45:␤Useless use of constant integer 43 in sink context (line 1)␤»
skids Just used to seeing "gather while"
TimToady which is fine 04:15
04:15 davercc` left 04:17 kid51 joined 04:25 cpage_ joined 04:27 secwang left 04:29 muraiki left 04:32 stmuk_ joined 04:38 molaf left
[Coke] So, what's the plan for making a roast release? Last I saw jnthn had a thing that was plan-shaped. 04:39
Is this something that we can hope to accomplish in a day? 04:40
skids mostly the possibly-affected loop/while I am seeing in the ecosystem are runloops in ".start" methods or such. Which should probably be fixed but are likely to be called in sink context. 04:45
adu TimToady: hey 04:50
04:51 Quantum joined 04:52 Quantum left 04:54 BenGoldberg left 04:57 BenGoldberg joined 05:00 Ben_Goldberg joined, kid51 left 05:01 Calibellus left 05:02 BenGoldberg left 05:07 rickbike left 05:09 raiph left 05:10 Actualeyes left
ugexe m: class X::Y { }; my $foo = ::("X::Y"); say $foo.gist; # why not (X::Y)? 05:14
camelia rakudo-moar 9a01b4: OUTPUT«(Y)␤»
TimToady well, I'll be jiggerd, it worked 05:17
because "gist" means just enought suggest something? 05:18
m: class X::Y { }; my $foo = ::("X::Y"); say $foo.perl;
camelia rakudo-moar 9a01b4: OUTPUT«X::Y␤»
PerlJam For those people that like Perl 5's Test::Class, have a look at github.com/perlpilot/p6-Test-Class
05:18 Ben_Goldberg left
PerlJam .zZ 05:19
AlexDaniel I am getting errors about EVAL during build-panda, any solution?
[Coke] (perlhist website) whois will give you an email address and a familiar looking name. 05:20
TimToady > p6 'my $x; say (while ++$x < 3 { UNDO say "undo"; say "done"; +$x })' 05:21
done
(1 2)
now I just need to make sure UNDO puts the loop into non-sink context
skids TimToday++ 05:22
ugexe i know .perl gives (X::Y), but im wondering why (Y) is considered more human readable
TimToady in context, you usually know what Y is 05:23
m: say True 05:24
camelia rakudo-moar 9a01b4: OUTPUT«True␤»
TimToady m: say True.perl
camelia rakudo-moar 9a01b4: OUTPUT«Bool::True␤»
TimToady do you really want it to print Bool::True everywhere?
AlexDaniel the problem is in this file: github.com/rakudo/rakudo/blob/nom/...t.pm6#L197 05:25
hmm 05:26
TimToady you have an old precompile of something, and probably need to blow away install/share/perl6 05:28
skids gist.github.com/skids/c4113575bb89625f4bf3 05:29
AlexDaniel skids: hmm why not grep the whole ecosystem? 05:30
05:31 Actualeyes joined 05:32 secwang joined
skids AlexDaniel: it wasn't as simple as "grep", I had to go look at surrounding code. 05:35
AlexDaniel skids: ah
05:37 sammers left 05:39 ellybelly_ left, ellybelly left, regreg joined 05:44 secwang left 05:46 labster left 05:50 sammers joined 05:54 adu left 05:57 gauss joined 06:02 labster joined
AlexDaniel hm. So let's say I do this: .say for run(‘curl’, ‘raw.githubusercontent.com/perl6/ec...ETA.list’, :out).out.lines.race 06:04
I get a segfault about half of the times I run it 06:05
is it a known issue? Or can anybody confirm it?
06:07 adhoc left
awwaiid llfourn: I fixed/pushed a fix for Pod::To::HTML 06:09
(re-enabled the test)
llfourn awwaiid: sweet thanks. Sorry about that :(
awwaiid: it seems that my pod bug fix perl6 was being tested rather than master which *was* passing the tests. Though I thought they both passed. 06:13
awwaiid no problem. Nothing like a failing test to force a fix :) . I kinda think this is working around an upstream issue though -- the meta on the link switches from [] to [""] -- but that data is provided in the $=pod
llfourn awwaiid: yep and my yet to be applied patch fixes that empty string being there :) 06:15
awwaiid ah! where is that at (for educational purposes)? 06:16
what does "post-curli" mean (looking at recent doc commit)?
llfourn awwaiid: yes give me 10mins and I'll make my PR 06:17
post-curli means post precompilation
llfourn can't remember what it stands for
awwaiid there was a recent rakudo change reguarding that?
oh, caching compiled bits? 06:18
llfourn I have no idea. It seems it might have been broken for ages (the cli doc cmd). 06:19
awwaiid in textual bits of the docs, "Perl 6" is used a lot more than "Perl6" -- on purpose, ya? 06:21
llfourn yes Perl 6 is the name of Perl 6 :)
m: say "$*PERL"
camelia rakudo-moar 9a01b4: OUTPUT«Perl 6␤»
llfourn see even Perl 6 says so! 06:22
awwaiid good, I'll use s/Perl6/Perl 6/ as my commit to force a build that will hopefully pull in the fixed Pod::To::HTML :)
llfourn sounds good :)
dalek c: 8a134ae | (Brock Wilcox)++ | doc/Language/ (3 files):
Mass change Perl6 -> Perl 6
 06:24
awwaiid hmm. one of those is wrong. oh well
labster Sometimes it's good to use the non-breaking space in Perl 6, too. 06:25
llfourn mmm good point
labster m: "\xa0".uniname 06:27
camelia ( no output )
labster m: "\xa0".uniname.say
camelia rakudo-moar 9a01b4: OUTPUT«NO-BREAK SPACE␤»
dalek c: 07a3825 | (Brock Wilcox)++ | doc/Language/5to6-nutshell.pod:
No space in reference to an actual module name
06:27 skids left
awwaiid There a POD6 for that? 06:27
llfourn awwaiid: I think you can do E<..> for unicode
E<x0a> (maybe?) 06:28
s/0a/a0/
awwaiid Looks like doc.perl6.org might still not build until someone tries to install Pod::To::HTML again 06:29
06:29 kanishka left
llfourn it looks like it has also broken: design.perl6.org/S26.html # pod design docs 06:30
llfourn thinks it is somewhat ironic 06:31
awwaiid llfourn: maybe there is something less unslightly, like a S<Perl 6> 06:35
yeah, I was just tring to reads S26.html also :)
llfourn wonders if S<> is implemented 06:36
I don't think it is :\ 06:37
awwaiid My panda is mad at trying to install the new Pod::To::HTML because it already has version "*" installed :( 06:38
alright. well, I give up for tonight -- will bang the fix-doc.perl6.org-drum a bit more tomorrow :) 06:39
gnight!
llfourn hmm it should probably be versioned
awwaiid: gnight :)
awwaiid++ thanks for fixing my goofs
06:39 yeahnoob left
awwaiid yes... should probably be a version other than "*" :) 06:39
06:40 geraud left 06:41 dayangkun joined, dayangkun left 06:42 dayangkun joined 06:43 lustlife left 06:44 yeahnoob joined
llfourn .tell moritz I merged a PR with a failing test in Pod::To::HTML -- the fallout seems to be that design.perl6.org/S26.html is gone and docs aint recompiling -- Yikes! Sorry :( 06:46
yoleaux llfourn: I'll pass your message to moritz.
06:47 lustlife joined 06:57 tmw1 joined 07:00 regreg left
thomax will p6.c released tomorrow or the day after?? 07:02
AlexDaniel thomax: hmm it also depends on the timezone I think 07:03
thomax: but I'm curious, what would be the difference? :)
thomax the difference is a big exclamation mark behind p6. 07:06
i hope there will be a wave of reviews 07:07
07:10 dayangkun left 07:11 llfourn left
AlexDaniel “use MONKEY-SEE-NO-EVAL to override, but only if you're VERY sure your data contains no injection attacks” – ehh… What a nice warning. Well, indeed, another solution to qqx problem is to introduce MONKEY-SEE-NO-QQX or something like that. Shell injection is about equal to eval injections, with the only difference that we usually say “everyone forgets to escape their stuff when passing it to shell but it's all their fault, they shou 07:11
have known better” and thus no protection…
07:14 hankache joined
AlexDaniel By the way, the startup time of my program that is using a bunch of modules has been reduced by half in the last few days. (Some angel)++ 07:14
which is about ⅓ of what it was a month ago
[Tux] test 50000 24.682 24.565 07:15
test-t 50000 14.302 14.185
csv-parser 50000 52.309 52.192
hankache morning #perl6
yoleaux 22 Dec 2015 17:52Z <TimToady> hankache: see irclog.perlgeek.de/perl6/2015-12-22#i_11759252 for how to put spaces around the replacement
hankache masak++ #advent post 07:17
.tell TimToady thanks boss 07:18
yoleaux hankache: I'll pass your message to TimToady.
hankache Tis the season to be jolly, Fa la la la la, la la la la 07:24
07:24 jczeus joined, gauss left, CIAvash joined 07:25 [Sno] left
moritz \o 07:28
yoleaux 06:46Z <llfourn> moritz: I merged a PR with a failing test in Pod::To::HTML -- the fallout seems to be that design.perl6.org/S26.html is gone and docs aint recompiling -- Yikes! Sorry :(
07:29 nwc10 joined
nwc10 good *, #perl6 07:30
dalek : 5657282 | moritz++ | util/update-design.perl6.org.sh:
design.perl6.org rebuild: avoid clobbering .html files on failure
 07:31
moritz .tell llfourn I've installed a new Pod::To::HTML, and made the build process more robust as to avoid emtpy S26.html in future (github.com/perl6/mu/commit/5657282404) 07:32
yoleaux moritz: I'll pass your message to llfourn.
hankache hello moritz nwc10
moritz good morning hankache, nwc10, *
nine Good morning! 07:33
nwc10 days are getting longer (in some parts of the world)
all-nighter hacking sessions are getting shorter :-)
hankache morning nine 07:34
07:35 vv3 joined 07:36 _nadim joined
[Tux] m: my @x = ^10; for ^@x.elems -> $idx { say $idx } 07:36
camelia rakudo-moar 9a01b4: OUTPUT«0␤1␤2␤3␤4␤5␤6␤7␤8␤9␤»
dalek c: 2cc8c2f | moritz++ | doc/Language/ (4 files):
SSLify some links
TimToady moritz++ SSL++ 07:38
yoleaux 07:18Z <hankache> TimToady: thanks boss
07:44 RabidGravy joined 07:46 pierre-vigier joined 07:47 pierre-vigier left
[Tux] All tests successful. 07:49
Files=27, Tests=21962, 35 wallclock secs ( 3.06 usr 0.17 sys + 89.73 cusr 1.83 csys = 94.79 CPU)
but I still want a :don't-be-smart option for \r\n (or have that implicit on :!chomp) 07:50
_nadim God morning all 07:52
hahainternet do we have anything equivalent to f#'s forward pipe operator?
i can't imagine it'd be common, but a friend is asking
07:53 pierre-vigier joined 07:54 sjn_ is now known as sjn 07:55 llfourn joined 07:57 [Sno] joined 08:01 adhoc joined
nine hahainternet: what does that operator do? 08:05
hahainternet nine: it passes the parameter on the left, to the function on the right
i don't really quite get why they have it tbqh
a random example: [1..10] |> List.map add1
(and yes i know about », but i couldn't find any equivalent to that specific operator) 08:06
hankache www.kevinberridge.com/2012/12/neat-...rward.html 08:09
hahainternet they're neat operators, but i have to say i don't quite 'get it'
i guess it's for the super functional nerds to describe a program on a single line and then get annoyed when you ask them to explain it ;)
hankache i think chaining in Perl6 does the same thing no? 08:10
TimToady m: 1..10 => say
camelia rakudo-moar 9a01b4: OUTPUT«5===SORRY!5===␤Argument to "say" seems to be malformed␤at /tmp/cE__j9wVCk:1␤------> 031..10 => say7⏏5<EOL>␤Other potential difficulties:␤ Unsupported use of bare "say"; in Perl 6 please use .say if you meant $_, or use an explicit in…»
TimToady m: 1..10 ==> say
camelia rakudo-moar 9a01b4: OUTPUT«5===SORRY!5===␤Argument to "say" seems to be malformed␤at /tmp/4g8RfTSZ6S:1␤------> 031..10 ==> say7⏏5<EOL>␤Other potential difficulties:␤ Unsupported use of bare "say"; in Perl 6 please use .say if you meant $_, or use an explicit i…»
TimToady m: 1..10 ==> say()
camelia rakudo-moar 9a01b4: OUTPUT«1..10␤»
TimToady something like that?
hahainternet nice, so that's the operator i was missing
hahainternet goes back to operators doc page
i did read through it first! i swear! 08:11
CIAvash perl6advent.wordpress.com/2010/12/...operators/
hahainternet ok no need to show me up further ;)
interestingly, it only appears once on the operators page, as a 'Sequencer'
but is not documented fully
CIAvash design.perl6.org/S03.html#Feed_operators 08:13
hahainternet thanks for your help TimToady; nine; CIAvash
CIAvash design.perl6.org/S06.html#Feed_operators
08:14 darutoko joined, eone joined
TimToady m: given 1,2,3 { say |$_ } 08:17
camelia rakudo-moar 9a01b4: OUTPUT«123␤»
TimToady another way
hahainternet interesting, i guess i'll have to look that up 08:18
didn't realise you could flatten like that
RabidGravy early morning dumb question: is it safe to assume that the "size" of a Str passed to a char * in a native sub is Str.encode.bytes? libgdbm wants to know 08:22
08:23 timotimo left
hahainternet RabidGravy: to my ignorant eyes, that seems logical 08:24
RabidGravy I think it's try and delete the coredumps afterwards ;-)
08:26 timotimo joined, vv3 left, vv3 joined 08:27 vv3 left 08:28 vv3 joined, vv3 left, vv3 joined, vv3 left 08:29 xinming left, vv3 joined 08:30 shmibs left 08:32 rafl joined 08:33 pierre-vigier left 08:34 vv3 left 08:35 vv3 joined, vv3 left 08:36 vv3 joined, vv3 left, vv3 joined, vv3 left 08:37 shmibs joined, vv3 joined
dalek kudo/nom: 3dec5c8 | TimToady++ | src/core/Any-iterable-methods.pm:
useless use of 1
 08:42
kudo/nom: 0258819 | TimToady++ | src/Perl6/ (3 files):
while/until loops can now return values

A loop that has KEEP or UNDO, or that is used in a spot indicating that a return value is expected (such as in parens), will now be translated to a map (well, a for loop) that knows how to deal with blocks returning values correctly. In the case that a loop has a KEEP/UNDO but is not in a context wanting the return value, it still does the transformation to all KEEP/UNDO to have a value to test, but then sinks the results of the loop as a whole.
Fixes #126005
ast: 2f126a3 | TimToady++ | S04-statements/while.t:
tests for returns from while/until, RT #126005
08:42 vv3 left, shmibs left 08:43 vv3 joined, vv3 left
dalek kudo/nom: 0734842 | (Stefan Seifert)++ | src/core/CompUnit/Repository/Installation.pm:
Include dist-id in names of installed files

The hash used for generating unique file names now contains:
  * the relative path of the source file including file name
  * the dist's id (name, version, auth, api)
  * the repo chain at time of installation
Including the dist's id should fix name collissions when two distros contain a lib/Foo/Bar.pm6.
As a side effect, the precompiled file will now have the same name as the source file which could make debugging easier.		 08:44
08:44 vv3 joined, vv3 left 08:45 vv3 joined, vv3 left, abraxxa joined, vv3 joined
abraxxa hi! 08:46
TimToady hullo
abraxxa everything going well for the Christmas release? 08:47
hankache hALO 08:48
TimToady we're down to one ticket
CIAvash TimToady: /(\w)<{$0.Str.succ}><{$0.Str.succ.succ}>/ is eating a lot of memory in:
abraxxa i'm trying to get DBDish::Oracle pod finished and merged into DBIish
hankache ho ho ho ho ho
CIAvash $_ = "abcdefgh"; say ++$_ until /(\w)<next_char={$0.Str.succ}><{$<next_char>.Str.succ}>/ and !/i|o|l/ and /(\w)$0.*(\w)$1/;
abraxxa awesome!
RabidGravy is that being saved for midnight tomorrow?
CIAvash Am I doing something wrong?
TimToady RabidGravy: I have to write an advent posting tomorrow, so if someone else doesn't fix it, it probably won't get fixed 08:49
RabidGravy what's the ticket?
08:49 baest_ joined
abraxxa where can I find docs for Perl 6 pod? 08:50
stmuk rt.perl.org/Ticket/Display.html?id=123766
08:50 vv3 left, firstdayonthejob joined
abraxxa i'm having a hard time with S26 synopses 08:51
TimToady CIAvash: alas, I cannot really help you tonight, it's already almost 1am here...
08:51 baest left
abraxxa i there a module with known good perl 6 pod which I can use as example? 08:51
nwc10 TimToady++ # fixing the last blocker bug 08:52
08:52 vv3 joined, vv3 left
stmuk rt.perl.org/Ticket/Display.html?id=123272 has been closed 08:52
08:52 dakkar joined
TimToady ⟶ 💤 08:52
CIAvash TimToady: OK 08:53
08:53 vv3 joined, vv3 left
TimToady oh, it's fixed? \o/ 08:53
nwc10 er, OK, closing it in RT :-)
stmuk :D 08:54
08:54 vv3 joined
nwc10 technically it does seem that it should be "tests needed" 08:54
TimToady well, whatever, we're sittin' pretty now
abraxxa Advent day 10 is pod, will take a look at that
nwc10 or atcually, "There are no tests there for with or whenever, but those did not appear to misbehave when I tried them on the command line. Feel free to add more tests for those." seems to be "more tests please"
TimToady now if only we still didn't have tests that hang when run in parallel... 08:55
nwc10 there's still one day before Christmas :-)
TimToady I'm not sure those ever misbehaved, so didn't think it really mattered for this bug
but it wouldn't hurt to have tests for 'em, is all
RabidGravy I never like these things that "just go away"
08:57 znpy left
TimToady if anyone is at loose ends for something to think about, we can figure out how to freeze a set of tests for 6.c without fudges 08:57
Poor Man's Version: we save a fudged version, and 6.c has to keep passing that 08:58
thomax wait, will this be the christmas eve release or the christmas day release?
08:58 rindolf joined, vv3 left
TimToady I suspect we'll generate a release on the 24th and someone will officially announce it on the 25th 08:59
but of course this depends on availability of voluntolds
abraxxa design.perl6.org/S26.html returns forbidden
RabidGravy that's what I was going to suggest, doesn't there need to be at least two fudged versions though - one for rakudo and one for jvm? (and also aren't there other fudges for architecture and OS and so forth?)
thomax cool. thats a big thing!
09:00 znpy joined
RabidGravy goes back to making coredumps 09:00
TimToady 6.c is whatever the moarvm fudges produce, so that also has to be taken into account. jvm doesn't get to use a different standard for 6.c, it'll just take a while to catch up, as we anticipated all year 09:01
RabidGravy right
thomax do you expect that the js implementation (of not only nqp) will be fast enough to get things done? 09:02
TimToady it will be fast enough to get some things done
it will not be fast enough to get other things done, likely, without further optimization 09:03
but same is true for all the other backends
just to differing degrees on differing work
thomax TimToady: yeah, my question is very unspecific..
hankache m: @array ==> sort() ==> map { $_ ** 2} ==> my @final-array; say @final-array; 09:04
camelia rakudo-moar 073484: OUTPUT«5===SORRY!5=== Error while compiling /tmp/p1H6_dybOf␤Variable '@array' is not declared. Did you mean any of these?␤ Array␤ array␤␤at /tmp/p1H6_dybOf:1␤------> 3<BOL>7⏏5@array ==> sort() ==> map { $_ ** 2} ==>␤»
09:04 vv3 joined
hankache m: my @array = <7 8 9 0 1 2 4 3 5 6>; @array ==> sort() ==> map { $_ ** 2} ==> my @final-array; say @final-array; 09:04
camelia rakudo-moar 073484: OUTPUT«[0 1 4 9 16 25 36 49 64 81]␤»
thomax i just think, it would be cool to use p6 in the browser environment to get a nice UI framework done 09:05
hankache hein?? why is this ^^ not working on my local rakudo :(
rindolf TimToady: what are "voluntolds"? en.wiktionary.org/wiki/Special:Sea...voluntolds
yoleaux 22 Dec 2015 22:34Z <nine> rindolf: your share/perl6 really doesn't contain anything but panda's modules and the CORE dist. And lots of pre-curli files. I really wonder where panda installed Shell::Command to
TimToady it's a recent neologism, "I was voluntold to do this." 09:06
nwc10 rindolf: I'm assuming that they're the plural of en.wiktionary.org/wiki/voluntold
09:06 Deepak_ joined
rindolf nwc10: ah, I see. 09:06
nwc10 (um, which is only given as a verb. but I think works as a noun) 09:07
09:07 cpage_ left, vv3 left
nwc10 I had not met it before. It amuses me. 09:07
09:07 cpage left
rindolf TimToady: thanks! 09:07
RabidGravy "one who is volunteered"
TimToady m: say (+$_ when *.is-prime while ++$_ < 100) 09:08
camelia rakudo-moar 073484: OUTPUT«(2 3 5 7 11 13 17 19 23 29 31 37 41 43 47 53 59 61 67 71 73 79 83 89 97)␤»
TimToady m: say ($_ when *.is-prime while ++$_ < 100)
camelia rakudo-moar 073484: OUTPUT«(100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100)␤»
TimToady maybe wants an automatic decont in there somewhere
timotimo o/ 09:09
TimToady but it now really is past 1am
TimToady ⟶ 💤 # This time fershure!
timotimo hah
thomax TimToady: hands of the keyboaard
timotimo the same dance i did last night
RabidGravy 9:09 am
timotimo anyway, BBL :)
09:10 vv3 joined
abraxxa moritz: are you around? 09:10
hankache ca someone try putting this on multiple lines: my @array = <7 8 9 0 1 2 4 3 5 6>; @array ==> sort() ==> map { $_ ** 2} ==> my @final-array; say @final-array;
and see if it works
not working if each ==> is on a new line 09:11
09:12 ChoHag left 09:15 vv3 left, Deepak_ left 09:18 pierre-vigier joined, vv3 joined
abraxxa i've just compiled moar using rakudobrew which defaults to the nom branch and now get an error with DBDish::Oracle: Cannot import symbol size_t from DBDish::Oracle::Native, because it already exists in this lexical scope 09:18
jczeus it fails for when i add the map
for me 09:19
RabidGravy abraxxa, probably yes, the size_t was just added to NativeCall in the last few days IIRC
abraxxa RabidGravy: I define it in DBDish::Oracle::Native; 09:20
constant size_t is export = long;
09:20 Begi joined
hankache jczeus indeed unless you write the map like this: map({ $_ ** 2}) 09:21
abraxxa does this mean I have to remove my definition?
nine abraxxa: seems like you can just remove it. Exporting such a generic name doesn't sound like the best idea anyway.
hankache no idea why though
abraxxa nine: i've taken over the names from the OCI header files
RabidGravy abraxxa, yeah it went in with 61f085ca92ba9221fa38ac753509e5285f88cacb
abraxxa RabidGravy: thanks 09:22
jczeus hankache: it also works when the closing brace is on the same line as the pipe operator
nine abraxxa: but size_t is a very standard C type. It's bound to be used by others, too.
abraxxa nine: got it
hankache my @array = <7 8 9 0 1 2 4 3 5 6>;
@array ==> sort()
==> map({ $_ ** 2})
==> reverse()
==> my @final-array;
say @final-array;
hankache likes the feed operator 09:23
09:23 vv3 left
nine I wonder why exactly we don't precompile scripts. With all the automatic precompilation logic already in place, it should actually be quite manageable. 09:23
09:24 vv3 joined, vv3 left 09:25 vv3 joined, vv3 left
jczeus hankache: that also works: ==> map { $_ ** 2 } \ 09:25
09:25 eone left, vv3 joined, vv3 left
hankache what happening with vv3? joining and quitting 09:25
TEttinger that's not uncommon with a bad connection on IRC 09:26
hankache bot?
ah ok
09:26 vv3 joined
TEttinger might be a snowstorm or windstorm or whatever messing with power lins 09:26
abraxxa and NativeCall doesn't find the lib any more 09:27
hankache jczeus yes 09:28
09:28 pierre-vigier left
nine abraxxa: we had a change there last night. Removing the "lib" in the libname should fix it. 09:31
RabidGravy abraxxa, it may be omit the 'lib' from the front
abraxxa nine: I've guessed that from the debug output
09:31 vv3 left
abraxxa also added the version like Skarsnik did here github.com/Skarsnik/DBIish/commit/...4fca73e964 09:31
09:31 yeahnoob left
abraxxa nine: is that wise? it will break all modules using nativecall 09:32
09:32 vv3 joined, vv3 left
abraxxa the version causes this error: Connect failed with error Type check failed in binding $version; expected Version but got Int 09:32
09:32 nwc10 left
nine abraxxa: better break them before christmas than afterwards. The goal is to abstract away platform conventions. What's liboracle.so.5.3 on Linux may be oracle.dll on Windows or liboracle.5.3.dylib on OS X 09:32
abraxxa i see 09:33
09:33 vv3 joined, vv3 left
RabidGravy the version thing is just use v1.5 or whatever 09:33
abraxxa it's libclntshcore.so.12.1 ;)
jczeus is this intended behavior? i'm looking at s04 to see if that's the case
09:33 pierre-vigier joined
abraxxa RabidGravy: then Skarsnik's commit is wrong too, maybe that's why it hasn't been merged so far 09:33
RabidGravy so "is native('clntshcore', v12.1)"
09:34 vv3 joined, vv3 left
hankache same question as RabidGravy ? 09:34
RabidGravy dunno, I'm going to have to fix eight modules today
abraxxa i'd prefer to not specify the exact version or a minimum version, is that possible?
09:34 vv3 joined
hankache how do you specify the ver now? 09:35
rindolf nine: here? I see Shell__Command in /home/shlomif/Download/unpack/perl/p6/rakudobrew/moar-nom/panda/ext .
abraxxa i now get # Connect failed with error Unknown type 'size_t' used for native call
hankache: v12.1 09:36
09:36 pierre-vigier left
nine abraxxa: does Oracle install a libclntshcore.so.12 symlink? 09:36
abraxxa: just specifying the major version could be enough
abraxxa sorry, wrong name before libclntsh.so.12.1 and yes libclntsh.so
nine rindolf: that's the source where it should be installed from.
abraxxa nine: version 11 should work fine too 09:37
nine rindolf: do you find any trace of it in your ~/.perl6?
abraxxa so if I specify v11 does it search only for .11 or will the symlink be used if it can't be found?
rindolf nine: grep -ri Shell ~/.perl6/ does not return anything. 09:38
09:38 pierre-vigier joined
nine abraxxa: usually there is a libfreetype.so.6.11.2 file and a libfreetype.so.6 symlink pointing to the file and maybe a libfreetype.so symlink. 09:38
09:39 virtualsue joined
nine abraxxa: AFAIK we do not have any automatic fallback in NativeCall 09:39
09:39 vv3 left
abraxxa nine: the Oracle InstantClient doesn't have any symlink, DBD::Oracle (the Perl 5 one) did create one as far as I remember 09:39
nine rindolf: that's...really odd. It should at least be found in some precomp files
AlexDaniel “As a side effect, the precompiled file will now have the same name as the rakudo/nom: source file which could make debugging easier.” – YESSSSS 09:40
if I read it correctly
09:41 xfix joined 09:42 vv3 joined
nine AlexDaniel: you probably didn't 09:43
timotimo hankache: a } at the end of a line gets a ; for free. that's what gives you trouble with the feed operator
nine AlexDaniel: it's still a SHA1 hash. It's just that both the source file and the precomp file will be called 6F2671BB083EFD2E16FD682C7A9D8498E3816ED0 09:44
timotimo i imagine we can special-case a check for that in the grammar that gives you a helpful hint if you're starting a line with a ==>, that perhaps the previous line ended in a } and needs an unspace
AlexDaniel nine: f…
nine AlexDaniel: otherwise my commit message would have been a lot more enthusiastic than "could make it easier" :)
dalek ast: 9d3e3c3 | labster++ | S17-promise/allof.t:
add test for RT #122802: non-invokable object is non-invokable
 09:45
hankache timotimo: ah!
nine AlexDaniel: I do believe that we can make those backtraces more useful. It's just nothing we can squeeze in this side of christmas.
AlexDaniel nine: by the way, although you've probably thought about that, is it possible to use SHA1 hash for directories while still keeping the same filename inside? 09:46
hankache timotimo i wrapped them in parens. what is the unspace?
abraxxa nine: any idea where that size_t error comes from?
--ll-exception didn't help
nine AlexDaniel: the trouble with file names is that not all file systems support Unicode which is the reason why Perl 5 for example doesn't allow Unicode module names.
AlexDaniel: we do give you much more freedom in Perl 6 but have to make sure we can actually store those files. 09:47
abraxxa: not without knowing anything about the code.
09:47 vv3 left
timotimo hankache: ending your line in "\" gives you an unspace 09:47
AlexDaniel nine: I'm not sure that I want to be affected by this if my filesystem does handle it correctly 09:48
nine: perhaps make it opt-in? :)
timotimo unspace lets you put something that looks a lot like spaces in places where spaces are otherwise not allowed
hankache thanks timotimo
timotimo so you can line up some stuff with that, like the parens of a function call
09:48 vv3 joined, vv3 left
timotimo m: say("hi", "how", "are", "you"); say ("hi", "how", "are", "you"); say\ ("hi", "how", "are", "you") 09:49
camelia rakudo-moar 073484: OUTPUT«hihowareyou␤(hi how are you)␤hihowareyou␤»
timotimo you see how the first and second are the same, but not the one in the middle?
abraxxa nine: I've commented out the CATCH in the test file so I see the error: Method 'prepare' not found for invocant of class 'Any'
nine AlexDaniel: if any people had joined the effort, I may have found some time to find a nicer solution. As it is, I will still work on the necessary on the evening before the release.
09:49 vv3 joined, vv3 left
hankache oki 09:50
09:50 vv3 joined, vv3 left
nine AlexDaniel: in other words: what you want or not really doesn't matter. What matters is what you do to make it happen. 09:50
abraxxa nine: ok, that happens because connect doesn't return
AlexDaniel nine: sure
09:50 espadrine joined 09:51 Actualeyes left, vv3 joined
Ulti m: say 8 ?? 09:51
camelia rakudo-moar 073484: OUTPUT«5===SORRY!5=== Error while compiling /tmp/xByXbim1ie␤Confused␤at /tmp/xByXbim1ie:1␤------> 3say 8 ??7⏏5<EOL>␤»
dalek ast: 03e62e2 | lizmat++ | S02-types/range.t:
Add tests for Range.minmax
09:52 pierre-vigier left, cpage_ joined
abraxxa nine: why can size_t be unknown? 09:52
AlexDaniel nine: by the way I've submitted a rakudobug about this hoping to see any movement
nine: though I didn't feel like diving in because I didn't know why it was implemented in first place 09:53
and not that I know now…
nine abraxxa: does the module that uses this type import NativeCall? 09:54
abraxxa nine: yes
jczeus hankache: haven't found anything conclusive in s04, but a line ending with a closing brace apparently terminates the expression. that's why your solution with the parens also worked 09:55
09:56 vv3 left
nine AlexDaniel: all the reasoning is there: gist.github.com/jnthn/47a42b2e86e7e552b2e2 09:56
09:56 mohae left, vv3 joined, vv3 left 09:57 virtualsue left, vv3 joined, vv3 left
hankache jczeus as timotimo said a closing brace will give you a ";" for free. so we should either use an unspace "\" like you proposed or wrapping them in parens 09:57
09:58 pierre-vigier joined, vv3 joined, vv3 left
timotimo jczeus, hankache design.perl6.org/S04.html#Statement...ing_blocks 09:58
hankache: alternatively put the ==> on the end of the previous line
09:59 vv3 joined
abraxxa nine: i have only two occurrences of size_t, both in Oracle::Native.pm6 09:59
10:00 cognominal left
abraxxa for the failing one I pass 0 to the parameter 10:00
10:00 pierre-vigier left
abraxxa nine: github.com/abraxxa/DBIish/blob/mas...ve.pm6#L89 10:01
10:01 cpage joined
AlexDaniel nine: oh nice! Thanks for the link 10:02
10:02 mohae joined 10:03 woolfy joined
jczeus timotimo: i read that bit, but it says only "if an end of statement can occur there" - not sure about that 10:03
10:04 vv3 left
timotimo combine that sentence with the global want for "single-pass parsing" and you'll see why it's the way it is 10:04
we're not going to re-parse the end-of-line as "not supposed to be a ;" when we discover the next line has a feed operator, for example
10:05 vv3 joined, vv3 left, pierre-vigier joined
stmuk dd seems to have lost the variable name 10:05
10:05 vv3 joined, vv3 left
dalek kudo/nom: 8266a1f | timotimo++ | src/ (2 files):
METAOP_ASSIGN shall use .DEFINITE instead of .defined

this will correctly let Failure objects in a variable blow up when the variable is +='d or so, rather than defusing and replacing the Failure object.
The optimization for METAOP_ASSIGN in the optimizer has sadly become a casualty, as it unexplicably leads to a
  "Cannot modify immutable *" error in many places.
That's also why the implementation uses if/else rather than a ?? !!. That leads to the same error.		 10:05
timotimo m: dd my $x = <hi what's up> 10:06
camelia rakudo-moar 073484: OUTPUT«List $var = $("hi", "what's", "up")␤»
timotimo huh, interesting
10:06 vv3 joined, vv3 left
lizmat m: dd :a(42) # stmuk: can use nameds to indicate the name 10:06
camelia rakudo-moar 073484: OUTPUT«Int a = 42␤»
10:07 kanishka joined, vv3 joined
stmuk ah 10:08
10:09 pierre-vigier left
lizmat I wanted to be able to use nameds for dd 10:10
but that broke the positional introspection :-(
10:12 pierre-vigier joined, vv3 left 10:13 hankache left, vv3 joined, vv3 left 10:14 mohae left, vv3 joined, vv3 left
abraxxa nine: interesting that this syntax error didn't cause the module to fail until now: github.com/abraxxa/DBIish/blob/mas...le.pm6#L88 10:14
10:14 mohae joined
abraxxa but this doesn't fix the size_t error ;( 10:15
10:15 vv3 joined, vv3 left 10:16 vv3 joined
abraxxa nine: it's the OCIEnvNlsCreate call that causes the error 10:16
10:18 kjs_ joined
lizmat stmuk: naming the positionals depended on being able to do nqp::p6argvmarray(); 10:19
stmuk: and that apparently only works with a (|) signature
stmuk: and *that* doesn't allow named params
abraxxa nine: it shouldn't come down there because %type_map contains size_t: github.com/rakudo/rakudo/blob/ddb8...l.pm6#L135 10:20
10:21 vv3 left, vv3 joined, vv3 left 10:22 Begi2 joined, vv3 joined, vv3 left 10:23 vv3 joined, vv3 left 10:24 Begi left, vv3 joined 10:26 AlexDaniel left
lizmat ok, I'm going to revert supporting nameds with dd 10:27
10:29 vv3 left 10:30 vv3 joined, vv3 left, iH2O joined
abraxxa m: use NativeCall; my size_t $foo; say $foo.^shortname; 10:30
10:30 vv3 joined, vv3 left
camelia rakudo-moar 8266a1: OUTPUT«Int␤» 10:30
abraxxa haha!
10:31 iH2O left
abraxxa nine: the ^shortname is used for the type_map lookup and Int maps to longlong 10:31
10:31 vv3 joined, vv3 left 10:32 rurban joined, vv3 joined
lizmat nine: confirm that all Inline::Perl5 tests now pass on OS X 10:34
nine: alas, install still fails with: Failed to copy '/Users/liz/Github/panda/.panda-work/1450866723_2/resources/p5helper.so' to '/Users/liz/Github/rakudo.moar/install/share/perl6/site/resources/6A67E44F00229F4678C75C6496AC23F4B5FA022A.so': Failed to copy file: no such file or directory 10:35
nine: feels like some of the new lib name guessing has not been applied yet ?
10:37 vv3 left 10:38 vv3 joined, vv3 left 10:39 vv3 joined, vv3 left
dalek kudo/nom: 9308de3 | lizmat++ | src/core/Any.pm:
No longer look at nameds in dd

It was a nice idea, but it broke the introspection on positionals. And since nqp::p6argvmarray only knows about positionals, and that is needed to introspect (afaics), dropping the nameds support was needed to reinstate positional introspection		 10:39
10:39 vv3 joined, vv3 left
abraxxa am I allowed to push a commit adding a test to nom? 10:40
shows that size_t isn't working
10:40 vv3 joined
lizmat as in the t/ directory? 10:40
sure
if you have a commit bit ? 10:41
abraxxa lizmat: not sure
lizmat otherwise a PR will do fine
abraxxa i don't
forking
RabidGravy right, all the methods added to GDBM::File off out to pub -> farm shop etc, don't go breaking anything ;-) 10:44
Juerd Looking for feedback on my unpack idea: gist.github.com/Juerd/ae574b87d40a66649692
timotimo lizmat: it doesn't work with (|c) signature? p6vmargarray, that is 10:46
argvmarray*
10:47 hankache joined
lizmat nope, any other sig than (|) eats it 10:48
10:48 mohae left
lizmat by the looks of it 10:48
10:48 vv3 left
timotimo hm, OK 10:48
10:49 mohae joined, RabidGravy left, kjs_ left
timotimo really all p6argvmarray does is grab a slurpy positional argument and hllize each entry 10:49
10:49 kjs_ joined
labster m: my $promise = start { fail 42 }; await $promise; say $promise.status #RT #123204 10:50
camelia rakudo-moar 8266a1: OUTPUT«42␤ in block <unit> at /tmp/LB3q0tp6Dx:1␤␤»
lizmat I guess that's what any sig other than (|) also does, hence I can't get at them anymore after that
labster j: my $promise = start { fail 42 }; await $promise; say $promise.status irc://chat.freenode.net:6665/#RT irc://chat.freenode.net:6665/#123204 10:51
10:51 vv3 joined
camelia rakudo-jvm 6c0f93: OUTPUT«5===SORRY!5=== Error while compiling /tmp/0blLkRtO2b␤Two terms in a row␤at /tmp/0blLkRtO2b:1␤------> 032 }; await $promise; say $promise.status7⏏5 irc://chat.freenode.net:6665/#RT irc://␤ expecting any of:␤ infix␤ inf…» 10:51
timotimo yeah
i expect p6argvmarray is an optimization so that no Capture object has to be created?
labster j: my $promise = start { fail 42 }; await $promise; say $promise.status
timotimo if you only need nameds
camelia rakudo-jvm 6c0f93: OUTPUT«Attempt to return outside of any Routine␤ in block <unit> at /tmp/QrAskKZvVG:1␤␤»
timotimo labster: yeah, fail doesn't work if you don't have a sub to fail from 10:52
hankache what is the status of rakudo on jvm ?
timotimo j: my $promise = start(anon sub($_) { fail 42 }); await $promise; say $promise.status
camelia rakudo-jvm 6c0f93: OUTPUT«5===SORRY!5===␤Type 'sub' is not declared. Did you mean 'Sub'?␤at /tmp/WEISKACa6Z:1␤------> 3my $promise = start(anon sub7⏏5($_) { fail 42 }); await $promise; say $␤Malformed anon␤at /tmp/WEISKACa6Z:1␤------> 3my $promise = start(anon…»
10:52 vv3 left
timotimo j: my $promise = start(anon sub ::($_) { fail 42 }); await $promise; say $promise.status 10:53
camelia rakudo-jvm 6c0f93: OUTPUT«===SORRY!===␤Name ::($_) is not compile-time known, and can not serve as a sub declaration␤»
timotimo j: my $promise = start(anon sub ($_) { fail 42 }); await $promise; say $promise.status
camelia rakudo-jvm 6c0f93: OUTPUT«5===SORRY!5=== Error while compiling /tmp/tdK_bFCcz8␤Undeclared routine:␤ start used at line 1. Did you mean 'spurt', 'sqrt', 'sort'?␤␤»
labster Wait, which is correct though, moar or jvm?
timotimo j: my $promise = start anon sub ($_) { fail 42 }; await $promise; say $promise.status
camelia rakudo-jvm 6c0f93: OUTPUT«Kept␤»
timotimo j: my $promise = start anon sub ($_) { fail 42 }; say await $promise; say $promise.status;
camelia rakudo-jvm 6c0f93: OUTPUT«sub ($_) { #`(Sub|93720058) ... }␤Kept␤»
timotimo oh damn
i forgot about that
does Promise.new schedule the promise, too? 10:54
no, it's not about creating code-backed promises
lizmat Promise.new just creates a Promise
nothing special about that
m: Promise.new.result # wait for the heat death
timotimo yeah 10:55
camelia rakudo-moar 9308de: OUTPUT«(timeout)»
timotimo how do i "start" a sub?
nine lizmat: yes, that part is still NYI. On my list for tonight
hankache chaining methods was available in Perl 5?
lizmat ++nine
hankache @array.sort.reverse
or is this new to 6 ? 10:56
lizmat m: sub a { "foo" }; Promise.start( &a ).result.say
camelia rakudo-moar 9308de: OUTPUT«foo␤»
llfourn m: my str $str = ''; $str.WHAT.say # how do I tell if the value contained is native?
yoleaux 07:32Z <moritz> llfourn: I've installed a new Pod::To::HTML, and made the build process more robust as to avoid emtpy S26.html in future (github.com/perl6/mu/commit/5657282404)
camelia rakudo-moar 9308de: OUTPUT«(Str)␤»
lizmat timotimo: ^^^
llfourn moritz++ thanks!
lizmat llfourn: you can't
timotimo ah, that's how!
great
labster Ah, makes sense.
timotimo j: my $promise = Promise.start(anon sub ($_) { fail 42 }); say await $promise; say $promise.status 10:57
lizmat llfourn: natives are not objects, so they have to be upgraded first before you can call anything on them
camelia rakudo-jvm 6c0f93: OUTPUT«Too few positionals passed; expected 1 arguments but got 0␤ in block <unit> at /tmp/np31QXWJ6x:1␤␤»
lizmat llfourn: hence you see Str instead of str
timotimo j: my $promise = Promise.start(anon sub { fail 42 }); say await $promise; say $promise.status
camelia rakudo-jvm 6c0f93: OUTPUT«===SORRY!===␤42␤»
llfourn lizmat: even in nqp?
timotimo m: my $promise = Promise.start(anon sub { fail 42 }); say await $promise; say $promise.status
camelia rakudo-moar 9308de: OUTPUT«===SORRY!===␤42␤»
lizmat ah, in nqp... hmmmm
timotimo labster: there we go
lizmat llfourn: timotimo might know 10:58
labster Thanks timotimo — but I'm not trying to do it correctly, I'm trying to figure out how RT #123204 should fail.
timotimo oh, about what exactly?
labster: it should say "can't fail outside of any routine"
llfourn timotimo: how to tell if value $contained is native or not :)?
in nqp context
timotimo a value cannot "be native"; only arguments and lexicals can 10:59
er, wait
that's BS :)
in nqp, you'll use nqp::objprimspec
the return value is 0 for object, 1, 2 or 3 for some order of int, num and str
nqp-m: my int $a := 1; say(nqp::objprimspec($a))
camelia nqp-moarvm: OUTPUT«1␤»
timotimo nqp-m: my $a := 1; say(nqp::objprimspec($a)) 11:00
camelia nqp-moarvm: OUTPUT«1␤»
timotimo nqp-m: my $a := "hello"; say(nqp::objprimspec($a))
camelia nqp-moarvm: OUTPUT«3␤»
timotimo nqp-m: my $a := 10.0e0; say(nqp::objprimspec($a))
camelia nqp-moarvm: OUTPUT«2␤»
timotimo there we go
llfourn thanks I'll give that a shot. The reason I wanted to know was that I see nqp::unbox_s used sometimes and sometimes not and just wanted to see what the difference is.
11:02 CQ left 11:05 xtreak joined
xtreak p6: my %a = <a b c d>; say %a.^methods; say %a.keyof() 11:06
camelia rakudo-moar 9308de: OUTPUT«(BIND-KEY STORE_AT_KEY name keyof of default dynamic push append classify-list categorize-list Method+{<anon|65740608>}.new Method+{<anon|65740608>}.new Method+{<anon|65740608>}.new perl gist DUMP Method+{<anon|65740608>}.new elems iterator STORE STORE_AT_…»
xtreak the keyof method doesn't seem to be implemented yet. Does it supposed to return the key for the value? 11:07
lizmat xtreak: where did you find "keyof" ???
it doesn't seem to be in the speculation (anymore) 11:08
11:09 pierre-vigier left
abraxxa lizmat: PR sent 11:10
xtreak src/core/Hash.pm method keyof . commits : fe792fdf37e8091c8b6892c826fa0ba6441664cc , cd927656d7a1a379c50cd4cf10edfc4bb66a6857
abraxxa nine: github.com/rakudo/rakudo/pull/648 shows the failure
11:11 pierre-vigier joined
xtreak Am on rakudobrew built at Nov 17. May be its removed at update? 11:11
timotimo oh 11:17
dalek kudo/nom: f5912bc | lizmat++ | src/core/Range.pm:
Doesn't make sense to fail, we want to throw!
timotimo .keyof gives you the type of the keys
dalek kudo/nom: 16a5f25 | (Alexander Hartmaier)++ | t/04-nativecall/02-simple-args. (2 files):
add test for size_t
 11:18
kudo/nom: 9f81d1d | (Alexander Hartmaier)++ | t/04-nativecall/15-rw-args. (2 files):
add tests for passing and returning data to C functions
kudo/nom: 47e04b6 | lizmat++ | t/04-nativecall/ (4 files):
Merge pull request #648 from abraxxa/nom

add tests for size_t and passing and returning data to C functions
abraxxa lizmat: thanks!
lizmat yw !
xtreak It always returns Any
11:18 Su-Shee joined
xtreak The method is `method keyof () { Any }` 11:19
timotimo xtreak: yeah, but it's overrided by the typed hash role, no?
lizmat yes it is
timotimo m: my %foo{Range}; say %foo.keyof
camelia rakudo-moar 9308de: OUTPUT«(Range)␤»
timotimo xtreak: here you can see that
bbiab 11:20
labster Well, managed to close 3 tickets before bedtime. 2 resolved, 1 rejected. I'll be back tomorrow to see if there are any more LHFs to pick before Christmas.
xtreak yes. I thought by keyof to return the key of a value. Thanks :) 11:21
nine lizmat: .oO(throw, throw, throw your $got, gently out of range...) 11:22
lizmat xtreak: you probably want to use .pairs on a hash, then you can do .key for the key and .value for the value 11:23
m: my %h = a => 42, b => 666; for %h.pairs { say .key; say .value } 11:24
camelia rakudo-moar 9308de: OUTPUT«a␤42␤b␤666␤»
lizmat m: my %h = a => 42, b => 666; for %h.kv => $key, $value { say $key; say $value } # alternate way for xtreak
camelia rakudo-moar 9308de: OUTPUT«5===SORRY!5=== Error while compiling /tmp/xd4B475o0B␤Variable '$key' is not declared␤at /tmp/xd4B475o0B:1␤------> 3my %h = a => 42, b => 666; for %h.kv => 7⏏5$key, $value { say $key; say $value } ␤»
lizmat m: my %h = a => 42, b => 666; for %h.kv -> $key, $value { say $key; say $value } # alternate way for xtreak
camelia rakudo-moar 9308de: OUTPUT«a␤42␤b␤666␤»
abraxxa do i have to use =begin pod and =end pod before/after each =METHOD when I want my perl 6 pod to be inline with the code?
lizmat =METHOD implies =begin pod, I would think 11:25
don't know about the =end
xtreak No. I thought to get a key of the given value like `my %caps = <UK London France Paris>` and `say %caps<London>.keyof()` to return UK
abraxxa lizmat: it seems everybody is writing perl 5 pod in the perl6 modules, at least the ones i've checked 11:26
is there a perldoc equivalent to render perl 6 pod?
lizmat abraxxa: tbh, I haven't followed the p6pod developments closely
xtreak: nope 11:27
llfourn abraxxa: you should be able to do perl6 --doc 11:28
abraxxa llfourn: thanks! I wasn't able to find that info anywhere
nine xtreak: it returns the type of the keys. Typing usually involves an .of method, that's where the name comes from.
11:29 rurban left
llfourn abraxxa: ironically I think pod is the least documented of anything on p6docs 11:29
abraxxa llfourn: yes, it seems to
llfourn: design.perl6.org/S26.html returns forbidden 11:30
11:30 labster left
nine abraxxa: make tests passes here? 11:30
llfourn moritz: ^^ see above if you are around :)
_nadim rindolf: hi, it seems that I was using an API that is 600 commits old. could you please try the latest?
abraxxa nine: hm?
nine abraxxa: the tests you committed pass. Are they supposed to pass or to fail? 11:31
rindolf _nadim: yes, let me see.
abraxxa nine: the size_t one fails
nine: perl6 -Ilib t/04-nativecall/02-simple-args.t
nine abraxxa: All tests successful
_nadim flussence: Could you please pull the latest and re-run the tests?
abraxxa nine: Unknown type 'size_t' used for native call 11:32
in method setup at /home/ahartmai/perl6/git/rakudo/lib/NativeCall.pm6:223
in method CALL-ME at /home/ahartmai/perl6/git/rakudo/lib/NativeCall.pm6:234
in block <unit> at t/04-nativecall/02-simple-args.t:66
nine: perl6::version=2015.11-740-g0734842
rakudobrew build moar nom
a few hours ago
nine abraxxa: indeed. It breaks with perl6 -Ilib t/04-nativecall/02-simple-args.t but passes with make test
abraxxa nine: weird 11:33
are they run?
nine yes
abraxxa nine: did you see my pointer to where the problem lies before?
rindolf _nadim: panda still won't run, much less install Text::Table::Simple though. 11:34
11:34 Skarsnik joined
_nadim rindolf: hmm, I thought I fixed that, let me check 11:34
rindolf _nadim: I didn't try. 11:35
nine It also passes when run by /usr/bin/perl t/harness --moar t/04-nativecall/02-simple-args.t
11:35 rurban joined
rindolf _nadim: OK, now I'm getting much fewer failures. 11:36
abraxxa Could not execute (./perl6-m t/04-nativecall/02-simple-args.t): open3: exec of ./perl6-m t/04-nativecall/02-simple-args.t failed at /usr/share/perl/5.20/TAP/Parser/Iterator/Process.pm line 165.
_nadim rindolf: well I missed one place, at t/01_all.t line 65, there is a use Text::.... that is not a in a try, so that may fail, although the test is a todo
rindolf _nadim: www.shlomifish.org/Files/files/text...a-Dump.txt 11:37
rindolf forgot what he wanted to do now.
11:37 kjs_ left
_nadim rindolf: thanks mate, I'll remove the dependency and fix the last error. almost there. 11:40
xtreak Does %caps{Int} means that only only integer keys can be used?
11:41 pierre-vigier left
xtreak Yes got it. I was again using 'my' for the assignment again and it got shadowed. 11:41
llfourn m: my %caps{Int} = (1 => 'str') 11:42
camelia ( no output )
llfourn m: my %caps{Int} = ('str' => 1)
camelia rakudo-moar 47e04b: OUTPUT«Type check failed in binding key; expected Int but got Str␤ in block <unit> at /tmp/ue26FsEVfy:1␤␤»
11:42 pierre-vigier joined
abraxxa nine: anything I can do to fix it? 11:47
dalek kudo/nom: 82fe5fb | lizmat++ | / (6 files):
Split Temporal into Dateish/DateTime/Date

Was starting to make errors not seeing in which class I was making changes. Plus it was causing confusion among would-be core developers looking for the code that makes Date/DateTime tick :-)
11:50 shmibs joined, shmibs left 11:52 kjs_ joined
abraxxa seen moritz? 11:56
moritz abraxxa: I'm around-ish
11:56 moritz sets mode: -o mohae, moritz sets mode: -o moritz
abraxxa moritz: DBDish::Oracle is almost ready to merge 11:56
not sure how I should handle the still failing tests 11:57
moritz abraxxa: you mean besides fixing them?
abraxxa moritz: things like different sort order by database 11:58
Oracle sorts NULLs last by default and it can only be changes through a different sql query
ORDER BY name NULLS FIRST
moritz abraxxa: if the other backends support that too, change the SQL
abraxxa moritz: guess what, they don't
moritz they don't? 11:59
abraxxa so another sql query that depends on the DBD
nine What about ORDER BY coalesce(name, '')?
moritz well, you can also ORDER BY COAELESE(thecolumn, 0) or so
nine ha :)
Great minds think alike!
moritz nine: except that you spelled it correctly, afaict :-) 12:00
abraxxa nine: not that Oracle handles NULL and '' differently ;) 12:01
|Tux| yes it does, in char fields 12:03
moritz there's also ORDER BY column IS NULL DESC, column DESC or so 12:04
that should work in all four, I hope
|Tux| tux.nl/Talks/DBDc/null.html
abraxxa ORDER BY COALESCE(name,'A') does the trick, do you recommend some other function? 12:06
ORDER BY name IS NULL DESC results in ORA-00933: SQL command not properly ended on Oracle 12c 12:07
moritz then use COALESCE
abraxxa ok, thanks!
moritz I know mysql supports it
abraxxa SQLite does too
don't have MySQL or Postgres installed locally to test though
moritz and postgres too 12:08
abraxxa great!
12:08 regreg joined
abraxxa then in only need a solution for 'FROM DUAL' for Oracle 12:09
12:10 administrator joined
abraxxa is there a done_testing instead of again having to plan the number of tests? 12:10
lizmat I wonder whether 'use v6.c' should set a dynamic variable that modules could inspect to see under what version they're supposed to be running
12:10 administrator is now known as Guest54696
abraxxa lizmat: pre v6.c compat isn't important, isn't it? 12:11
lizmat: but going forward it will be
lizmat yes, indeed
12:11 hankache left 12:12 lichtkind joined
lizmat probably want to engrain this at an early stage 12:12
m: say $*PERL # perhaps that's the one ? 12:13
camelia rakudo-moar 82fe5f: OUTPUT«Perl 6 (6.b)␤»
12:13 kjs_ left, loren joined
abraxxa what does @a ~~ @b do? 12:13
Skarsnik m: say $*PERL.version;
camelia rakudo-moar 82fe5f: OUTPUT«v6.b␤»
abraxxa Skarsnik: hi!
Skarsnik Hello 12:14
12:14 cognominal joined
moritz design.perl6.org/S03#Smart_matching Positional List lists are comparable $_ »~~« X (but dwims ** wildcards!) 12:14
timotimo now to write tests for += and friends triggerint failures
loren m: say (1, 23) ~~ (1, 23); say (1, 23) ~~ (1, 32); 12:15
camelia rakudo-moar 82fe5f: OUTPUT«True␤False␤»
Skarsnik abraxxa, did you figure why is-deeply fail? x)
abraxxa Skarsnik: no, sadly not. but nom breaks DBDish and I'm waiting on nine to look into it, already commited a test for it
Skarsnik: i'm working on getting the tests in 99-common.t work together nicely with DBDish::Oracle, when I got this I'll send the pull request and continue its development in the main repo 12:16
12:17 pierre-vigier left
Skarsnik the break is because of lib not requiered in the libname anymore? 12:17
abraxxa moritz: thanks. So using it for comparing two arrays which should be the same length and contain the same typed values or Type objects should be good for a test? 12:18
12:18 vv3 joined, vv3 left
abraxxa Skarsnik: for the other DBDs yes, Oracle also breaks because of size_t 12:18
12:19 vv3 joined, vv3 left, vv3 joined, vv3 left
lichtkind if i get it .hyper makes any list list generating function into an supply? 12:20
loren Hi, I wonder whether Array has an method can swap two element
12:20 vv3 joined
loren I look for document about Array, but not found/ 12:20
timotimo loren: you can @arr[1, 10] .= reverse 12:21
Skarsnik abraxxa, you should add a test in xt/ for oracle (copy past one of the other) x) 12:22
loren m: say (1, 2, 3)[1,2] .= reverse;
camelia rakudo-moar 82fe5f: OUTPUT«Cannot modify an immutable Int␤ in block <unit> at /tmp/QiIgOL4ZeQ:1␤␤»
abraxxa Skarsnik: ah, haven't seen those
loren m: my @x = (1, 2, 3); @x[1,2] .= reverse; say @x;
camelia rakudo-moar 82fe5f: OUTPUT«[1 3 2]␤»
12:22 Actualeyes joined
loren timotimo, thanks 12:22
timotimo yw
@x[1,2] = @x[2,1] also works 12:23
12:23 vv3 left, vv3 joined
loren em, got it . ^_^ 12:23
12:23 kid51 joined
abraxxa Skarsnik: the 04_pgnative.t has sqlite headers in int 12:24
12:24 vv3 left
dalek kudo/nom: cc46351 | lizmat++ | src/core/Date (3 files):
Make .daycount a lazy Dateish attribute

Before, it was always calculated for Date's, but DateTime internally also used it in some cases by first creating a Date object for it. Seemed daycount wasn't always needed anyway, for many cases, so that making it lazy and cleaning up the interface seemed like a good idea.		 12:24
12:25 vv3 joined, vv3 left 12:26 vv3 joined, vv3 left
abraxxa Skarsnik: Could not find NativeCall::TypeDiag in any of: 12:26
is it newer than 2015.11?
hm, same for nom
Skarsnik It's a module
12:26 vv3 joined
abraxxa a non-core one? 12:27
Skarsnik Yes
abraxxa i see
Skarsnik hm, I did not run the pg test? x)
abraxxa the two vars aren't used because the first test is commented out 12:28
Skarsnik Oh yes, Pg does not define any CStruct
too much copy paste x)
moritz Pg works with OpaquePointer only 12:29
I love that C API
abraxxa moritz: OCI too
only pass and return by pointer
doesn't panda do dependency resolution for modules? 12:30
Skarsnik how do you get information about a field?
I did not put NC::Typediag in the dep, since it's only an author test 12:31
abraxxa Skarsnik: github.com/abraxxa/DBIish/blob/mas...e.pm6#L483
TypeDiag doesn't list File::Temp, I had to install it manually
Skarsnik ho, my bad
12:31 vv3 left
abraxxa NativeCall::TypeDiag:ver<*>:auth<Sylvain Colinet>:api<> already installed 12:31
and I can't install it even after installing File::Temp 12:32
moritz panda install -force maybe?
12:32 vv3 joined, vv3 left
abraxxa it seems it's panda --force install 12:33
nine lizmat: indeed, I've heard lots about language backwards compatibility. But what about the runtime? How will e.g. List's method know how to behave?
abraxxa Could not find NativeCall::TypeDiag in:
;(
12:33 rurban left, vv3 joined, vv3 left
abraxxa Skarsnik: does it work on nom for you? 12:34
timotimo how do i best test for a failure to blow up?
lizmat nine: well, that would be my point: by checking on $*PERL ? if that's being set by use v6.n
abraxxa panda fails to install on 2015.11, so I can't test on that either
moritz timotimo: dies-ok { ~things-that-fail() } or so
timotimo i tried class X::ThisFailure { }; my $foo = Failure.new(X::ThisFailure.new()), but i get an X::AdHoc from throws-like
abraxxa timotimo: catch the exception?
timotimo moritz: dies-ok seems like a terrible idea :) 12:35
like, in general
Skarsnik erf
all my vmbox save are dead ><
moritz m: class X::F { has $.message = 'boo' }; sub f () { fail X::F.new }; say f().^name 12:36
camelia rakudo-moar cc4635: OUTPUT«Failure␤»
Skarsnik great it again lost inter vx tek, I can't boot my 64 bits vm with nom ><
moritz m: class X::F { has $.message = 'boo' }; sub f () { fail X::F.new }; say f().exception.^name
camelia rakudo-moar cc4635: OUTPUT«X::AdHoc␤»
12:36 vv3 joined
moritz that looks...wrong 12:36
El_Che hi my favorite perl6 people!
moritz m: class X::F is Exception { has $.message = 'boo' }; sub f () { fail X::F.new }; say f().exception.^name 12:37
camelia rakudo-moar cc4635: OUTPUT«X::F␤»
moritz oh, it works like that
loren El_Che, Hi...
lizmat wrt to dies-ok, jnthn changed a lot of checks for X::AdHoc to dies-ok :-(
llfourn moritz: thanks for fixing docs build and imporving robustness. fyi design.perl6.org/S26.html is note completely blank but is 403 :)
/note/not/
moritz lizmat: well, X::AdHoc is wrong when we start to introduce typed exceptions
lizmat: better to test for Exception than to test X::AdHoc 12:38
lizmat well, that's the same as dies-ok
so we get false positives
:-(
Skarsnik how to test the fail message btw? 12:39
12:39 vv3 left
moritz Skarsnik: once it's handled, you can stringify it 12:40
12:40 vv3 joined, vv3 left 12:41 vv3 joined, vv3 left
nine abraxxa: you do have NativeCall::TypeDiag installed but cannot load it? 12:42
12:42 vv3 joined, vv3 left
Skarsnik I think having a missing dep screw it? 12:43
12:43 vv3 joined
abraxxa nine: panda --force install NativeCall::TypeDiag said install successful on my nom build 12:43
nine abraxxa: but perl6 -e 'use NativeCall::TypeDiag' fails? 12:44
abraxxa nine: yes
Could not find NativeCall::TypeDiag in:
nine Oh, I can even reproduce this! 12:45
Skarsnik I just fixed the missing dep xD
12:45 kjs_ joined
nine And that's the reason: "Gumbo" : "lib/NativeCall/TypeDiag.pm6" in META.info 12:45
perl6 -e 'use Gumbo' # obvious fix 12:46
Skarsnik lol
I should use/write something to generate the base code for my modules x) 12:47
12:47 Begi2 is now known as begi, begi is now known as Begi
moritz Skarsnik: please take over Module::Starter 12:47
Skarsnik abraxxa, ok try remove and install again?
12:48 vv3 left
abraxxa Skarsnik: does panda have an uninstall/remove option? 12:48
12:48 pierre-vigier joined
Skarsnik I wanted to use mi6 but using it to run test rebuild the META, it screw me the dbiish meta file by changing the author x) 12:48
12:49 vv3 joined, vv3 left, kjs_ left
Skarsnik No idea. it should ? 12:49
nine abraxxa: not yet
12:49 vv3 joined, vv3 left
abraxxa panda --force install NativeCall::TypeDiag again said ==> Successfully installed NativeCall::TypeDiag but it still can't be found 12:49
Skarsnik update?
12:50 vv3 joined, vv3 left
nine abraxxa: panda update before installing again 12:50
abraxxa nine: oh, thanks 12:51
12:51 vv3 joined
abraxxa still had to force it but now it seems to work 12:51
12:52 kid51 left
nine We all should start versioning our modules. Otherwise people will have to use --force all the time 12:52
Skarsnik panda has a gen-meta option, fun
abraxxa Skarsnik: OCI has multiple header files but InstantClient doesn't install into the regular lib dirs, can I specify the dir somehow? 12:53
Skarsnik you could always set @nctd-extracompileroptions 12:54
abraxxa Skarsnik: where is that documented? 12:55
nine: anything I can do to fix size_t?
Skarsnik but the diag-function compile nothing, it just look at the routine signature
12:56 vv3 left
nine abraxxa: find out why the hell the test fails when run with perl6 directly and why it passes when run in the harness 12:56
Skarsnik constant size_t is export = long; remove that
size_t exist
now
abraxxa nine: I can't run the harness
nine abraxxa: why? 12:57
12:57 vv3 joined, vv3 left, vv3 joined, vv3 left
abraxxa nine: Could not execute (./perl6-m t/04-nativecall/02-simple-args.t): open3: exec of ./perl6-m t/04-nativecall/02-simple-args.t failed at /home/ahartmai/perl5/perlbrew/perls/22.1/lib/5.22.1/TAP/Parser/Iterator/Process.pm line 165. 12:58
when I run perl t/harness --moar t/04-nativecall/02-simple-args.t
12:58 vv3 joined, vv3 left
Skarsnik just run perl6 -I lib t/04-nc/02 ? 12:59
12:59 vv3 joined
nine Skarsnik: doesn't help. The test fails (as it should) when run that way but passen when run with the harness (make test) 12:59
13:00 rurban joined
Skarsnik duh 13:00
abraxxa how can i build a List from one column of an Array of Arrays? 13:01
lizmat abraxxa: the nice syntax for that using * is still NYI, afaik 13:02
abraxxa @array.map({ .[3] })? 13:04
13:04 vv3 left
Skarsnik btw if you have the same issue with is-deeply that I have, it's probably a rakudo bug on eqv 13:05
13:05 vv3 joined, vv3 left
abraxxa works but returns a Seq 13:05
13:07 xinming joined, vv3 joined 13:08 pierre-vigier left
abraxxa nine: make test passes here too 13:09
13:10 AndyDee left, RabidGravy joined
abraxxa and ./perl6-m -Ilib t/04-nativecall/02-simple-args.t fails 13:10
nine: but as I said before, size_t.^shortname is Int which maps to longlong instead of size_t in %type_map 13:11
13:12 vv3 left, xinming_ joined 13:13 vv3 joined, vv3 left 13:14 vv3 joined, vv3 left 13:15 vv3 joined, vv3 left, xinming left
abraxxa what is the difference between :amount($(Rat, Rat, 2.4, 4.95, Rat, 4.85)) and :amount($[Rat, Rat, 2.4, 4.95, Rat, 4.85]) ? 13:15
the first is generated by the map().list
13:15 vv3 joined, kjs_ joined 13:18 kjs_ left 13:19 kjs_ joined 13:20 vv3 left, shmibs joined 13:21 vv3 joined, vv3 left
moritz abraxxa: () vs [] 13:22
abraxxa: () is a list, [] is an Array
abraxxa yes, but I can't figure out what is what
13:22 vv3 joined, vv3 left
moritz m: say $[].^name 13:22
camelia rakudo-moar cc4635: OUTPUT«Array␤»
moritz m: say $().^name
camelia rakudo-moar cc4635: OUTPUT«Use of Nil in string context in block <unit> at /tmp/MVusf41i2J:1␤Str␤»
moritz m: say $(1, 2).^name
camelia rakudo-moar cc4635: OUTPUT«List␤»
Skarsnik m: use NativeCall; say size_t.^shortname;
camelia rakudo-moar cc4635: OUTPUT«size_t␤»
abraxxa i still hate that there are two types!
as the code pushes it only can return an Array, correct?
moritz right 13:23
13:23 vv3 joined, vv3 left
abraxxa i tried initializing it with () instead of [] but that didn't work 13:23
so how do i get an Array out of the Seq returned by map?
Skarsnik I don't get your issue with size_t x)
RabidGravy flat
abraxxa RabidGravy: flat returns a List too
13:24 vv3 joined
abraxxa hm, doc bug 13:24
Interprets the invocant as a list, flattens it, and returns that list.
but: method flat() return Iterable
the first is from Class Any, the later from role Iterable
13:25 andreoss joined
abraxxa Skarsnik: does the test fail for you too? 13:25
moritz List ~~ Iterable
so not a bug
abraxxa moritz: more of an annoyance with the docs that reload the page when you click a method and this new page shows all possible sources and not only the one used by the Class or Type you clicked the link in 13:27
so is there a way to get an Array from a Seq?
lizmat [$seq.list] ?
moritz [$seq] ?
$seq.Array? 13:28
m: say (gather { take 1; take 2}).Array.perl
camelia rakudo-moar cc4635: OUTPUT«[1, 2]␤»
13:28 kjs_ left
moritz m: say [gather { take 1; take 2}].perl 13:28
camelia rakudo-moar cc4635: OUTPUT«[1, 2]␤»
13:28 darutoko left
abraxxa [ $seq.list ] works, thanks 13:28
13:28 vv3 left
Skarsnik moritz, can you look at dbiish to have row usable in for? (without being greedy) 13:29
abraxxa this needs BIG red letters in the docs: [] is an Array, () is a List
lizmat abraxxa: from what moritz said, $seq.Array should also work
abraxxa which is weird for p5 people where you write my @rray = ();
13:29 vv3 joined, vv3 left
abraxxa not listed here: doc.perl6.org/type/Seq 13:30
nine or even simpler: my @column = @array.map({ .[3] });
13:30 vv3 joined, vv3 left
lizmat m: dd my @a = (1,2,3) # abraxxa: that still works in P6 13:30
camelia rakudo-moar cc4635: OUTPUT«Array @a = [1, 2, 3]␤»
abraxxa nine: i have that in a hash
lizmat m: dd my @a = 1,2,3 # abraxxa: but the parens are superstitious 13:31
camelia rakudo-moar cc4635: OUTPUT«Array @a = [1, 2, 3]␤»
nine m: my @array = [1, 2, 3], [2, 3, 4], [3, 4, 5]; my @column = @array.map(*.[2]); say @column;
camelia rakudo-moar cc4635: OUTPUT«[3 4 5]␤»
abraxxa lizmat: yes, which makes it even more confusing! does this build a List which is then assigned to the Array?
13:31 vv3 joined, vv3 left
lizmat abraxxa: yup 13:31
, is the list creator
m: dd (42); dd (42,)
camelia rakudo-moar cc4635: OUTPUT«42␤(42,)␤»
abraxxa lizmat: so my @array = []; performs better because no List object is generated and thrown away immediatly? 13:32
13:32 vv3 joined
abraxxa .Array works, can we please add it to the docs? 13:32
not that is_deeply is satified by it, put at least the printout is the same 13:33
lizmat abraxxa: could be, would need to benchmark that
m: dd my @a = []; dd my @b = [],[] # abraxxa: single arg rule at play! 13:34
camelia rakudo-moar cc4635: OUTPUT«Array @a = []␤Array @b = [[], []]␤»
abraxxa because if yes all example code should only assign [] to @rray
so what is a hash then, i guess {}? 13:35
Skarsnik: if yes this makes no sense: github.com/perl6/DBIish/blob/maste...n.pl6#L179
nine No, no, no. Even if benchmarks say that it's faster (which they won't), let's not optimize example code for the _current_ performance characteristics that may change as soon as someone plays half an hour with the optimizer.
abraxxa as it's an Array of key-value pairs
nine: agreed, but assigning an Array to an Array makes more sense for newcomers then assigning a List to it 13:36
just to not have some sort of type conversion going on in the simplest examples
nine Arrays _are_ Lists
13:36 leont joined
abraxxa no, they are two different types with some identical apis 13:36
nine my @a = 1, 2, 3; is a perfect idiom to learn
lizmat m: say Array.^mro
camelia rakudo-moar cc4635: OUTPUT«((Array) (List) (Cool) (Any) (Mu))␤»
Skarsnik m: say (1,2,3) eqv [1, 2, 3]
camelia rakudo-moar cc4635: OUTPUT«False␤»
abraxxa if they would be the same one would be useless 13:37
13:37 vv3 left
nine Arrays _are_ Lists. We are object oriented now and we do have inheritance (aka is-a relationships) 13:37
abraxxa Skarsnik: exaclty
^^
colomon m: say Array ~~ List
camelia rakudo-moar cc4635: OUTPUT«True␤»
colomon m: say List ~~ Array
camelia rakudo-moar cc4635: OUTPUT«False␤»
abraxxa nine: how do you define _are_
colomon what nine++ said
13:37 vv3 joined, vv3 left
Skarsnik hm, maybe the second test should work with @ref = [stuff ] instead of () 13:37
abraxxa Lists are immuatble, Arrays aren't 13:38
lizmat Array isa List # p5 think
nine m: say Array.isa(List
camelia rakudo-moar cc4635: OUTPUT«5===SORRY!5=== Error while compiling /tmp/lHKG2l_Jp5␤Unable to parse expression in argument list; couldn't find final ')' ␤at /tmp/lHKG2l_Jp5:1␤------> 3say Array.isa(List7⏏5<EOL>␤»
nine m: say Array.isa(List)
camelia rakudo-moar cc4635: OUTPUT«True␤»
nine abraxxa: that way ^^^
Skarsnik err that mean eqv is buggy?
nine just learned, that we still have .isa in Perl 6
abraxxa Arrays are a subclass of List is correct
13:38 vv3 joined, vv3 left
Skarsnik m: say (1,2,3) eqv [1, 2, 3]; say [1, 2, 3] eqv (1,2,3); 13:38
camelia rakudo-moar cc4635: OUTPUT«False␤False␤»
Skarsnik or not
abraxxa so is {} a Hash?
nine m: say {}.^name 13:39
camelia rakudo-moar cc4635: OUTPUT«Hash␤»
13:39 vv3 joined, vv3 left
leont Hey *, how's christmas coming? :-) 13:39
13:39 lucasb joined
Skarsnik m: my %ref = (a => 2); say %ref.WHAT; 13:40
camelia rakudo-moar cc4635: OUTPUT«(Hash)␤»
13:40 vv3 joined
nine leont: fine I dare say :) jnthn++ cut down the Xmas RT list to 0 with help of a few others. I want to land just one more patch 13:40
leont Awesome :-) 13:41
Skarsnik nine, the issue, this dumb test fail (is-deeply) github.com/perl6/DBIish/blob/maste...n.pl6#L175 even if Test print the same string on expected/got 13:42
nine abraxxa: note that I never said that Arrays are the _same_ as Lists. They are a special kind of List, but wherever code wants a List, you can give it an Array and it will work just fine
lucasb m: say (loop (my $i = 0; $i < 3; $i++) { 42 }) 13:44
camelia rakudo-moar cc4635: OUTPUT«(timeout)»
RabidGravy :-\ 13:45
13:45 vv3 left
lucasb I'm happy that 'while' now return values, so I thought I would try that with 'loop' 13:45
Skarsnik abraxxa, t/04-nativecall/02-simple-args.t ..... ok
thomax nsty bug it is
13:46 vv3 joined, vv3 left, vv3 joined, vv3 left
lucasb star: say (while $++ < 3 { 1+2 }) 13:47
camelia star-m 2015.09: OUTPUT«sub infix:<<> (Mu $?, Mu $?) { #`(Sub+{<anon>}|42774000) ... }␤»
lucasb star: say (loop (my $i = 0; $i < 3; $i++) { 42 })
camelia star-m 2015.09: OUTPUT«sub infix:<<> (Mu $?, Mu $?) { #`(Sub+{<anon>}|41680576) ... }␤»
abraxxa nine: in this test I want to be explicit as possible as it tests the API
lucasb ^^ just as a data point, before while and loop returned the same thing... the last operation executed maybe
13:47 vv3 joined, vv3 left
abraxxa Skarsnik: make test or running it directly? 13:48
Skarsnik make test
13:48 vv3 joined
abraxxa Skarsnik: same here, run it direclty please 13:48
Skarsnik why it's so slow ~~ 13:50
nine Skarsnik: it compiles a lot of native libs 13:51
thomax m: say (loop (my $i = 0; $i < 3; $i++) { print $i; })
Skarsnik Oh yeah, that's true
camelia rakudo-moar cc4635: OUTPUT«(timeout)0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000…» 13:52
Begi O_o
13:53 vv3 left
Skarsnik Oh I see abraxxa 13:53
that's weird
nine Skarsnik: t/35-pg-common.t passes here?
13:54 vv3 joined, vv3 left, hankache joined 13:55 vv3 joined, vv3 left
Skarsnik hm, I don't have pg setup for test ~~ 13:55
13:55 vv3 joined, vv3 left 13:56 vv3 joined
nine Skarsnik: I guess it doesn't mean much that they pass: Cannot locate native library 'liblibpq.so' 13:56
Skarsnik I did not update DBIis for api version/new native trait because it will fail to work with 2 week old rakudo x) 13:57
loren m: my @source = 1, 2, 3; my @display; @display.push: @source.clone(); my @temp = @source.clone(); @display.push: @temp; @source[1,2]=@source[2,1]; say @display; 13:59
camelia rakudo-moar cc4635: OUTPUT«[[1 3 2] [1 2 3]]␤»
14:00 skids joined
loren Hi, i have a question about above code result.. 14:00
Has anyone can have a look about it ? 14:01
14:01 vv3 left 14:02 vv3 joined, vv3 left
Skarsnik github.com/MoarVM/MoarVM/blob/mast...call.c#L48 this code is dumb lol but yes, it missing bool and size_t 14:02
geekosaur looks right to me?
loren I want to use clone copy a Array, but when i use like this [email@hidden.address] @source.clone()', it's not working .. 14:03
14:03 vv3 joined, vv3 left
geekosaur oh, missed that it's cloned twice 14:03
14:03 vv3 joined, vv3 left
loren geekosaur, em the result a little strange .. 14:04
14:04 vv3 joined 14:05 xinming joined
RabidGravy append 14:05
14:05 llfourn left
loren I don't want to append 14:05
'@display' shound be an Array of Array
RabidGravy then maybe say what you expect?
loren m: my @source = 1, 2, 3; my @display; my @temp = @source.clone(); @display.push: @temp; @source[1,2]=@source[2,1]; say @display; 14:06
camelia rakudo-moar cc4635: OUTPUT«[[1 2 3]]␤»
loren These code work fine
but [email@hidden.address] @source.clone()' not
abraxxa Skarsnik: please look at my latest commits: github.com/abraxxa/DBIish/commits/master 14:07
Skarsnik .tell FROGGS You forget to update the plan on t/04-nc/02.t to 13 and make the test succes on make test, when it fail for the 13th because size_t (and bool) are not listed on moar/src/core/nc.c 14:08
yoleaux Skarsnik: I'll pass your message to FROGGS.
14:08 xinming_ left, skids left
Skarsnik abraxxa, ~~ will compare the type only probably 14:09
hankache m: my @source = 1, 2, 3; my @display; @display.push: @source.clone(); say @display;
camelia rakudo-moar cc4635: OUTPUT«[[1 2 3]]␤»
14:09 vv3 left
hankache loren ^^ 14:09
how is it not working?
loren ..I'm now mean that 14:10
m: my @source = 1, 2, 3; my @display; @display.push: @source.clone(); @source[1,2]=@source[2,1]; say @display;
camelia rakudo-moar cc4635: OUTPUT«[[1 3 2]]␤»
14:10 vv3 joined, vv3 left
Skarsnik abraxxa, for size_t, well... keep your constant but with another name (like FIXMELATER?) x) 14:10
loren hankache, '@display' also change when i change '@source' 14:11
14:11 vv3 joined, vv3 left 14:12 vv3 joined, vv3 left, vv3 joined
hankache loren ah! it seems like clone binds instead of assigning 14:13
loren yeah, hankache 14:14
hankache m: my $a = 1; my $b = $a.clone; $a = 3; say $b; 14:15
camelia rakudo-moar cc4635: OUTPUT«1␤»
hankache weird 14:16
m: my @a = [1,2]; my @b = @a.clone; @a = [9,9]; say @b; 14:17
camelia rakudo-moar cc4635: OUTPUT«[1 2]␤»
14:17 vv3 left
loren m: my @a = [1,2]; my @b = @a.clone; @a[1,2] = @[2,1]; say @b; 14:17
camelia rakudo-moar cc4635: OUTPUT«[1 2]␤»
abraxxa Skarsnik: good idea 14:18
loren m: my @a = [1,2]; my @b = @a.clone; @a[1,2] = @a[2,1]; say @b;
camelia rakudo-moar cc4635: OUTPUT«[1 2]␤»
hankache loren it doesn't seem like it is binding
let's see your initial example
loren hankache, has problem when push it to an Array 14:19
m: my @a = [1,2]; my @b; @b.push: @a.clone; @a[0,1] = @a[1,0]; say @b; 14:20
camelia rakudo-moar cc4635: OUTPUT«[[2 1]]␤»
Skarsnik nine, can you change NC.pm:183 to return $*VM.platform-library-name($libname.IO, :version($apiversion ?? Version.new($apiversion) !! Version)).Str; It allow to write native('foo', 1) instead of forcing to v1
loren m: my @a = [1,2]; my @b = @a.clone; @a[0,1] = @a[1,0]; say @b;
camelia rakudo-moar cc4635: OUTPUT«[1 2]␤»
14:21 vv3 joined
nine Skarsnik: I'm not sure I want that. I figured, we should start out being very strict about the types we require. We can always relax the requirements later but we may never tighten them up. 14:21
Skarsnik ok 14:22
m: say v1.WHAT
camelia rakudo-moar cc4635: OUTPUT«v1.WHAT␤»
Skarsnik m: say v1.0.WHAT 14:23
camelia rakudo-moar cc4635: OUTPUT«v1.0.WHAT␤»
Skarsnik m: say (v1).WHAT
camelia rakudo-moar cc4635: OUTPUT«(Version)␤»
Begi m: say (v1).WHAT;
camelia rakudo-moar cc4635: OUTPUT«(Version)␤»
hankache m: my @a = [1,2]; my @b; @b.push: |@a; @a[0,1] = @a[1,0]; say @b;
Skarsnik say (v0).WHAT;
camelia rakudo-moar cc4635: OUTPUT«[1 2]␤»
dalek kudo/nom: 214debb | (Stefan Seifert)++ | t/04-nativecall/02-simple-args.t:
Fix wrong test plan hiding a failing test

15:08 < Skarsnik> .tell FROGGS You forget to update the plan on t/04-nc/02.t to 13 and make the test succes on make test, when it fail for the 13th because size_t (and bool) are not listed on moar/src/core/nc.c		 14:24
hankache loren : my @a = [1,2]; my @b; @b.push: |@a; @a[0,1] = @a[1,0]; say @b;
Skarsnik say (v0).WHAT;
err
m: say (v0).WHAT;
camelia rakudo-moar cc4635: OUTPUT«(Version)␤»
Skarsnik it look weird x)
Begi m: say (v6).WHAT; 14:25
camelia rakudo-moar cc4635: OUTPUT«(Version)␤»
hankache loren you are pushing @a into @b. @a is an array in other words: a container or a reference
14:25 vv3 left 14:26 darutoko joined
loren hankache, I want an Array of Array 14:26
14:26 vv3 joined, vv3 left
nine loren: try .append instead of .push 14:27
hankache loren forget what i just said... i am lost :(
14:27 vv3 joined, vv3 left
loren hankache, it doesn't matter 14:27
hankache Ladies and gents can someone help us please 14:28
m: my @a = [1,2]; my @b; @b.push: @a; @a[0,1] = @a[1,0]; say @b;
camelia rakudo-moar cc4635: OUTPUT«[[2 1]]␤»
14:28 vv3 joined, vv3 left
leont Just tried building a new rakudo, but it failed with a «Could not find symbol '&bool'» during tools/build/install-core-dist.pl 14:28
loren nine, I don't want use append cause i need an Array of Array
nine leont: remove install/share/perl6 please
hankache why when we modify @a then @b also gets affected?
RabidGravy "unsigned long long int" - that's special 14:29
14:29 vv3 joined
nine hankache: [[2 1]] it's an array containing @a 14:29
Skarsnik I have the same can't find Shell::Command on panda hm
grondilu hankache: see @a as a reference.
nine Skarsnik: is there any trace (try ack Shell::Command) of Shell::Command in your ~/.perl6 or install/share/perl6? 14:30
hankache m: my $a = 1; my $b = $a; $a = 3; say $b; 14:31
camelia rakudo-moar cc4635: OUTPUT«1␤»
grondilu m: my @a = [1,2]; my @b; @b.push: @a[]; @a[0,1] = @a[1,0]; say @b;
camelia rakudo-moar cc4635: OUTPUT«[[2 1]]␤»
grondilu hum
that's unexpected
loren hankache, the code is not correct
grondilu m: my @a = [1,2]; my @b; @b.push: @@a; @a[0,1] = @a[1,0]; say @b;
camelia rakudo-moar 214deb: OUTPUT«[[2 1]]␤»
loren m: my @source = 1, 2, 3; my @display; @display.push: @source.clone(); @source[1,2]=@source[2,1]; say @display;
camelia rakudo-moar 214deb: OUTPUT«[[1 3 2]]␤»
grondilu m: my @a = [1,2]; my @b; @b.push: @a.clone; @a[0,1] = @a[1,0]; say @b;
camelia rakudo-moar 214deb: OUTPUT«[[2 1]]␤»
loren I mean like that
grondilu wth
nine m: my @a = [1,2]; my @b; @b.push: [|@a]; @a[0,1] = @a[1,0]; say @b; 14:32
camelia rakudo-moar 214deb: OUTPUT«[[1 2]]␤»
Skarsnik not here
loren grondilu, yep , use clone
lucasb ok, can we all agree that array cloning is not cloning? :)
grondilu but clone did not work.
14:32 xtreak left
nine My version does work :) 14:32
Skarsnik Oh could it because I did not update rakudobrew?
grondilu m: my @a = [1,2]; my @b; @b.push: @a.List; @a[0,1] = @a[1,0]; say @b;
camelia rakudo-moar 214deb: OUTPUT«[(1 2)]␤»
Skarsnik I just rakudobrew nuke
geekosaur nine, I think the point here is that .clone works with a temp var but not directly in .push 14:33
loren m: my @source = 1, 2, 3; my @display; my @t = @source.clone(); @display.push: @t; @source[1,2]=@source[2,1]; say @display;
camelia rakudo-moar 214deb: OUTPUT«[[1 2 3]]␤»
grondilu m: my @a = [1,2]; my @b; @b.push: @a[*]; @a[0,1] = @a[1,0]; say @b;
camelia rakudo-moar 214deb: OUTPUT«[(2 1)]␤»
nine I wish rakudobrew would not exist...
loren use temp variable is working fine
hankache so is &clone the culprit? 14:34
14:34 vv3 left
loren I don't know about detail.. 14:34
nine loren: because clone is a very low level too that you just shouldn't use unless you know it's the right tool.
14:35 vv3 joined, vv3 left
grondilu m: my @a = 1; my @b = @a.clone; @a[0] = pi; say @b 14:35
camelia rakudo-moar 214deb: OUTPUT«[1]␤»
lucasb m: my @a = 1,2; my @b := @a.clone; say @b; @a.push(42); say @b
camelia rakudo-moar 214deb: OUTPUT«[1 2]␤[1 2 42]␤»
14:35 vv3 joined, vv3 left
hankache m: my @a = 1,2; my @b = @a.clone; say @b; @a.push(42); say @b 14:36
camelia rakudo-moar 214deb: OUTPUT«[1 2]␤[1 2]␤»
loren nine, so when i don't want a reference. what should i do ?
dalek blets: 11b52a7 | (Herbert Breunung)++ | docs/appendix- (3 files):
+ nav anchor
blets: 6f40514 | (Herbert Breunung)++ | docs/appendix-b-grouped.txt:
Merge branch 'master' of github.com:perl6/tablets
blets: 0d83447 | (Herbert Breunung)++ | docs/ (4 files):
fixed added navigation in A and B
grondilu m: my @a = [1,2]; my @b; @b.push: [@a]; @a[0,1] = @a[1,0]; say @b;
camelia rakudo-moar 214deb: OUTPUT«[[1 2]]␤»
hankache loren [|@arrayname]
nine loren: like I showed, the easiest way to copy an array seems to be [|@a]
grondilu | is not necessay here
14:37 Ven joined
nine true, it's the single arg rule 14:37
grondilu it is
Ven irclog.perlgeek.de/perl6/2015-12-23#i_11761990 TimToady , I'm not sure why the UNDO phasers are triggered there?
nine m: my @a = 1, 2, 3; say [@a];
camelia rakudo-moar 214deb: OUTPUT«[1 2 3]␤»
14:37 Begi left
loren If i want do like that, why i use append instead 14:38
grondilu m: say grep 'append', Array.^methods
camelia rakudo-moar 214deb: OUTPUT«Method object coerced to string (please use .gist or .perl to do that) in block <unit> at /tmp/tc2dBx9Lhq:1␤Method object coerced to string (please use .gist or .perl to do that) in block <unit> at /tmp/tc2dBx9Lhq:1␤Method object coerced to string (p…»
loren why not ..
14:39 vv3 joined
Skarsnik damn even with self-update shell:::cmd is not found 14:39
Ven m: say grep 'append', Array.^methods.map(*.name)
camelia rakudo-moar 214deb: OUTPUT«(append append)␤»
Ven grondilu: ^^ 14:40
grondilu what's this append method? I can't find it in S32. 14:41
RabidGravy append is a "flattening" push
Ven append is like .push: Slip.new(...) 14:42
loren m: my @a = 1, 2, 3; my @x; @x.push: @a; my @y ; @y.append: @a; say @x; say @y;
camelia rakudo-moar 214deb: OUTPUT«[[1 2 3]]␤[1 2 3]␤»
14:42 vv3 left
grondilu ok 14:42
loren m: my @a = 1, 2, 3; my @x; @x.push: |@a; my @y ; @y.append: @a; say @x; say @y;
camelia rakudo-moar 214deb: OUTPUT«[1 2 3]␤[1 2 3]␤»
nine Skarsnik: are there traces of Shell::Command anywhere in your .rakudobrew?
Skarsnik it just does not get installed 14:43
hankache append @a is like push |@a
Skarsnik even running bootstrap.pl on panda did not work
loren Do i need report a bug about clone ?
nine If panda tells you that it successfully installed, then I believe it. The question is: where?
14:43 vv3 joined, vv3 left
Skarsnik with ack 14:44
1:unit module Shell::Command;
/root/.rakudobrew/moar-nom/panda/ext/Shell__Command/lib/Shell/Command.pm
only in this file
nine That's where it's installed from
14:44 vv3 joined, vv3 left
Skarsnik not in the install path 14:44
nine Skarsnik: you do this as root?
lucasb this dude keeps entering and leaving the channel... he must be waiting for some release announcement 14:45
14:45 vv3 joined
Skarsnik Yes, I work on root on this dumb vm x) 14:45
dalek kudo/nom: a860e8f | lizmat++ | src/core/Date (3 files):
Streamline Date handling

  - new()/gist moved to Dateish
  - main new logic now in .new($y,$m,$d) which allows for coercion
  - .new($str) no longer parses, but just splits and catches any fallout
  - no special format for small/large years
  - ACCEPTS now checks day/mon/year instead of year/mon/day
   since day is most likely to differ
  - DAYS-IN-MONTH is now a private method
  - no longer use check-date, but directly use Range,in-range
Skarsnik anyways, I am off for a while x) 14:46
nine loren: no, clone is fine as it is. It's just not the tool you need.
lucasb nine: sorry, but the clone behavior surprised me... idk if it's how it's supposed to work 14:47
loren nine, but it has problem obviously
hankache well from the docs it seems clone is defined in mu
nine loren: no, it doesn't! Just because you think or wish it would do something different, doesn't make it so. clone gives you a shallow copy of the object with the possibility of overriding some attributes. If you don't override, you get exactly the same attributes. Like Array's $!storage. 14:48
loren: So you get a new Array object that shares it's storage with the old one. That's why you see this behavior which is totally fine. 14:49
14:49 molaf joined
loren m: my @source = 1, 2, 3; my @display; @display.push: @source.clone(); @source[1,2]=@source[2,1]; say @display; 14:49
camelia rakudo-moar 214deb: OUTPUT«[[1 3 2]]␤»
loren nine, how you can explain this ? 14:50
14:50 vv3 left, raiph joined
nine loren: I just did. You're using a screwdriver for hammering in a nail and you're blaming the screwdriver for being a bad hammer. 14:50
14:51 vv3 joined, vv3 left
loren I know what am i doing , nine 14:52
FROGGS_ Skarsnik: I did not even add a test to github.com/rakudo/rakudo/commits/n...ple-args.t :o)
14:52 vv3 joined, vv3 left, FROGGS_ is now known as FROGGS, vv3 joined, vv3 left
nine lucasb: design.perl6.org/S12.html#Construct...ialization 14:53
loren: design.perl6.org/S12.html#Construct...ialization
14:53 vv3 joined
hankache nine: it is just that the word "clone" insinuates that you're (creating a copy) 14:54
14:55 cdg joined
nine hankache: and that's what it does. It creates a very shallow copy. So shallow, that even an Array's $!storage attribute is copied 14:55
hankache nine: indeed 14:57
lucasb nine: thanks for the link, I'll take a read at it later.
RabidGravy m: my @a = <1 2 3>; say @a.WHICH; @a.list>>.WHICH.say; my @b = @a.clone; say @b.WHICH; @b.list>>.WHICH.say 14:58
camelia rakudo-moar a860e8: OUTPUT«Array|56649600␤[IntStr|1 IntStr|2 IntStr|3]␤Array|56649728␤[IntStr|1 IntStr|2 IntStr|3]␤»
14:58 tipdbmp left, vv3 left
RabidGravy different arrays same containers as elements 14:58
simple
14:59 vv3 joined, vv3 left 15:00 vv3 joined, vv3 left, llfourn joined 15:01 vv3 joined, vv3 left
loren m: my @source = 1, 2, 3; my @display; @display.push: @source.clone(); .WHICH.say for @display; @source[1,2]=@source[2,1]; say @display;say @source.WHICH; say @display.WHICH; .WHICH.say for @display; 15:01
camelia rakudo-moar a860e8: OUTPUT«Array|50465856␤[[1 3 2]]␤Array|50465984␤Array|50465920␤Array|50465856␤»
15:01 vv3 joined
hankache see ya later #perl6 15:02
15:04 hankache left 15:05 llfourn left
loren I agree with your opinion about *clone*'s mean, and i read Array document carefully before i use it. but you see above result, i change '@source' but '@display' also changed! 15:06
15:06 vv3 left
geekosaur please rename "clone" since apparently people are going to trust the name over what it actually does 15:07
and inisit
*insist
15:07 ZoffixW joined
nine m: my @a = 1, 2, 3; my @b := @a.clone; push @a: 4; say @b; 15:07
camelia rakudo-moar a860e8: OUTPUT«[1 2 3 4]␤»
15:07 vv3 joined, vv3 left
nine RabidGravy: ^^^ I still think it's actually the sharing of the $!reified attribute that causes this 15:08
15:08 hankache joined
hankache back for a quick question 15:08
15:08 vv3 joined, vv3 left
ZoffixW loren, right, but display contains one element: the shallow clone of @source 15:09
RabidGravy nine, I think you're right
ZoffixW loren, and thus, by changing @source, you're changing it in @display too
hankache if i use module::xyz and this module has use MONKEY-SEE-NO-EVAL will i inherit it?
RabidGravy is it the repr that actually implements the clone
15:09 davercc left
nine hankache: no, MONKEY-SEE-NO-EVAL is lexical 15:09
loren ZoffixW, but Array address is not same.
ZoffixW loren, yes it is
hankache nine thank you :) 15:10
RabidGravy hankache, also a very polite author would put "no MONKEY-SEE-NO-EVAL" when they're done with it ;-) 15:11
nine m: use nqp; my @a = 1, 2, 3; say @a.WHICH; say nqp::getattr(@a, List, <$!reified>).WHICH; my @b := @a.clone; say @b.WHICH; say nqp::getattr(@b, List, <$!reified>).WHICH;
camelia rakudo-moar a860e8: OUTPUT«Array|49371680␤IterationBuffer|54820320␤Array|49371744␤IterationBuffer|54820320␤»
nine loren: ^^^
ZoffixW loren, err, I mean the addresses of stuff contained in the array
m: my @source = 1, 2, 3; my @display; @display.push: @source.clone(); @source[1,2]=@source[2,1]; .WHICH.say for @source; .WHICH.say for |@display[0]; 15:12
camelia rakudo-moar a860e8: OUTPUT«Int|1␤Int|3␤Int|2␤Int|1␤Int|3␤Int|2␤»
loren m: my @source = 1, 2, 3; my @display; my @t = @source.clone(); @display.push: @t; @source[1,2]=@source[2,1]; say @display;
camelia rakudo-moar a860e8: OUTPUT«[[1 2 3]]␤»
15:12 jczeus left, vv3 joined
loren ZoffixW, how about these ? 15:12
RabidGravy I'm all for slapping a big "THIS IS NOT WHAT YOU ARE LOOKING FOR" in Array.clone 15:13
nine Array doesn't even have its own .clone. It's Mu's which should already give you a hint that it's kinda low level.
hankache clone is defined in Mu
nine++ 15:14
RabidGravy so it's never going to work properly for Array really
hankache loren you can define a multi clone for arrays
RabidGravy or rather "as people think it should work"
15:14 vv3 left 15:15 vv3 joined, vv3 left
loren nine, i got what your point .. 15:15
nine loren: what's the problem with just using [@a]? 15:16
loren nine, i got your point ..
nine m: my @a = 1,2,3; my @b; @b.push: [@a]; @a[0,1] = @a[1,0]; say @b;
camelia rakudo-moar a860e8: OUTPUT«[[1 2 3]]␤»
nine [@a] is even shorter than @a.clone
hankache RabidGravy except for politeness forgetting no MONKEY-SEE-NO-EVAL is harmless, yes?
loren Em, nine, it's a good way 15:17
15:17 vv3 joined, vv3 left
ZoffixW I don't 15:18
15:18 vv3 joined
ZoffixW Nm 15:18
loren Last i want say 'using Perl 6' not go well ... 15:22
-_-
RabidGravy hankache, yes 15:23
15:23 vv3 left, khw joined
ZoffixW Well, no, I don't get it. I can repro that [@a] is not same as @a.clone 15:23
loren Many trap for me ..
ZoffixW And here's what I don't get:
m: my @one = 1, 2, 3; my @clone = @one.clone; my @two.push: @clone; say @one.WHICH; say @clone.WHICH; say @two[0].WHICH; @one[1,2] = @one[2,1]; say [ @one, @clone, @two[0] ]
camelia rakudo-moar a860e8: OUTPUT«Array|58993248␤Array|58993312␤Array|58993312␤[[1 3 2] [1 2 3] [1 2 3]]␤»
ZoffixW Here, changing the @one does not affect neither the clone nor the array with a clone
15:23 vv3 joined, vv3 left
ZoffixW m: my @one = 1, 2, 3; my @two.push: @one; @one[1,2] = @one[2,1]; @one.WHICH.say; @two[0].WHICH.say; [ @one, @two ].say 15:24
camelia rakudo-moar a860e8: OUTPUT«Array|60278704␤Array|60278704␤[[1 3 2] [[1 3 2]]]␤»
ZoffixW err
This one:
m: my @one = 1, 2, 3; my @two.push: @one.clone; @one[1,2] = @one[2,1]; @one.WHICH.say; @two[0].WHICH.say; [ @one, @two ].say
camelia rakudo-moar a860e8: OUTPUT«Array|59631600␤Array|59631664␤[[1 3 2] [[1 3 2]]]␤»
15:24 vv3 joined, vv3 left
ZoffixW Here, changing @one does affect the cloned one and if I rewrite @one.clone as [@one] the effect dissappears. 15:24
Or is this what you were talking about the $!reified stuff? 15:25
15:25 vv3 joined, vv3 left
loren ZoffixW, that's what i confused 15:26
15:28 vv3 joined
ZoffixW I think I get it: :push uses the address, thus modifications can be be made, but the first example creates a new array, thus modifications don't propagate 15:28
m: my @one = 1, 2, 3; my @two.push: @(@one.clone); @one[1,2] = @one[2,1]; @one.WHICH.say; @two[0].WHICH.say; [ @one, @two ].say
camelia rakudo-moar a860e8: OUTPUT«Array|76265504␤Array|76265568␤[[1 3 2] [[1 3 2]]]␤»
ZoffixW m: my @one = 1, 2, 3; my @z = @one.clone; my @two.push: @z; @one[1,2] = @one[2,1]; @one.WHICH.say; @two[0].WHICH.say; [ @one, @two ].say 15:29
camelia rakudo-moar a860e8: OUTPUT«Array|56014976␤Array|56015040␤[[1 3 2] [[1 2 3]]]␤»
stmuk lizmat++ # fixing dd
ZoffixW m: my @one = 1, 2, 3; my @two.append: @one.clone; @one[1,2] = @one[2,1]; @one.WHICH.say; @two[0].WHICH.say; [ @one, @two ].say
camelia rakudo-moar a860e8: OUTPUT«Array|63703072␤Int|1␤[[1 3 2] [1 2 3]]␤»
ZoffixW loren, so it's really the behaviour of .push at play, rather than that of .clone 15:30
15:30 vv3 left
lucasb after today's advent post, I'm considering taking a serious look at 007. masak++ :) 15:35
loren ZoffixW, so .clone is correct, .push cause that problem
ZoffixW loren, I think the real cause of the problem is our lack of understanding of what .clone does and returns :P 15:41
m: my @one = 1, 2, 3; my $clone = @one.clone; my @two.push: $clone; @one[1,2] = @one[2,1]; @one.WHICH.say; @two[0].WHICH.say; [ @one, @two ].say 15:42
camelia rakudo-moar a860e8: OUTPUT«Array|68298992␤Array|68299056␤[[1 3 2] [[1 3 2]]]␤»
loren ZoffixW, please explain to me .. 15:44
Ven m: my $a = [1, 2]; my @a = @$a; @a.push: 3; say $a; 15:45
camelia rakudo-moar a860e8: OUTPUT«[1 2]␤»
Ven m: my $a = [1, 2]; my @a = $a; @a.push: 3; say $a;
camelia rakudo-moar a860e8: OUTPUT«[1 2]␤»
15:48 skids joined
_nadim rindolf: same procedure, git pull ; t6 t/ please. I hope most is ironed out now 15:51
ZoffixW loren, you want the blind to lead the deaf? Very well then... Clone returns a shallow copy, so if you use it in a $ or with a .push (that takes "a" thing), the you get the contents of the array too. But if you assign it into @ or use append, it gets a new storage thing and thus modification is not propagated 15:53
jnthn afternoon, #perl6
yoleaux 22 Dec 2015 20:33Z <TimToady> jnthn: I think the answer to #126005 is to transform any loop that wants its return value into a map, including KEEP/UNDO, but also loops in non-statementlist context, as the specs ask for; that should give us the correct semantics for KEEP/UNDO and should allow for sink optimization of maps to throw away values later, hopefully after checking return for success
ZoffixW loren, which is why it was stated above, .clone doesn't work properly on Arrays and is not doing what one might think it should be doing.
FROGGS o/
yoleaux 14:08Z <Skarsnik> FROGGS: You forget to update the plan on t/04-nc/02.t to 13 and make the test succes on make test, when it fail for the 13th because size_t (and bool) are not listed on moar/src/core/nc.c
jnthn .tell TimToady See you fixed things up by now; I'd actually implemented Seq.from-loop while doing GLR prototyping 15:54
yoleaux jnthn: I'll pass your message to TimToady.
jnthn .tell TimToady So we'd thunk the various bits of the condition. But, so long as we get the semantics right for now, can do it that way :) 15:55
yoleaux jnthn: I'll pass your message to TimToady.
15:55 rurban left
jnthn is happy to see all the xmas RTs 15:56
uh, all the xmas RTs are fixed \o/
FROGGS damn, install_core_dist.pl infiniloops here
ZoffixW Wooohooo
jnthn++
jnthn Well, others++ took on the last couple
jnthn nailed the third to last :) 15:57
Then got sick.
loren ZoffixW, thanks for your explanation
ZoffixW :)
m: my @a = 1, 2, 3; say [@a.head, @a.tail]
camelia rakudo-moar a860e8: OUTPUT«[(1) (3)]␤»
ZoffixW m: my @a = 1, 2, 3; my $head = @a.head; my $tail = @a.tail; say [$head, $tail]
camelia rakudo-moar a860e8: OUTPUT«[(1) (3)]␤»
jnthn has one more change to sneak in :) 15:58
uruwi m: my @a = 1, 2, 3; say (@a.head, @a.tail).flat;
camelia rakudo-moar a860e8: OUTPUT«(1 3)␤»
FROGGS ahh, had to remove ~/.perl6/*
loren It's confusing me so much... 15:59
ZoffixW finally remembers what they were gumbling about a few weeks ago
m: my @a = 1, 2, 3; say [ @a.first, @a.last ]
camelia rakudo-moar a860e8: OUTPUT«Method 'last' not found for invocant of class 'Array'␤ in block <unit> at /tmp/fRMkqXSvNX:1␤␤»
ZoffixW m: my @a = 1, 2, 3; say [ @a.first, @a.tail ]
camelia rakudo-moar a860e8: OUTPUT«[1 (3)]␤»
ZoffixW Ah.. dreams
loren, .clone is a low level thing that you shouldn't be touching :) There! Now the confusion is gone :) 16:00
:P
loren So $ reference that Array, and a .push use $
RabidGravy I think the takeaway from the last two hours is that DWIM is not DWIW, DWIM has it's limits, treat apparently advanced functionality conservatively 16:01
ZoffixW ¯\_(ツ)_/¯ hence my blind leading the deaf comment :)
16:01 llfourn joined
ZoffixW m: my @a = ^3; my $a = |^3; say [ @a, $a ] 16:02
camelia rakudo-moar a860e8: OUTPUT«[[0 1 2] 0 1 2]␤»
loren ZoffixW, document not say what method can we use ....
ZoffixW doesn't get what's going on here
loren, do you know what "shallow clone" of the Array object is? 16:03
dalek kudo/nom: 29723e3 | FROGGS++ | lib/NativeCall.pm6:
map bool/size_t in nativecall to right sized int

This mean the don't have to change the NativeCall implementation in the backends, where bool is partially supported by dynall/libffi, and size_t is not. This makes the test pass in t/04-nativecall/02-simple-args.t and will hopefully aid DBDish::Oracle.
ZoffixW m: my @a = ^3; my $a = |^3; my $b = ^3; say [ @a.WHAT, $a.WHAT, $b.WHAT ]
camelia rakudo-moar a860e8: OUTPUT«[(Array) (Slip) (Range)]␤»
ZoffixW 0.o
dalek c: d8f36f3 | (Steve Mynott)++ | bin/p6doc-index:
fix p6doc CLI index generation
 16:04
c: ec2d529 | (Steve Mynott)++ | bin/p6doc:
use EVALFILE not EVAL
FROGGS .tell abraxxa please see github.com/rakudo/rakudo/commit/29723e31e7
yoleaux FROGGS: I'll pass your message to abraxxa.
loren ZoffixW, i misunderstand it ..
ZoffixW, sorry to everyone ..
ZoffixW m: my @one = 1, 2, 3; my $clone = @one.clone; say [ @one.WHAT, $clone.WHAT]
camelia rakudo-moar a860e8: OUTPUT«[(Array) (Array)]␤»
RabidGravy rebuilds rakudo *again* (third time today,)
rindolf _nadim: thanks! Let me see.
ZoffixW m: my @one = 1, 2, 3; my $clone = @one.clone; say [ @one.WHICH, $clone.WHICH] 16:05
camelia rakudo-moar a860e8: OUTPUT«[Array|70161760 Array|70161824]␤»
RabidGravy lorra, lorra modules to check in the next day or so
dalek osystem: fc09664 | thundergnat++ | META.list:
Regularize my modules to META6.json format
ZoffixW "lorra"?
m: my @a = ^3; my $a = [ ^3 ]; say [ @a, $a ] 16:06
camelia rakudo-moar a860e8: OUTPUT«[[0 1 2] [0 1 2]]␤»
ZoffixW still doesn't get what a "container" is 16:07
stmuk its like a beer can and beer
16:08 llfourn left
jnthn On latest, I get "Unknown type 'size_t' used for native call" 16:08
I thought I had latest of all the things...
rindolf _nadim: still fails here - www.shlomifish.org/Files/files/text...a-Dump.txt
ZoffixW @a = ^3; my $a = [ ^3 ]; say [ @a, $a ]
FROGGS jnthn: I just fixed it 16:09
RabidGravy ZoffixW, I think it's google "Cilla Black lorra"
jnthn FROGGS: ok :)
FROGGS jnthn: like, six minutes ago :o)
ZoffixW m: my @a = ^3; my $a = [ ^3 ]; my @b = $a; say [ @a, $a, @b, @a.WHICH, $a.WHICH, @b.WHICH ]
camelia rakudo-moar a860e8: OUTPUT«[[0 1 2] [0 1 2] [[0 1 2]] Array|79373904 Array|79373968 Array|79374032]␤»
RabidGravy it's a, er, cultural thing
ZoffixW lol
jnthn FROGGS: Ah, OK. :)
jnthn proceeds to spectest his change :)
loren $a is like a pointer, @a is a container ?
moritz @a is a compound type, $a is Scalar container 16:10
ZoffixW So @a is not even a container at all?
stmuk are these "[DISLOCATED MESSAGE]" intended to remain user faceing or are they internal debugging messages?
loren What is a scalar container ....
ZoffixW loren, docs.perl6.org/language/containers 16:11
ZoffixW re-reads that too
16:11 lustlife left
lizmat stmuk: they are bugs, afaik 16:11
loren ZoffixW, thanks 16:12
16:12 lustlife joined
moritz loren: it's a thing to hold another thing :-) 16:12
lizmat stmuk: as in: they shouldn't happen
loren moritz, -_-..
moritz, let me read document .. 16:13
Usually called a container, it's also be a pointer .. 16:14
FROGGS loren: a pointer is just one way to do it 16:16
16:18 cdg left
lucasb m: say (while $++ < 3 { 42 }) 16:18
camelia rakudo-moar 29723e: OUTPUT«(42 42 42)␤»
loren FROGGS, em, so it does.
lucasb m: say (repeat while $++ < 3 { 42 })
camelia rakudo-moar 29723e: OUTPUT«3␤»
lucasb m: say (repeat while $++ < 3 { 1+2 }) 16:19
camelia rakudo-moar 29723e: OUTPUT«sub infix:«<» (Mu $?, Mu $?) { #`(Sub+{<anon|67987232>}|50361816) ... }␤»
ZoffixW 0.o 16:20
dalek kudo/nom: d995c87 | coke++ | tools/build/gen-version.pl:
Update language version in prep for Christmas.
stmuk wooo 16:21
16:21 vv3 joined 16:22 virtualsue joined
RabidGravy grooviness 16:22
16:23 vv3 left
abraxxa FROGGS: re 16:23
yoleaux 16:04Z <FROGGS> abraxxa: please see github.com/rakudo/rakudo/commit/29723e31e7
lucasb m: say ({ 42 } while $++ < 3) 16:25
camelia rakudo-moar 29723e: OUTPUT«===SORRY!===␤Cannot find method 'CURSOR'␤»
timotimo m: my &metaassign = -> Mu \a, Mu \b { &[+]( (if a.DEFINITE { a } else { &[+]() }), b ) }; my $i = 0; metaassign($i, 1); say $i 16:26
camelia rakudo-moar 29723e: OUTPUT«0␤»
timotimo m: my &metaassign = -> Mu \a, Mu \b { a = &[+]((if a.DEFINITE { a } else { &[+]() }), b ) }; my $i = 0; metaassign($i, 1); say $i
camelia rakudo-moar 29723e: OUTPUT«1␤»
timotimo m: my &metaassign = -> Mu \a, Mu \b { a = &[+]((a.DEFINITE ?? a !! &[+]() ), b ) }; my $i = 0; metaassign($i, 1); say $i
camelia rakudo-moar 29723e: OUTPUT«Cannot modify an immutable Int␤ in block <unit> at /tmp/YXMmas40fS:1␤␤»
FROGGS lucasb: nice one
lucasb FROGGS: :)
timotimo ^- jnthn can i have a helping hand with this? i have no clue why it breaks :(
abraxxa why does rakudobrew build panda now by default? 16:28
16:28 nanis joined 16:29 mohae left
andreoss m: say(while $++ < 3 { LAST { $-- } }) 16:29
camelia rakudo-moar 29723e: OUTPUT«5===SORRY!5=== Error while compiling /tmp/YaLg3u3xLH␤Unable to parse expression in argument list; couldn't find final ')' ␤at /tmp/YaLg3u3xLH:1␤------> 3say(while 7⏏5$++ < 3 { LAST { $-- } })␤»
andreoss m: say(do while $++ < 3 { LAST { $-- } })
camelia rakudo-moar 29723e: OUTPUT«(Nil Nil Nil)␤»
16:29 RabidGravy left 16:30 loren left
jnthn timotimo: Did you manage to golf it down any? 16:30
nanis Just built 2015.11-750-ga860e8f with MS VS2015. When running `nmake test`, one test fails: 16:31
t\04-nativecall\02-simple-args.t with "Unknown type 'size_t' used for native call in method setup at D:\Src\rakudo\lib\NativeCall.pm6:223 in method CALL-ME at D:\Src\rakudo\lib\NativeCall.pm6:234 in block <unit> at t\04-nativecall\02-simple-args.t:66
hoelzro o/ #perl6
nanis Where are the known types deduced?
abraxxa nanis: does your build include github.com/rakudo/rakudo/commit/29...025e90d759 ? 16:32
nanis abraxxa I'll check, but FYI perl6::build-date=2015-12-23T16:05:07Z and I did a fresh clone, and nuked the install directory before building. 16:33
abraxxa nanis: i'm just building nom because of that, you can wait for my tests if you want\
timotimo jnthn: that was already a single golf step ... maybe i can make it even better
m: my $i = 0; $i = &[+](($i.DEFINITE ?? $i !! &[+]() ), 1 ) }; 16:34
camelia rakudo-moar d995c8: OUTPUT«5===SORRY!5=== Error while compiling /tmp/K1lpf8L_zI␤Unexpected closing bracket␤at /tmp/K1lpf8L_zI:1␤------> 3[+](($i.DEFINITE ?? $i !! &[+]() ), 1 ) 7⏏5};␤»
timotimo m: my $i = 0; $i = &[+](($i.DEFINITE ?? $i !! &[+]() ), 1 );
camelia ( no output )
timotimo m: my $i = 0; my \a := $i; a = &[+]((a.DEFINITE ?? a !! &[+]() ), 1 );
camelia rakudo-moar d995c8: OUTPUT«Cannot modify an immutable Int␤ in block <unit> at /tmp/fa4EHYYBqk:1␤␤»
timotimo m: my $i = 0; my \a := $i; a = 2;
camelia ( no output )
nanis abbraxxa: Good call. I missed that commit by a few mintues. I just wanted to see if things would build with VS 2015, and they did.
timotimo m: my $i = 0; my \a := $i; a = a + 1;
camelia ( no output )
timotimo m: my $i = 0; my \a := $i; a = (True ?? a !! 2) + 1; 16:35
camelia ( no output )
abraxxa FROGGS: i've built nom twice and every time on perl6 -V I get Unhandled exception: Missing or wrong version of dependency 'gen/moar/stage2/QRegex.nqp' (from 'src/Perl6/Pod.nqp')
timotimo m: my $i = 0; my \a := $i; a = (a.DEFINITE ?? a !! 2) + 1;
camelia rakudo-moar d995c8: OUTPUT«Cannot modify an immutable Int␤ in block <unit> at /tmp/Bo5OlxBTiW:1␤␤»
stmuk abraxxa: it doesn't but it reinstalls modules previousily installed
timotimo jnthn: is that weird enough for you? :D
abraxxa stmuk: ah, great!
timotimo m: my $i = 0; my \a := $i; a = 2 + 1;
camelia ( no output )
timotimo m: my $i = 0; my \a := $i; say a.DEFINITE;
camelia rakudo-moar d995c8: OUTPUT«True␤»
abraxxa i nuked it on the second attemplt but still this error
timotimo m: my $i = 0; my \a := $i; say a.DEFINITE; a = 5;
camelia rakudo-moar d995c8: OUTPUT«True␤Cannot modify an immutable Int␤ in block <unit> at /tmp/mhZiivNy9a:1␤␤»
timotimo m: my $i = 0; my \a := $i; a = 5;
camelia ( no output )
timotimo okay, that's really a little bit weird.
just calling .DEFINITE on it makes it unassignable? 16:36
hm. perhaps i see the error
nanis Thanks everyone, and good luck. Looking forward to Christmas. Bye.
16:36 nanis left
abraxxa maybe because I run rakudobrew in a nom checkout? 16:37
timotimo um... am i seeing this right? 16:38
abraxxa yes, that was the problem
timotimo no, i wasn't
jnthn timotimo: Yes, that's odd
m: my $i = 0; my \a := $i; a = a + 1; 16:39
camelia ( no output )
jnthn m: my $i = 0; my \a := $i; a = (a.DEFINITE ?? a !! a) + 1;
camelia rakudo-moar d995c8: OUTPUT«Cannot modify an immutable Int␤ in block <unit> at /tmp/gHAFTGeyl9:1␤␤»
abraxxa FROGGS: same error
jnthn m: my $i = 0; my \a := $i; a = (a ?? a !! a) + 1;
camelia ( no output )
16:39 dakkar left
timotimo jnthn: i expect it's because the p6definite code deconts over the register that the result is in 16:39
16:39 kanishka left
jnthn timotimo: Yes, that's what I'm suspecting 16:39
That's naughty.
And wants fixing 16:40
dalek kudo/nom: bcf63ef | jnthn++ | src/core/Promise.pm:
Promise.[anyof|allof] no longer break.

They now only care about promises completing *somehow* rather than about the nature of that completion. For `anyof` you'll be doing some follow-up work to get a result anyway, and when you care about getting all of a set of results, you just use `await` with a list of promises. This makes anyof/allof a little easier to work with for their common uses, which is waiting until one/all of a set of things have taken place, when you'll then go on to be interested in them individually.
timotimo it is, eh? but why does that influence what a is after the fact?
jnthn timotimo: 'cus a is lowered to a local
timotimo oh, damn.
jnthn I'll have a look at it now
timotimo that'd be it, then.
oh, you will?
neat :)
hankache what's up everybody 16:41
timotimo i'll rewrite the piece of the optimizer that triggers on METAOP_ASSIGN and the METAOP_ASSIGN implementation to be ready when you are done :)
TimToady is up
yoleaux 15:54Z <jnthn> TimToady: See you fixed things up by now; I'd actually implemented Seq.from-loop while doing GLR prototyping
15:55Z <jnthn> TimToady: So we'd thunk the various bits of the condition. But, so long as we get the semantics right for now, can do it that way :)
timotimo welcome back, TimToady
16:41 RabidGravy joined
RabidGravy ooh 16:41
timotimo hey gravy 16:42
RabidGravy Rarr!
FROGGS abraxxa: what error exactly?
abraxxa: ahh, so you're still not able to build rakudo? 16:43
dalek ast: d1fe0c9 | jnthn++ | S17-promise/a (2 files):
Chase anyof/allof semantic changes.
abraxxa FROGGS: sorry, my fault, hadn't commented out my size_t constant definition 16:44
16:44 mohae joined
abraxxa as a type check become more strict? I'm getting a failure which I hadn't before 16:45
Type check failed in assignment to $col_namep; expected NativeCall::Types::CArray[NativeCall::Types::CArray[int8]] but got NativeCall::Types::CArray[NativeCall::Types::CArray[int8]].new
in block at /home/ahartmai/perl6/git/DBIish/lib/DBDish/Oracle/StatementHandle.pm6:494
the line in question is: my CArray[CArray[int8]] $col_namep .= new;
timotimo abraxxa: maybe it's not able to .new a nested CArray properly? 16:47
jnthn timotimo: Building a patch now.
RabidGravy tthis may be a bug abraxxa
abraxxa timotimo: worked with 2015.11
timotimo huh.
lucasb TimToady: when you have time, can you look at the 'while', 'repeat while' and 'loop' list comprehensions in the backlog? thanks 16:48
16:49 mohae left
jnthn timotimo: Yeah, it was that decont screwup 16:49
flussence mornin', #perl6 16:50
jnthn FROGGS: Confirm 02-simple-args.t fixed on latest \o/
ZoffixW \o
FROGGS \o/
jnthn timotimo: Can you add a regression test of the cut-down example once I push the patch?
FROGGS hi ZoffixW
timotimo jnthn: i feel silly now that it only took like a minute to pinpoint the problem after i started bothering you; i could totally have done that on my own
16:50 mohae joined
timotimo surely can! are regression tests without a RT number okay, though? :P 16:50
jnthn Sure :) 16:51
dalek kudo/nom: 86c1009 | jnthn++ | src/vm/moar/Perl6/Ops.nqp:
Fix p6definite code-gen bug found by timotimo++.
FROGGS lucasb: can you rakudobug the one-liner you posted?
jnthn Didn't spectest that...lol yolo
(passes make test though)
FROGGS hehe
timotimo does make test even have DEFINITE in it? :) 16:52
jnthn Will see if I can fix something else to get two patches for my spectest time :)
I'm not definite it does...
I checked it fixed your thing though :)
lucasb FROGGS: I'm shy to write emails to rakudobug :) But TimToady will see the backlog later... 16:53
nine jnthn: easy! Fix run_alt and become the hero of a generation! Oh wait...you already are
16:53 mohae left
FROGGS lucasb: just write your one-liner and the output to [email@hidden.address] you dont have to add anything to that message :o) 16:54
TimToady will certainly see the backlog, but must reserve time today to write advent post, so no promises on fixing
FROGGS TimToady: I've got a patch already I think
16:54 Ven left
TimToady ooh, and I don't even know what the bug is yet :) 16:54
FROGGS m: say ({ 42 } while $++ < 3) 16:55
camelia rakudo-moar bcf63e: OUTPUT«===SORRY!===␤Cannot find method 'CURSOR'␤»
FROGGS just a one-line patch also
TimToady heh, ok
lucasb m: say (while $++ < 3 { FIRST { 42 } })
camelia rakudo-moar bcf63e: OUTPUT«===SORRY!===␤Cannot find method 'CURSOR'␤»
lucasb idk how these list comprehensions are supposed to deal with loop phasers... 16:56
idk FIRST, LAST, etc. is called a loop phasers, however :) 16:57
TimToady well, it's still in a loop, just a different kind of loop (for now)
16:57 ZoffixW left
jnthn nine: What's the easiest known reproduction of alt_nfa? 16:58
TimToady we might have an extra level of {} snuck in there to mess up things like FIRST
dalek ast: 795db8d | timotimo++ | S12-introspection/definite.t:
regression test for DEFINITE code gen bug

where a decont was accidentally clobbering the local register that contained a variable that we wanted to write to later.
16:59 Deepak_ joined
nine jnthn: mid November. But it can easily be much older than that. The bisect stopped at a totally unrelated commit 17:00
uruwi Anyone know why this is segfaulting? pastebin.com/THzTTAFP
If you ask, then I'm willing to give read access to the full repository.
(line 58 in the paste) 17:01
17:02 mohae joined
nine jnthn the information I've gathered so far points at a precomp related corruption of an NQP meta model cache 17:02
jnthn nine: Yeah, sounds like... 17:04
Or sounds like a reasonable hypothesis. I've no idea how it could be happening.
Is there a reliable way I can reproduce it? 17:05
dalek ast: 60d1ad6 | timotimo++ | S03-operators/assign.t:
test that += on a Failure dies
 17:06
kudo/nom: 587a801 | timotimo++ | src/ (2 files):
Optimize METAOP_ASSIGN again
 17:07
17:07 tipdbmp joined
uruwi Anyone know why this is segfaulting? pastebin.com/THzTTAFP 17:08
jnthn uruwi: Not from just looking at the code, now...
*no
17:09 Begi joined
uruwi Do you want to try running it yourself? 17:09
(the full program I mean, not just that file)
jnthn Given it has a mass of dependencies, not especially...
RabidGravy uruwi, or just make the smallest possible example that still segfaults 17:10
jnthn If you can run it under one of perl6-gdb or perl6-valgrind it may produce something I can make a good guess from.
uruwi perl6-gdb-m output pastebin.com/bKeRexiF 17:11
The error message refers to some file or directory not exiting 17:12
lucasb is the locale set to japanese?
uruwi * existing
17:12 mohae left
uruwi Yes 17:12
Skarsnik Let's see if panda work now ~~
jnthn uruwi: can you type "bt" and press enter and see if it produces any more output?
lichtkind is 25 dez still standing as date? 17:13
uruwi pastebin.com/DGgU4G92
timotimo jnthn: probably b0rked output thanks to the jit :)
oh, look, it's the mutex destruction bug
didn't we put in a workaround for that, though?
jnthn Yeah, it's the free of a held mutex one :( 17:15
Not sure if we did nor not
*or
uruwi Link to somewhere mentioning it? 17:16
17:16 mohae joined 17:17 Deepak_ left, espadrine left
jnthn nine: I tried the catdir thing in rt.perl.org/Ticket/Display.html?id=126832 and couldn't reproduce the run_alt thing 17:19
abraxxa RabidGravy, timotimo: can someone confirm that this is a bug?
jnthn uruwi: It's been discussed a bit on #moarvm
17:21 muraiki joined
Skarsnik panda still can't find shell::comand, it does not get installled >< 17:21
17:21 andreoss left
RabidGravy abraxxa, it's a regression related to precomp can't reproduce without putting code in a module, don't know if it's RTd 17:21
abraxxa RabidGravy: should I open an RT for it? 17:22
timotimo abraxxa: sorry, what now?
oh, the CArray[CArray[...]] thing?
abraxxa timotimo: yes
i don't find a test for that
timotimo seems buggy, yeah. are you interested in bisecting rakudo to figure out where it started happening?
abraxxa is it even NativeCall specific of all nested Types? 17:23
17:23 mohae left
abraxxa m: my Array[Array[Str]] $foo .= new; 17:23
camelia ( no output )
abraxxa m: use NativeCall; my CArray[CArray[int8]] $foo .= new;
camelia ( no output )
RabidGravy timotimo, abraxxa 's example overcomplicates it, in a precomped module a typeconstraint with a typed CArray fails 17:24
it didn't used to
Skarsnik duh
timotimo "in a precompiled module" ;_;
RabidGravy i.e 17:25
abraxxa add precompilation so close to the release isn't a wise decision imho as it's not changing the API but an optimization only that can be added later
timotimo it changed a lot of api, actually 17:26
RabidGravy m: use NativeCall; sub foo() returns CArray[iuint8] { return CArray[uint8].new }; say foo()
jnthn Indeed.
camelia rakudo-moar 587a80: OUTPUT«5===SORRY!5=== Error while compiling /tmp/1r8O6do6Nb␤An exception occurred while parameterizing CArray␤at /tmp/1r8O6do6Nb:1␤Exception details:␤ 5===SORRY!5=== Error while compiling ␤ Cannot invoke this object␤ at :␤»
timotimo well, it was part of the whole curli refactor which changed a lot of api
Skarsnik gah now shell::command is installed but not precompiled and panda still does not find it ><
perlpilot abraxxa: +1 (and all of the .precomp dirs littered everywhere)
RabidGravy m: use NativeCall; sub foo() returns CArray[uint8] { return CArray[uint8].new }; say foo()
camelia rakudo-moar 587a80: OUTPUT«NativeCall::Types::CArray[uint8].new␤»
timotimo in theory, we could release a rakudo completely without precompilation
abraxxa timotimo: even worse!
17:27 boegel|quassel is now known as boegel, boegel left
RabidGravy that fails if it's in a module that has been precompiled 17:27
17:27 boegel joined
abraxxa did 2015.11 have precomp? I've never seen the .precomp dirs before 17:27
jnthn Precomp has been around for ages 17:28
abraxxa i've removed the .= new; as a workaround
jnthn But was handled at install time
By Panda
abraxxa i see
jnthn More recently it's been moved in Rakudo and made a bunch more robust
abraxxa that will root installed, non precompiled code fail
17:29 raiph left
abraxxa WTF 17:29
Type check failed in assignment to $col_namep; expected NativeCall::Types::CArray[NativeCall::Types::CArray[int8]] but got Array
Skarsnik it's normal that install/share/perl6/site/lib/ is not in where panda search module?
jnthn Yes, of course it would have been nice if we'd had precomp stuff in Rakudo landed earlier.
Skarsnik You need the .= new sadly
RabidGravy yeah, it was at least 14 days ago github.com/jonathanstowe/Audio-Enc...a8d928d034 17:30
jnthn But that's not really how things work.
17:30 mohae joined
timotimo dinner time \o/ 17:30
jnthn We make the best effort we can with the limited resource we have.
abraxxa it's the only point where I need a pointer to a pointer 17:31
RabidGravy exactly
[Coke] sees catdir in backscroll and reads catbug
dalek kudo/nom: 82e6ca2 | jnthn++ | src/Perl6/Grammar.nqp:
Fix errors on unknown symbols in constants, enums.

Rids us of a copule of cases of "Cannot invoke this object".
flussence
.oO( nobody from outside showed up for the beta release, so we're tricking them into beta-testing now :)
RabidGravy well even if more people here had made more effort to keep up 17:32
dalek ast: a4ace08 | jnthn++ | S32-exceptions/misc.t:
Tests for RT #126987.
Skarsnik who/what could I blame? I did rakudobrew nuke && rakudobrew build moar. panda does not work. I even rerun the bootstreap.pl
17:35 mohae left
moritz sighs. Not again, this close before the release 17:37
dalek osystem: 859d327 | PerlJam++ | META.list:
Add Test::Class
 17:38
Skarsnik let's try on e anew user
17:39 mohae joined
leedo fails here as well, with a lot of EVAL is dangerous errors 17:42
that is running an existing panda install, and bootstrapping
17:42 mohae left
jnthn Haven't done a Panda bootstrap in quite a while in this laptop. Seems to have completed fine on Rakudo HEAD. 17:43
Skarsnik just wanted to patch some modules and write some user code today
leedo i'll try nuking install, and redoing it
jnthn runs panda install HTTP::UserAgent to see if he can make the run_alt thing appear that way 17:45
RabidGravy jnthn, that was "fixed"
jnthn RabidGravy: "fixed" as in "magically vanished"? :) 17:46
Skarsnik no precompil
in the code x)
RabidGravy no precompilation
jnthn oh
leedo nuking install seems to let panda install here :)
dalek p: c353663 | (Zoffix Znet)++ | README.pod:
Fix 404 URLs
p: a1339e2 | lizmat++ | README.pod:
Merge pull request #264 from zoffixznet/fix-url

Fix 404 URLs in README.pod
jnthn But it reliably produced the error without "no precompilation"?
Skarsnik Not cool to have essential module broken x) 17:47
RabidGravy but if you were to remove them and run the t/030-cookies.t they would reappear 17:48
jnthn OK
RabidGravy yes
jnthn Getting headachey again, so probably time to take a break for now
Skarsnik Oh I found what is wrong 17:49
RabidGravy there is an RT with a very small replication I think
nine jnthn: sorry, was on my way home 17:50
jnthn: run_alt has become more difficult to reproduce in the past week
17:51 mohae joined
jnthn RabidGravy: Yes, I tried the small replication and couldn't replicate it.. 17:51
RabidGravy: Thus why I'm onto the larger one now
Skarsnik rakudobrew nuke did not nuke panda repository so panda did not fetch/build/install modules like Shell::Commad did not get properly reinstalled
RabidGravy hmm
jnthn Alas, 03-cookies.t passed
RabidGravy strange 17:52
nine jnthn: yes, my checkout from Dec 13 also passes :/
Skarsnik so shell::comand did not get precompiled and the precompiled file install
jnthn nine: Aww
Skarsnik ah yes and I can't install XML ~~
who can merge PR in XML?
jnthn OK, time for a break. bbl
RabidGravy :)
nine Skarsnik: at this point I consider rakudobrew to be actively harmful :/ 17:53
17:54 rurban joined
flussence has always built rakudo "by hand" with a shell script that nukes everything using a «find -exec git-clean», and missed out on all this fun people seem to have with rakudobrew :( 17:54
Skarsnik want to fork XML in perl6
lucasb for the record, I too brew my own rakudo :) 17:55
17:57 mohae left
lucasb my adhoc shell script has 'rm -rf ~/.perl6' in it, and it dies if it finds PERL6LIB set in the environment 17:57
^^ these things have made the thing sane for me
muraiki I've always used rakudobrew :( 17:58
stmuk I've found rakudobrew nuke and rm -rf ~/.perl6 fine 18:01
Skarsnik There is no way to put a p5 module in the depends in the Meta file?
flussence nope 18:02
nine Skarsnik: not yet anyway
Skarsnik I think it's something needed to at least have panda warn that something additionnal is needed ~~
nine Skarsnik: we can do better than that :) 18:03
abraxxa i'm annoyed by the breakage every time I touch rakudo especially this close to its first release, maybe I'm coming back in half a year or so
Skarsnik especially for DBIish that has not obvious library name
you search for a libmysql, not libmysqlclient. libpostgressql? not libpq xD 18:04
18:04 jevin joined
Skarsnik s/not/nop/ 18:04
18:05 mohae joined
Skarsnik nine, depend what you can more? if you mean for p5 calling cpan.... I personnaly don't want that. I want the debian package (or build the package with dh_make) installed 18:06
*mean
dalek p: 23538e5 | (Zoffix Znet)++ | docs/ops.markdown:
Add Table of Contents for easier navigation
 18:08
p: 9d05ece | lizmat++ | docs/ops.markdown:
Merge pull request #265 from zoffixznet/ops-toc

Add Table of Contents to docs/ops for easier navigation
nine Skarsnik: like I told AlexDaniel today: what you want is not nearly as important as what you're gonna do about it ;)
Skarsnik And now I need to edit all the NC doc because of you x) 18:11
nebuchadnezzar I read the 5to6-nutshell and I was surprise by the module import, I had the perl5 habit to always list what I'm using and with perl6 I can do it only if the module writer planned to put some tags 18:13
it empoyer module writers and not module users :-/
lucasb nqp/ops.markdown already had a little ToC. now it has 2. maybe the earlier can be removed now?
*the earlier one 18:14
18:14 mohae left
_nadim So i found time between two Christmas dishes preparation to update Rakudo (600+ commits), it broke a thing or two that were surprising but what I liked most is the speed boost. I haven't checked but Imaybe two or three times faster. Maybe precomp has something to do with it. Great work all of you, so far I am enjoying it more than it frustrates me. 18:15
18:15 smls joined
timotimo _nadim: it's probably a stupid performance regression i introduced in the JIT compiler that was completely killing the multi-dispatch cache 18:16
it was recently fixed, so maybe that's the difference you saw
18:16 mohae joined, aindilis` left 18:17 dwarring joined
Juerd Feedback wanted on gist.github.com/Juerd/ae574b87d40a66649692 an idea for an "unpack" replacement. 18:17
18:17 ZoffixW joined
ZoffixW lucasb, I think it's fine to keep the old one in. I kinda see it more as a list of categories (that are also links) than necessarily a TOC 18:17
18:17 virtualsue left
timotimo Juerd: we don't have anything that can determine the position of a named argument 18:18
Juerd timotimo: These aren't named arguments
timotimo: I'll add that
abraxxa is t/04-nativecall/06-struct.t a good place for a CArray[CArray[int8]] test? 18:19
18:19 aindilis joined
lucasb ZoffixW: ok, both can stay 18:19
nine nebuchadnezzar: yes that's one of the very few changes that I'm not comfortable with
dalek c: 78410e8 | (Sylvain Colinet)++ | doc/Language/nativecall.pod:
Fix all the libfoo to foo in NativeCall documentation (thank to nine x))
c: f23805a | (Sylvain Colinet)++ | / (19 files):
Merge branch 'master' of github.com/perl6/doc
18:19 xtreak joined, mohae left
Juerd timotimo: I've added that the signature is a mere *@foo 18:19
Skarsnik git I hate you, stop doing useless merge commit ><
El_Che that feeling when your .t is bigger than your .pm6 18:20
mst Skarsnik: rebase moar
ZoffixW :D
Juerd El_Che: Write more POD :)
ZoffixW Skarsnik, git pull --rebase ?
mst El_Che: that's the feeling that you might -just- be approaching having written enough tests ;) 18:21
timotimo Juerd: if it's a sub, that'll throw an error, if it's a method, it'll silently swallowing the named arguments :(
El_Che mst: I ended rewriting some part of the pm for corner cases I hadn't think of
so yeah for testing 18:22
Juerd m: class X { method y (*@foo) { say +@foo } }; X.new.y(pair => 1, :pair(2), :3pair)
camelia rakudo-moar 82e6ca: OUTPUT«0␤»
Juerd Daamn
ZoffixW El_Che, I had 65 lines of tests for a single letter of code :P here github.com/perl6/modules.perl6.org...b080f95e8b
_nadim timotimo: In aby case, Great job. 18:23
Juerd timotimo: Have to wrap everything in an array then. Ugly :(
nebuchadnezzar nine: I made some tests with two modules defining a sub with same name, “use Module1” followed by “use Module2” will happily override things without any warning :-/
El_Che ZoffixW: hardcore
Juerd timotimo: Thanks for the heads up though. I was completely unaware of this.
nebuchadnezzar nine: so, if I do not take care, I can finish by using a sub from Module2 where I want the one of Module1
s/of/from/ 18:24
smls "my @i = $blob.decode(:big, uint32, uint16, uint8);" looks pretty neat. Juerd++.
hankache yahou! That's the first time i can update rakudo without nuking everything
This is Rakudo version 2015.11-756-g82e6ca2 built on MoarVM version 2015.11-113-gbd56e2e
implementing Perl 6.c.
ZoffixW \o/
timotimo www.reddit.com/r/perl/comments/3xt...se/cy80n7e - here's new measurements for +=
El_Che hankache: what the fun in updating without nuking stuff?
nine didn't have to nuke for weeks 18:25
smls 6.c has been released?
timotimo Juerd: thanks for taking the time to write up a proposal :)
hankache why does it say 6.c?
nine hankache: because we're gonna release 6.c quite soon
ZoffixW smls, no, just the version bumped github.com/rakudo/rakudo/commit/d995c87b51
smls ok
ZoffixW hankache, preparing for release 18:26
hankache aha ok
18:26 ZoffixW left, mohae joined
smls Is there a countdown or something? :) 18:26
hankache I wouldn't want to miss it
Juerd smls: Refresh the page, it looks a little less neat now
Skarsnik Juerd, if you make this type usable with NC, you are my hero x)
Juerd Skarsnik: What do you mean by 'usable with'? Also, I'm not a C coder so probably not... 18:27
timotimo: I've updated the gist to include [] around every template
That does have the advantage of being able to use multi methods to get both .decode("utf8") and .decode([...]) 18:28
smls still looks better than unpack :) 18:29
Juerd Agreed 18:30
18:30 mohae_ joined
Juerd And thanks :) 18:30
18:31 mohae left
lizmat m: (^61).ACCEPTS(60.9).say # this feels wrong 18:31
camelia rakudo-moar 82e6ca: OUTPUT«True␤»
lizmat ah, but of course, it isn't
Juerd Hi lizmat. Do you have time to look at my unpack proposal? 18:32
Skarsnik Well we need a type to catch char* from C (0 terminated) as raw data
lizmat I scanned it...
but haven't thought much about it yet
Skarsnik and not encoded in a Str
18:33 xtreak left
Juerd lizmat: That's okay. At least you didn't hate it then :) 18:33
Skarsnik: Blob?
flussence random thought: could we convince .{/de|en/}code('foo') to automatically use an Encode::foo module that's in-scope? would provide future expansion for string charsets as well as any other craziness people dream up. 18:35
Juerd flussence: I hope so, because I think that things like hex and binary are string encodings. 18:36
flussence: And there are more things like that, like base64...
Bit unwieldy to have those all in the mechanism for text encodings.
muraiki I missed the earlier context for the rakudobrew discussion so I don't know if this is a new thing or not, but the optimizing part of the build is failing
Stage mast : MVM_platform_alloc_pages failed: 12
El_Che is there a difference between die die X::AdHoc.new(:payload<foo>).throw and die X::AdHoc.new(:payload<foo>) ? or just syntax? 18:37
18:38 stanrifkin joined
nine El_Che: if you call .throw you don't need a die 18:38
lizmat die(Exception) basically does a Exception.throw afaik 18:39
chansen_ Juerd: The syntax seems very verbose compared to the "traditional" pack/unpack template
Juerd chansen_: That is definitely true. 18:40
chansen_: Do you see this as a good thing or a bad thing?
timotimo muraiki: that looks a bit like you're running out of ram perhaps?
muraiki whoa. someone lowered the ram on my development VM to 2GB T_T 18:41
18:41 smls_ joined, smls left
abraxxa is t/04-nativecall/06-struct.t a good place for a CArray[CArray[int8]] test? 18:41
i'm adding it to 05-arrays.t
timotimo muraiki: damn, 2gb isn't enough to build rakudo? :(
abraxxa it was on my raspberry pi 18:42
240mb ram + 1gb swap file
at least 2015.09 or 10
flussence my netbook has 1GB and builds rakudo fine...
muraiki timotimo: I've got about 1.1GB free
hmm
I'm going to try again, but only building moar manually 18:43
abraxxa how to run tests in a checkout? 18:45
RabidGravy abraxxa, that sounds too specific a test, as I said earlier it's all typed CArrays yhat fail typeconstraints
abraxxa I always get Unhandled exception: Missing or wrong version of dependency 'gen/moar/stage2/QRegex.nqp' (from 'src/Perl6/Pod.nqp')
chansen_ Juerd: I guess it depends who you ask! I'm used to the concise pack/unpack template, but your syntax looks very clear and extensible!
abraxxa RabidGravy: better one failing test than none 18:46
lizmat Juerd: far from it
abraxxa RabidGravy: all other CArrays work for me, only CArray[CArray
nine abraxxa: ./perl6-m instead of just perl6
RabidGravy have you got a failing example outside your code?
abraxxa nine: thanks 18:47
Juerd chansen_: Well, I was asking you. Thanks for the feedback :)
abraxxa why does the test pass?
chansen_ Juerd: Why not propose it as an alternative instead of a replacement? 18:48
RabidGravy what test?
Juerd chansen_: Because the current .unpack is very incomplete and impossible to extend while keeping backwards compatibility.
abraxxa RabidGravy: ok my CArray[CArray[int8]] $pointer-to-string .= new, 'can instantiate CArray[CArray[int8]]';
muraiki building moar on its own was ok. so I tried rakudobrew again and during the optimize stage saw my available memory rapidly declining to nothing 18:49
abraxxa yes, lets change the whole pack/unpack syntax one day before release!
Juerd chansen_: And I don't think there's anything to gain by having both, to be honest
RabidGravy as I said earlier as well it only fails in precompiled modules
abraxxa isn't there a code freeze policy?
muraiki moar was using 99.2% CPU and 59.1% memory.
Juerd abraxxa: Nope, I don't see this as an immediate change, but I do hope that pack/unpack get marked experimental
abraxxa Juerd: don't you think you're a bit late? 18:50
nine .tell jnthn fde3d620bf1b22f1b78420e1294a04c663145efe is the last commit with which I can repro run_alt. With b6f0532c2b6fcf95b65f1de63537e91821a123fd it disappeared again. Didn't have to downgrade nqp for that.
yoleaux nine: I'll pass your message to jnthn.
Juerd abraxxa: I'll let other people decide that
lizmat Juerd: ah... eh... hmmm... 18:51
Juerd abraxxa: I've contributed to Perl 5 long after its first release and wasn't too late :)
nine Well if our unpack sucks (and I don't remember anyone working on it in the past year), marking it experimental seems like a good idea.
flussence chansen_: besides, if we get this verbose stuff in, it'd be trivial to write a Pack::Terse (or *) module that expands the p5 notation to it 18:52
Juerd It most definitely suck. I'm glad it's there, but it's not ready for prime time.
s/suck/sucks/
RabidGravy unpack, works for me and I have made code with it
chansen_ Juerd: OK, I guess it's better with your verbose syntax that can be extended. We could release a simple wrapper module that provides the concise template API using the new syntax
flussence: ^^
abraxxa is perl6 searching in ./lib for libs? 18:53
flussence not by default
Juerd RabidGravy: It ignores unknown characters in the template, which means it can't be extended. And it has some abstraction breakage in a*, because it deals with Str where that should be Blob.
abraxxa that was removed from Perl 5 because it's a security risk
RabidGravy but that's fine I'll just stop making modules until everything is perfect ;-p
Juerd RabidGravy: These things can't be fixed without breaking compatibility
abraxxa so why does running my rakudobrewed perl6 fail when run from a rakudo git clone then?
Juerd RabidGravy: If we mark it as experimental, we can keep the current implementation and use that until we have something better.
RabidGravy: But then at least users are warned that things will probably change. 18:54
lizmat Juerd: how does your idea conflict with pack/unpack?
abraxxa i thought the whole API was though out and discussed for FIFTEEN YEARS?
and now unpack/pack sucks?
who wrote the synopsis for it and who agreed that it's good?
lizmat abraxxa: it's a P5 API, really
Juerd abraxxa: pack and unpack templates aren't specced. 18:55
abraxxa lizmat: which seems to handle everything you throw at it fine
Juerd abraxxa: Even if you think Perl 5 pack/unpack templates are great and should be kept, then still the ones we have now in Perl 6 are incompatible and cannot be made compatible without breaking compatibility.
lizmat abraxxa: S32/Str:503 18:56
abraxxa so why not keep it, marks it as 'like Perl 5 to get going without learning anything new' and add something easier on the brain with a different name?
lizmat =item unpack
B<XXX To be defined>
RabidGravy I've got a working xbase implementation here, which uses unpack
abraxxa great, so someone implemented something unspeced
lizmat abraxxa: hardly 15 years of thinking and speccing
abraxxa goes reading Python 3 docs
Juerd abraxxa: Well, I don't think anyone can actually implement a Perl5-like pack/unpack in the few days we've got left until Christmas... :)
18:56 cdg joined
flussence abraxxa: you've had 15 years to complain about it and you wait until *now*? 18:56
lizmat Juerd: I actually started on that but then the flu hit :-( 18:57
smls_ I wouldn't mind it being marked experimental for now, like the 'is cached' trait was.
Juerd lizmat: I got struck last week. Three days in bed... And now I'm procrastinating dayjob-things because it's too much to do before I'm leaving on a christmas vacation... 18:58
lizmat Juerd: sounds familiar
lucasb I'm on a brand new rakudo and panda hangs in '==> Installing Bailador'
Juerd Heck, the whole .decode to replace .unpack proposal was pure procrastination (from sleeping, though)
chansen_ flussence++ # Nice counter comment ;o) 18:59
stmuk lucasb: I just saw that issue
muraiki I was able to make it past optimization stage in rakudobrew . it looks like about 1.1GB of memory was in use, which is about how much free memory I had on the vm where rakudobrew ran out of memory. unfortunately the build then died at install-core-dist.pl with "An exception occurred while evaluating a constant" 19:00
lucasb stmuk: thanks for confirming it!
abraxxa RabidGravy: the test I wrote is ok but it fails in my code, no idea why
m: use NativeCall; my CArray[CArray[int8]] $pointer-to-string .= new; $pointer-to-string[0] = CArray[int8].new;
camelia ( no output )
lizmat dinner& 19:01
smls_ Another thing that has a "very Perl 5" API, is the 'Test' module. While that good for P5->P6 newcomers now, I predict that in 10 years it will be seen as baggage in core, and will be hard to get rid of.
Juerd lizmat: Eet smakelijk
chansen_ When is it Christmas according to Rakudo/MoarVM developers? I celebrate Christmas on the 24th, do I get my Christmas present? ;o)
Juerd smls_: That's okay. Test::More moved to Test, so the next thing can be called Test::More again ;) 19:02
smls_ heh ;)
muraiki chansen_: Orthodox Old Calendar Christmas, Jan 7
chansen_: just kidding :)
but yeah, if you guys need some extra days, there's always the old calendar
chansen_ muraiki: Bah, Julian calendar! ;)
stmuk its probably 25th CLT (Cola Late Time) 19:03
Skarsnik quick survey. what is the version of your libmysqclient on your system? (the libmysqlclient.so.18 for example)
stmuk 11:59 "just one leetle more commit"
lucasb another thing I hit is when trying to install 007: 'Could not find _007::Runtime::Builtins in ...'
abraxxa Skarsnik: how can i determine the version? just by looking at the filename? 19:04
nine Skarsnik: 18
Skarsnik abraxxa, yes, that how it's should work. (on unix) so.major.minor.subminor.areyoucrazy? 19:05
19:05 virtualsue joined
smls_ Skarsnik: /usr/lib/libmysqlclient.so.18.0.0 19:06
though that's actually MariaDB
abraxxa it seems Ubuntu mysql-client package nor its dependencies contain that lib
Skarsnik it's libmysqlclient 19:07
that has nothing to do with mysql-client actually
I think?
abraxxa a client that doesn require its client library, yeah looks like mysql
there is no package named like that 19:08
it's called libmysqlclient18 which should answer your question without installing
i still don't get it why the nested CArray stuff only fails in DBDish Oracle 19:09
TimToady well, after all that agonizing, my advent entry for the 24th just kinda wrote itself in an hour :)
abraxxa TimToady: did you delay the release until a code freeze period was established and bugs fixed in a not-overhastly manner? 19:10
RabidGravy abraxxa, I think you will find that if you put "no precompilation;" in the top of the file that *does* fail then it will work again
abraxxa RabidGravy: does the test not use precompilation? 19:11
RabidGravy not a script, no
Skarsnik it's just mysql found interesting to not follow a major abi version so I am not sure putting mysqlclient, v18 in DBIis is a good idea, since the v16 is not that old
and I am pretty sure the C api is exactly the same 19:12
araujo checks perl6.org after a long while .... nice redesign!
abraxxa RabidGravy: so precompilation doesn't get tested although it's on by default for production code?
RabidGravy only for modules
abraxxa finds that troubling 19:13
can it be tested?
TimToady it's tested when you run Test :)
hankache :) 19:14
19:14 cdg left, zengargoylew joined
abraxxa which doesn't use NativeCall an nested vars 19:14
muraiki ok, after `rakudobrew nuke moar`, building did not result in the exception I previously saw in install-core-dist.pl. it build successfully, yay!
hankache congrats muraiki
muraiki now I feel too tired to do anything with it though. heh 19:15
(not because of these problems, just in general)
I think it's time for caffeine
timotimo christmas tires you out, doesn't it ...
abraxxa there seems to be no 'use precompilation;'
timotimo yeah, because that's the default 19:16
abraxxa not for scripts
timotimo hm. perhaps it'd make sense to turn it on and off for individual statements
araujo still needs to go out tomorrow and buy the Christmas food/candies
timotimo yeah, we don't precompile scripts
abraxxa and what are .t files?
scripts!
timotimo yeah
so?
abraxxa i can't write a failing test case
Skarsnik you need to put the no precompilation in your module file
timotimo of course you can
abraxxa i could be moving it into a module
nine abraxxa: test scripts use quite a few modules
timotimo you can "use" the test code from your ... yeah
abraxxa so every t file should be duplicated in a module so it gets tested with and without precomp 19:17
timotimo perhaps we can build a second test runner that just pretends every test file is a module and uses it
RabidGravy yeah, if the failure is in a module put the code in a module
stmuk hmm Bailador freezes at the install stage even without using rakudobrew
abraxxa Skarsnik: the other way round, I want precompilation to get used and show the failure
nine timotimo: I do wonder wether we could actually precompile scriptes
timotimo nine: we can, but you'll have to run it with -e ''
Skarsnik for installed App probably
RabidGravy abraxxa, fwiw I can't find any way to get the failure in or out of a module 19:18
abraxxa i find it just too brittle as it is now because most of the tests are in scripts and so don't get precompiled
RabidGravy: interesting
RabidGravy it's very possibly some other set of circumstances that aren't immediately obvious 19:19
19:19 rickbike joined
Skarsnik could be because the module is loaded with require and not use? 19:19
abraxxa Skarsnik: need DBDish::Oracle::StatementHandle; 19:20
nine Skarsnik: no, both do precompile
Skarsnik yes but DBIish will require it later x) 19:21
in DBDIsh install_driver
is there a way to have Test display the type of stuff in got/expected?
or does is just call .gist? 19:22
abraxxa Skarsnik: maybe we should improve is-deeply
Skarsnik is-deeply is eqv
abraxxa as it's output is for human and we don't touch the Test API it shouldn't have negative effects
does this help? paste.scsys.co.uk/503418 19:23
nine RabidGravy: do we want to keep resources a hash with keys like 'libraries/p5helper' and 'images/foo.png' or do we want to follow the spec that would make it nested like $*RESOURCES<libraries><p5helper>? 19:24
lizmat: ^^^
stmuk oh it does successfully install Bailador but take ages
nine I'm trying to implement the spec right now, but it makes everything more complicated and I don't know if it actually helps with anything
Skarsnik Remove Bailador from star and put xml x) 19:25
nine Except... / is a platform dependent thing :/
Skarsnik for path
keep the /
and have stuff that work with it, convert it at the end 19:26
stmuk I liked the API of the p6 XML module -- which is rare for me and XML libraries
RabidGravy nine, I'm cool either way, spec if you can other if you can't
Skarsnik I don't like it because I think it leak x)
hankache lazy list is .. or ... 19:27
?
RabidGravy er
.. is a Range, ... isn't
e.g 19:29
lizmat nine: yeah, so making it in the hash, would take the platform dependency away 19:30
Skarsnik m: my $a = 4.4e0; say $a.perl; say $a.Str;
camelia rakudo-moar 82e6ca: OUTPUT«4.4e0␤4.4␤»
dalek blets: 45e82d3 | (Herbert Breunung)++ | docs/appendix-b-grouped.txt:
set op table completed
timotimo i don't think there's a problem with just letting the hash key on paths that always have /
RabidGravy m: say ( 1 .. 2 ).WHAT; say ( 1 ... 2).WHAT;
camelia rakudo-moar 82e6ca: OUTPUT«(Range)␤(Seq)␤»
abraxxa RabidGravy: i'm compiling moar again, maybe some later commit already fixed my issue
nine lizmat: OTOH we could declare the / to be logical, split the path on / and let the excellent IO::Path do the joining
lizmat timotimo: well, that was the essence of the newio branch, but that didn't make it 19:31
timotimo yeah :(
flussence
.oO( let's use UNC paths on windows, then we can have / everywhere! )
 19:32
lizmat but under pressure, everyting becomes fluid
abraxxa which is then incompressible ;)
lizmat so nine: if you think it is over enginieerd, then we'll skip
abraxxa nope, still fails
nine lizmat: that way, resources in the META6.json would always be an array, its members either strings or hashes for files with special attributes. With the nested structure, it's always hard to know what a hash means (directory or special file)
Skarsnik my panda is super slow :( 19:33
lizmat nine: fine by me... :-)
abraxxa Skarsnik: did you feed him too much?
stmuk hahaha
nine lizmat: well I do like the ideas and the $*RESOURCES<libraries><p5helper> interface. It's just that implementing it feels like walking uphill through mud :)
19:33 rurban left
stmuk panda seems slow on DBIish and Bailidor 19:33
lizmat well, that's us core devs being tortured for the greater good
nine lizmat: it may also just be exhaustion and stress caused by the relentles release count down 19:34
abraxxa what does this mean? Cannot find method 'CALL-ME' 19:35
at gen/moar/m-BOOTSTRAP.nqp:2854 (/home/ahartmai/.rakudobrew/moar-nom/install/share/nqp/lib/Perl6/BOOTSTRAP.moarvm::76)
hankache we do have a countdown?
where?
when?
how?
nine hankache: in my head ;)
japhb nine: Not relentless. It will relent for you as a Christmas present.
hankache: You could consider the advent to be our countdown
mst nine: trout.me.uk/tick.html 19:36
abraxxa never mind, typo
hankache nine: my @brain = @nine-brain.clone ;)
nine Oh and there's my girlfriend coming home in about an hour and I promised her to actually spend time with her this evening
Skarsnik lol
El_Che nine: you lier!
-a 19:37
mst this is why you need more than one; then you tell them all you're seeing a different one that night, and spend the time hacking instead
lizmat nine: could you point me to where you are working on this ?
Skarsnik Oh great
everything that use Str given by a NC routine will leak 19:38
japhb mst: That logic is so twisted it got knotted
timotimo Skarsnik: does "explicitly manage" help?
leont japhb: don't get him started on knots ;-)
japhb Point taken
mst japhb: there's an old joke about that being why mathematicians tend to have a wife -and- a mistress
Skarsnik There are explicitly managed and should not
*They
mst I just sort of reworked it for the oddities of my life ;) 19:39
japhb heh
abraxxa timotimo: isn't that the source of the leak? because it tells Perl to leave it alone in regards to gc?
nine lizmat: well with that decision out of the way it should actually be a little patch to Panda::Installer::install for detecting paths beginning with 'libraries/' and applying $*VM.platform-library-name and a similar patch to CompUnit::Repository::FileSystem::resource
timotimo not actually sure 19:40
hankache a lazy lisy is a (Seq) or a (List) ? 19:41
s/lisy/list
timotimo Seq is an iterator that will not let you fetch older elements once it's iterated past them
Skarsnik timotimo, if you do a my $a = native_sub(); and the native_sub is marked as returns Str you get a explicitly-managed Str and... it's probably not a good idea
timotimo a list is one part "reified" elements, one part iterator(s)
Skarsnik and I fear a my $b = $a will clone and keep the explicitlymanaged trait 19:43
19:43 stanrifkin left
timotimo clone? why would it clone? 19:44
abraxxa a module use NativeCall; and the test script too results in P6M Merging GLOBAL symbols failed: duplicate definition of symbol NativeCall 19:45
Skarsnik lol
and github.com/tony-o/perl6-data-dump should be the output of Test. it show the type of stuff. it's better than just .perl x) 19:46
hankache ++what Skarsnik said 19:48
19:49 labster joined
RabidGravy right now I'm not feeling confident that I will be able to get a release standard rakudo built *and* test all my modules tomorrow 19:49
abraxxa +1
and I only have one module ;(
Skarsnik someone know a C function that return a char*? 19:50
RabidGravy yeah, I have 29 :-\ (how did that happen)
Skarsnik and take like no argument x)
RabidGravy getlogin()
Skarsnik Oh thx 19:51
nine lizmat: maybe...we don't even need the patch to rakudo. NativeCall already maps between p5helper and libp5helper.so :)
abraxxa Skarsnik: PassChar in 15-rw-args.c
Juerd m: method Str::foo { "bar" }; say "foo".foo 19:53
camelia rakudo-moar 82e6ca: OUTPUT«Potential difficulties:␤ Useless declaration of a has-scoped method in mainline (did you mean 'my method Str::foo'?)␤ at /tmp/aS8p7RAPmR:1␤ ------> 3method7⏏5 Str::foo { "bar" }; say "foo".foo␤Method 'foo' not found for invocant of c…»
Juerd m: class Str { method foo { "bar" } };say "foo".foo
camelia rakudo-moar 82e6ca: OUTPUT«===SORRY!===␤P6opaque: must compose before allocating␤»
Juerd Is there a way to extend core classes? I'm looking to move Blob.unpack to experimental.
RabidGravy m: use MONKEY-TYPING; augment class Str { method foo { "bar" } };say "foo".foo 19:54
camelia rakudo-moar 82e6ca: OUTPUT«bar␤»
Juerd Great, thank you :)
abraxxa is something wrong with that code? paste.scsys.co.uk/503421 19:55
19:55 rurban joined
Juerd "Cannot augment Blob because it is closed" :( 19:57
flussence Juerd: ah yeah, that sounds familiar to me... never figured out a way around it and ended up making subs. 19:58
...which doesn't help for what you're doing, of course. :( 19:59
hankache m: my $lazylist = (1 ... Inf); say $lazylist.eager; 20:03
camelia rakudo-moar 82e6ca: OUTPUT«(timeout)»
nine abraxxa: we don't allow use lib in a module
hankache this doesn't time out in the latest rakudo ^^ 20:04
abraxxa nine: you mean I have to move it before the unit module declaration?
nine: ok, works but not so good error message 20:05
nine abraxxa: move it into the script or use -I or PERL6LIB
abraxxa: to quote jnthn++ "Note that for Perl 6.christmas, we will only support the use of use lib in scripts, not in modules, as its interaction with precompilation is more complex than we have time to reasonably consider (and it's better to wait until we've a good answer than to use a hacky one now)."
hankache does rakudo timeout? 20:06
nine hankache: no, why should it?
hankache nine: my $lazylist = (1 ... Inf); say $lazylist.eager;
it's hanging 20:07
nine hankache: yes, you ask it to do infinite work and it happily starts doing it
hankache nine then shouldn't we forbid Inf + eager ? 20:08
RabidGravy Skarsnik, this "Consider adding the api version of the library you want to use" is it possible to shut it up without specifiying the version?
flussence hankache: you're asking perl6 to solve the halting problem.
20:09 [Sno] left
Skarsnik RabidGravy, I was thinking of that. but in most cast it's a fault in the library 20:09
hankache flussence halting problem?
Skarsnik RabidGravy, do you have an example on where you can't put the version? 20:10
lizmat m: use MONKEY-TYPING; augment class Blob { method foo { "foo" } } # hmmm... this seems to hang ?
camelia rakudo-moar 82e6ca: OUTPUT«5===SORRY!5=== Error while compiling /tmp/PRk5uheAut␤Cannot augment Blob because it is closed␤at /tmp/PRk5uheAut:1␤------> 3use MONKEY-TYPING; augment class Blob7⏏5 { method foo { "foo" } } # hmmm... th␤ expecting any of:␤ g…»
lizmat ^^^ hangs for me locally, eating all mem 20:11
20:11 bobv joined
Skarsnik I am worried to receive that from NC gist.github.com/Skarsnik/181144a2efa6dc50af99 these string should not be expmanaged (and why only the first) 20:11
RabidGravy Skarsnik, yes, for files that I have created - the version is irrelevant
hankache flussence not really, just to throw a warning/error
Skarsnik RabidGravy, it should not bother if you put a path to the file, it only warn when it's 'foo' 20:12
try ./foo ?
20:12 AlexDaniel joined
RabidGravy also I have 8 NativeCall libraries to check and fix 20:14
which really, really sucks
dalek line-Perl5: 320187e | (Stefan Seifert)++ | / (5 files):
Use platform independent library name

Requires latest panda		 20:16
nine lizmat: there ^^^
abraxxa I wish you all a Happy Christmas! 20:17
RabidGravy nine++ # got it 20:18
abraxxa, have a good one 20:22
abraxxa my updated code is on github, I'll open an RT for the CArray[CArray bug 20:23
RabidGravy: thanks!
20:24 darutoko left
abraxxa hahaha 20:26
my RT is localhost
127001
rt.perl.org/Ticket/Display.html?id=127001 20:27
tadzik :D
20:28 skids left
moritz did you know that you can write localhost as 0 too? 20:29
point your browser to 0:5000/ for example :-)
tadzik oh, I thought it's 0.0.0.0 :)
flussence depends on the browser. Some crazy code interprets a bare number as a uint32 IP address 20:30
abraxxa it's ::1
Skarsnik hey tadzik, did you see the new commit to my pr on json-unmarshal? x)
Juerd m: pack()
camelia rakudo-moar 82e6ca: OUTPUT«5===SORRY!5=== Error while compiling /tmp/9llXJ0PwsG␤Calling pack() will never work with any of these multi signatures:␤ (Str $template, *@items) ␤ (@template, *@items)␤at /tmp/9llXJ0PwsG:1␤------> 3<BOL>7⏏5pack()␤»
flussence (doesn't work in chromium here)
Juerd I dont understand the error message
Will never work with any of these...
tadzik Skarsnik: not yet, will look now :)
Juerd Shouldn't that be s/never/only/?
RabidGravy nah, 20:31
lizmat Juerd: hmmm....
flussence Juerd: "will never work, "
RabidGravy m: pack(42)
camelia rakudo-moar 82e6ca: OUTPUT«5===SORRY!5=== Error while compiling /tmp/jW4_r92yEz␤Calling pack(Int) will never work with any of these multi signatures:␤ (Str $template, *@items) ␤ (@template, *@items)␤at /tmp/jW4_r92yEz:1␤------> 3<BOL>7⏏5pack(42)␤»
Juerd flussence: Same
TimToady for each of these signatures, what you said will never work with it
Juerd RabidGravy: Oh, like that :) 20:32
abraxxa bye!
flussence maybe it should say "Calling pack() with no arguments" in that one case
20:32 abraxxa left
RabidGravy yeah, it's just not clear when there are no args 20:32
Juerd I interpreted the () in 'pack()' as a function sigil, not as an empty signature :)
tadzik Skarsnik: ooh, I see it now :) Nice!
TimToady has scheduled his advent post 20:33
RabidGravy TimToady++
tadzik ooooh:)
flussence gets hyped for 4 hours from now
Skarsnik tadzik, for the "42" stuff, I had a webservice that send id at string because they don't fit in a int32, but since perl6 Int is a bigint it map nicely x) 20:34
20:35 vv3 joined, vv3 left
[Coke] it's not too late to mark pack/unpack experimental. 20:35
20:35 vv3 joined, vv3 left
Juerd [Coke]: I'm working on it, actually :) 20:35
tadzik Skarsnik: yeah, I can see the applications :) I just wonder if maybe there should be a way to keep the module "strict" 20:36
'cos now, with your change, it kind of coerces types
20:36 vv3 joined, vv3 left
tadzik and I can imagine someone using J::M as sort of a validator, a bit like Json Schemas 20:36
Skarsnik Well it still depend on the type of the attribute 20:37
20:37 vv3 joined, vv3 left
tadzik m: my $a = Int("99bottles"); say $a.perl 20:37
camelia rakudo-moar 82e6ca: OUTPUT«Cannot convert string to number: trailing characters after number in '0399⏏5bottles' (indicated by ⏏)␤ in block <unit> at /tmp/saagdwaJ3_:1␤␤Actually thrown at:␤ in block <unit> at /tmp/saagdwaJ3_:1␤␤»
tadzik perl6++
Skarsnik I can't think if it's a NC bug or a failuyre to not have Explicitly managed returned char * gist.github.com/Skarsnik/7f2fee97d56c5acc1587
tadzik Skarsnik: I'm trying to be a devil's advocate and find a case where it could possibly bite someone :P 20:38
20:38 vv3 joined, vv3 left, virtualsue left
Skarsnik There is nothing that make these value explicitly managed in the code 20:38
tadzik though I very much value that we can now handle bigints like in that case you mentioned
20:39 vv3 joined, vv3 left
tadzik one could argue that CustomUnmarshaler already covers that, though this is definitely much nicer when it Just Works :) 20:39
20:39 vv3 joined
Skarsnik Yes x) 20:39
I also bothered RabidGravy to have automaticly _ case converted to -, but it was more me beeing lazy to not mark everything with is json-name for plop_id that end in plop-id x) 20:41
tadzik haha
Juerd I wish _ and - were considered the same thing
tadzik that's a good question though: how does JSON::Marshal handle bigints?
Skarsnik err
Juerd Since I've become accustomed to the kebab casing, underscores are annoying to type 20:42
tadzik 'cos if it turns them into strings, or we can agree that it should, then I have no other doubts :)
20:42 bobv left
RabidGravy it will just push them out as they are, if they needed to be strings for an API I'd suggest a custom marshaller 20:43
20:43 mohae joined
Skarsnik Also having a way to add unmarshaling to some core type could nice too. like webservice send timestamp in ISO-something than DateTime understand. I wanted to have t x) 20:43
*to add that
It's probably just a matter to be able do add a _unmarshal(DateTime, $json) 20:44
RabidGravy Skarsnik, tadzik I'd suggest another module that provides a bunch of "standard" marshallers/unmarshallers
tadzik RabidGravy: I just thought about that exactly!
20:44 vv3 left
tadzik well, my idea was to have it in core, but I can be convinced 20:45
RabidGravy so if it's loaded it looks the type up and gets a marshaller
20:45 mohae_ left, khw left
RabidGravy so if you want the bog standard behaviour do nothing, otherwise load the module 20:46
20:46 khw joined
tadzik that's not a bad idea 20:46
still, I think it's kind of our Core Duty to make bigints DTRT
Skarsnik Yes, but you will need to add something in marsh/unmarch to be able to 'extend' tehm?
tadzik even more so with marshal than unmarshal
nine I can't believe my last TODO before the release is done and it didn't even involve a patch to rakudo :)
tadzik and since all three of us are at our keyboards, we can probably come up with something sensible :) 20:47
RabidGravy I'm not quite sure what the problem with bigints is
Skarsnik Discord send me BigInt id (aka does not fit on a int32)
so it send it as string
tadzik so JSON defines its ints to be int32?
Skarsnik but perl6 Int are big enought
20:48 virtualsue joined
tadzik (I assume so from your example, but I haven't actually checked) 20:48
Skarsnik I have no idea
20:48 vv3 joined
tadzik I'll check 20:48
RabidGravy no, it's some other software can't handle it
Juerd Making something experimental is hard :) 20:49
Damn lexicals!
tadzik json spec says nothing about range of "number"
RabidGravy m say from-json('{ "foo": 11111111111112222222222222222222222222222222222222222222233333333333333333333333333333333}').perl
m: say from-json('{ "foo": 11111111111112222222222222222222222222222222222222222222233333333333333333333333333333333}').perl 20:50
camelia rakudo-moar 82e6ca: OUTPUT«${:foo(11111111111112222222222222222222222222222222222222222222233333333333333333333333333333333)}␤»
20:50 yqt joined
RabidGravy we can parse it 20:50
tadzik right
everything looks like it's actually Discord doing something weird here
[Coke] Do panda modules need to be updated? just ran a freshly built perl6 on bootstrap.pl in a freshly updated panda: gist.github.com/coke/726361e325c678d0472c 20:51
tadzik so I'm thinking, maybe it's actually better to have some nice, readymade tool (an unmarshaler probably) to handle this sort of odd documents rather than have it as a standard behaviour 20:52
20:52 vv3 left
[Coke] how do you login to perl6 advent with the new layout? 20:53
RabidGravy which is why I'd say if it is application specific it should be possible but not built in
tadzik [Coke]: I went to wordpress.com, logged in, then went to the advent blog
20:53 punter joined
tadzik yeah, I agree on that 20:53
what do you say, Skarsnik
20:53 vv3 joined, vv3 left 20:54 vv3 joined, vv3 left
tadzik I think we can come up with some nice syntax for this sort of stuff, so keep it explicit, but not make it a hassle 20:54
I see no possible harm in that :)
20:55 vv3 joined, vv3 left
RabidGravy but that's what "unmarshalled-by", "marshalled-by" are for 20:55
tadzik right
and is unmarshalled-by(Int(*)) enough to make it work?
20:56 vv3 joined
jnthn Juerd: Another way to make pack/unpack experimental is to have them look for a certain symbol, say $?ALLOW-EXPERIMENTAL-PACK, and then look for it to exist (nqp::getlexcaller would do it), and then just export that symbol from experimental when the :pack tag is used. 20:56
yoleaux 18:50Z <nine> jnthn: fde3d620bf1b22f1b78420e1294a04c663145efe is the last commit with which I can repro run_alt. With b6f0532c2b6fcf95b65f1de63537e91821a123fd it disappeared again. Didn't have to downgrade nqp for that.
RabidGravy so "class F { has $.very-big-number is marshalled-by('Str') is unmarshalled-by('Int') }
tadzik even that, huh
that's solid work on those traits, I see :)
Juerd jnthn: I just got something working, will submit PR in a minute.
jnthn OK, nice :) 20:57
20:57 lichtkind left
[Coke] tadzik: thanks, that did the trick! 20:57
RabidGravy yeah, it's a string method name or a sub :)
Juerd jnthn: github.com/rakudo/rakudo/pull/649
jnthn Juerd: looking
I'm +1 to making them experimental
Juerd my constant EXPERIMENTAL-PACK := True; is a hack to make it work for the next callee 20:58
jnthn Yes, was gonna say 20:59
I think it's brokne otherwise too
What happens if you do use experimental :pack; sub foo() { ...use pack... }; foo()
I think both can be fixed by using CALLERS:: instead of CALLER::
Which searches all the way down the dynamic chain
Juerd Ah 21:00
Will try that instead
21:01 vv3 left, vv3 joined, vv3 left 21:02 vv3 joined, vv3 left 21:03 vv3 joined, vv3 left 21:04 khw left, vv3 joined
Juerd jnthn: Nope, doesn't work with CALLERS 21:05
21:05 atweiden joined
Juerd jnthn: And I'm out of time, quite possibly for the next few days. :( 21:05
21:06 geraud joined
jnthn OK, lemme see if I can do it before I get sent to bed :) 21:08
lizmat fwiw, I'm ambivalent to making pack/unpack experimental at this stage 21:09
21:09 vv3 left
Skarsnik tadzik, RabidGravy I don't see the issue of doing it in the 'core' since you have to write the Attribute as Int for the coerce to Int to be done 21:09
RabidGravy but if it's coming as a string then it's a string 21:10
hankache anyone has an example of this: doc.perl6.org/type/Seq#method_from-loop
Skarsnik It's not like you say has $.foo and $foo is automaticly a Int
tadzik RabidGravy: on the other hand, we already cheat with Rat
21:10 vv3 joined, vv3 left
Skarsnik and Num x) 21:11
tadzik or do we, hmm
21:11 vv3 joined, vv3 left
tadzik yeah, it's the same story: declaring an attr as Rat means (for JSON::Unmarshal): try to make Rat from this, whatever it is 21:11
RabidGravy there was one that is cheated, because the json doesn't distinguish
tadzik it will be consistent if Int were to work the same way
we also do that with Bool 21:12
21:12 vv3 joined
tadzik and here there's no excuse 21:12
labster Good morning, #perl6. Any LHF needing some work? 21:13
RabidGravy I'm cool with it 21:14
Skarsnik LHF?
lizmat RabidGravy: cool with pack/unpack becoming experimental ?
Skarsnik Oh low hanging fruit x)
lucasb labster: LHF: Add --make-install description to rakudo's Configure.pl :)
... Configure.pl help message 21:15
RabidGravy lizmat, that as well ;-)
labster that sounds pretty low!
Skarsnik guess_libname in NC should be cached ~~ 21:16
RabidGravy Skarsnik, however for marshalling I'm still pretty certain that it will have to be a marshalled-by('Str') for those cases 21:17
21:17 vv3 left
Skarsnik probably 21:17
21:18 vv3 joined, vv3 left
Skarsnik How I can marshal/unmash a timestamp with the 2 trait? well it's marshalled by Str but it can be created like that DateTime.new('string value'); 21:18
Juerd jnthn: Great! Thanks 21:19
jnthn Got a patch that seems to work :)
Juerd To everyone who's going to celebrate: merry christmas!
I'm disallowing myself to participate in #perl6 for the next few days because of dayjob deadlines. 21:20
21:20 Juerd left
RabidGravy DateTime::Parse or something 21:21
whatever H::UA uses
Skarsnik I mean how I use the trait? x) 21:22
Hotkeys_ Haven't been reading this channel for a few days, any new developments?
RabidGravy oh
Hotkeys_ How's the release looking
dalek kudo-star-daily: d68d5bd | coke++ | log/ (8 files):
today (automated commit)
Skarsnik like has DateTime $timestamp is marshalled-by('Str') is un-marshalled-by('new') ? 21:23
[Coke] so, is fresh panda working for anyone? 21:24
lucasb [Coke]: worked for me when I bootstraped 1 or 2 hours ago. didn't hit the EVAL usage errors you found
RabidGravy something like "... is marshalled-by('Str') is unmarshalled-by(sub ($d) { DateTime.new($d) })" 21:25
21:25 CIAvash left
stmuk panda seems to generally work with a few modules slow to install 21:25
RabidGravy for the method name it has to be a method that is called on the value that you have rather than the type
tadzik [Coke]: works fine here 21:28
labster m: "»ö\x[342]\x[20f0]«".say # Camelia with santa hat 21:30
camelia rakudo-moar 82e6ca: OUTPUT«»ö͂⃰«␤»
flussence [Coke]: works here too (my daily reinstall process includes `panda install Linenoise`) 21:31
21:31 Hotkeys_ is now known as Hotkeys
Hotkeys last time I used panda it took so long to start doing things I thought it had hanged 21:32
flussence should get to work on replacing "daily" there with a script that monitors irssi's logfiles for commit messages...
Hotkeys lol
RabidGravy [Coke], how fresh?
Hotkeys why not just monitor git
[Coke] both are running the latest commit. 21:33
flussence I don't think inotify works over remote git yet
[Coke] trying to remove the install directory for perl6 and do a complete rebuild.
RabidGravy just trying 21:34
Skarsnik err my modules does not work now 21:37
Cannot call infix:«<»(Any, Int); none of these signatures match:
on a given block
21:38 cognominal left
Skarsnik it worked yesterday 21:38
lizmat Skarsnik: any idea where / how that happens 21:39
anything to do with dates or instants ?
Skarsnik no
lizmat *phew*
flussence [Coke]: just finished building stuff and reinstalling panda. It takes *ages* at "==> Fetching panda" but eventually completes. 21:40
Skarsnik github.com/Skarsnik/perl6-gumbo/bl...er.pm6#L63 it fail on that and the last travis build was working
lizmat flussence: pretty trivial to pump IO.watch-path supply onto an IO::Socket::Async supply ?
Skarsnik There not even a < comparaison on this
[Coke] going better this time... 21:41
lizmat Skarsnik: which module are we talking about? something in th eecosystem ?
Skarsnik Gumbo in the ecosystem (you will need to install XML with notests 21:42
dalek kudo/nom: bdc8003 | (Juerd Waalboer)++ | / (2 files):
Make pack and unpack experimental

The current pack and unpack are incomplete and not extensible. Future improvements will be necessary, but can't be made without breaking compatibility.
There is currently no language specification for pack and unpack.		 21:43
[Coke] the panda output "==> Please make sure that inst#/Users/williamcoleda/sandbox/rakudo/install/share/perl6/site/bin is in your PATH" is confusing.
ast: fde8fcf | jnthn++ | S (4 files):
Tests using pack/unpack should `use experimental`.
[Coke] it's giving a perl6 lib style path but it's meant for the system path.
jnthn++ Juerd++
afk a bit 21:44
tadzik [Coke]: yeah, I guess that message is a bit obsolete :) 21:45
dalek kudo/nom: 4355eac | jnthn++ | src/ (3 files):
Remove "as Foo" coercion trait.

Which has long been replaced by Foo() coercion types.		 21:46
jnthn 'night, #perl6
timotimo gnite jnthn
have a nice 24th :)
RabidGravy be groovy 21:47
AlexDaniel m: enum Animal (Cat => 25, Dog)
camelia rakudo-moar 82e6ca: OUTPUT«5===SORRY!5=== Error while compiling /tmp/MoGn2qU4cx␤Undeclared name:␤ Dog used at line 1␤␤»
stmuk so everything using NC needs to remove 'lib' from the front?
timotimo oh?
Skarsnik Yes, blame nine :)
timotimo huh? 21:48
RabidGravy panda working on a just finished installing rakudo here
stmuk Cannot locate native library 'liblibcurl.so': liblibcurl.so: cannot open shared object file: No such file or directory 21:49
Skarsnik it's not that stupid, since on windows lib are just foo.dll
RabidGravy yeah jusr 'curl'
21:49 terjekv joined
timotimo yeah, it'll put the lib in front for you 21:49
Skarsnik fun fact: there is some lib that are called liblibfoo x) 21:50
timotimo there's also libowfat (so you can -lowfat)
RabidGravy and libiberty 21:51
lucasb and libuv (so you -luv it) :)
geekosaur libroken
timotimo oh!
i never noticed that
geekosaur (which admittedly goes the other direction)
terjekv Hi. We're trying to resolve a circular module call using stubs. It works fine if the modules are located in the same file, but if they are spread across different files we get the message "the following packages were stubbed but not defined". Is there a (documented) way to have circular module calls across different files?
Skarsnik Should I merge than now? github.com/perl6/DBIish/pull/40 DBIish will not work for new rakudo without it x) 21:52
terjekv, You can't sadly I think
I solved by being loose on some type
+it 21:53
terjekv Ah. Right. Well, that at least explains a lot. And thanks for the idea.
21:54 [Sno] joined
RabidGravy well if you're not using them for typeconstraints you can require the module and the ::('Foo') it to get the package/class whatever 21:54
Skarsnik Maybe it can work if you could do unit module Foo in each file
RabidGravy I've used that in a couple of places
Skarsnik but I did not try it
21:54 muraiki left
Skarsnik I think it's weird to not be able to have that 21:55
RabidGravy so e.g. require Foo; my $f = ::('Foo').new
Skarsnik I have a channel class that has an owner (so a User) and the user class has channel on it
terjekv That sounds oddly familiar. I'm not quite sure what unit module actually does, looking at the module documentation for a few. 21:56
(and the ::('Foo').new idiom kinda hurts my eyes)
RabidGravy yeah it's not pretty 21:57
Skarsnik For me it's a bug x)
21:58 Begi left
Skarsnik I understand if for the precompile to work 21:58
terjekv Single pass etc, yeah. But it's a bit odd that you can get around it in the same file but not across files. 21:59
Also, uhm, what does the "unit" in "unit module Foo" do?
Skarsnik it declare the file as the Foo unit
21:59 hlafarge joined
Skarsnik if you want it's like putting module Foo { } around everything 21:59
RabidGravy which is what I prefer 22:00
Skarsnik but I think remember reading that you can have multi file of the same unit
terjekv Ah. Right. Akin to package {} (conceptually) in perl5? Makes sense though. 22:01
RabidGravy yeah
terjekv I think there will be some stabbing at this, thanks for the input! 22:02
RabidGravy there is package, module and class
Skarsnik I feel like RabidGravy, having code accross multiple module/patch in rakudo/modules to write something that I never finished writing because I spend time patching other stuff xD 22:03
terjekv (It's not bad until a branch -l lists enough output that you reach for grep) 22:04
flussence is pleasantly surprised to find that `st` has about 17 times more unicode support than `urxvt`
22:04 erisch_ joined 22:05 erisch_ left, punter left
RabidGravy gets jiggy with a Proxy in an AT-KEY 22:06
stmuk ha a script I last fixed 2 weeks ago needs more changes more
lizmat RabidGravy: in core?
22:06 terjekv left
RabidGravy nah, in this GDBM module which I earlier claimed to be able to finish today but I fell asleep for a couple of hours 22:08
22:08 pierre-vigier joined
Skarsnik I love at-key x) 22:09
22:11 maettu joined
labster m: (* // "foo").WHAT.say 22:13
camelia rakudo-moar 4355ea: OUTPUT«(Whatever)␤»
22:13 xfix left 22:15 hlafarge left
RabidGravy yeah it's defined 22:15
labster But should it curry? 22:16
lucasb iow, &&, || and // don't participate in whatever currying. (right?)
RabidGravy PUT IT IN THE CURRY!
22:17 nmpa joined
timotimo looks like, lucasb 22:17
22:17 raiph joined
captain-adequate TimToady: Nice post today. 22:18
dalek ast: d8d63b4 | lizmat++ | S03-smartmatch/capture-signature.t:
Fix fallout of " as " removal
 22:19
ast: b0e1278 | lizmat++ | S32-temporal/Date.t:
Fix no special format large/small years fallout
timotimo huh? where's toady's post?
stmuk it was emailed out to those with WP access it seems 22:20
22:20 nmpa left
lucasb
.oO( today's toady's post )
 22:20
captain-adequate Oh... The Perl6 advent calendar.
Can't seem to find it on the website... Must've just been emailed I guess. 22:22
timotimo i didn't get a post :( 22:24
maybe it was accidentally published, then quickly deleted again because it was meant to go up in an hour or so?
labster Don't open your blog posts before christmas, lucasb 22:25
perlpilot reads timtoady's post now as it's likely he'll have spotty internet later 22:28
timotimo are recorded talks from the LPW going to be up somewhere? 22:30
... did talks get recorded?
22:30 krshn joined
stmuk mdk tweeted about looking at the video an evening or two ago 22:32
hankache when will the post go live?
stmuk so I imagine the wheels are turning 22:33
hankache waiting and waiting and waiting
timotimo hankache: i can't actually see the post as scheduled in wordpress
perlpilot hankache, when its meant to be published. 12/24
dalek kudo/nom: 047db4b | lizmat++ | src/core/Date (3 files):
Streamline DateTime handling

  - main new logic now in .new($y,$m,$d,$h,$mi,$s,$t) which allows for coercion
  - no special format for small/large years
  - check-date/check-value helper functions removed
  - fixed some cases where multiple of a unit name would not be accepted
  - use native ints where possible
  - optimized earlier/later a bit
  - truncate-parts is now private method truncate-ymd
timotimo not sure if i've got sufficient privileges to see it if it was there
lizmat learned a lot about dates/instants/datetimes and posix today
diakopter TimToady: should the title have a Day 24: prefix
hankache perlpilot there are a lot of 12/24 :)
nine lizmat: it's a ridiculously complex subject 22:34
lizmat I actually found a few issues that I will try to look at tomorrow
hankache In my part of the word it's already 12/24
perlpilot hankache: okay, WP says ... Scheduled 2015/12/24
hankache perlpilot timezone? 22:35
perlpilot hankache, I dunno really. I was kind of guessing UTC
hankache, see ... your problem is that you're from the unevenly distributed future :) 22:36
hankache perlpilot :) 22:37
stmuk has the behaviour of File::Find changed in the last 2 weeks? 22:38
perlpilot Under WP's general config is a checkbox for "Show falling snow on my blog until January 4th" Why is that so much easier to find than the timezone
?
stmuk: my plan is to eventually make File::Find obsolete-ish by adding an option to dir() for recursion (there's already a :test for constraining what it returns) 22:40
(or ... maybe if I just keep mentioning it here, someone will do it for me, 'cause it doesn't look like I'll have time to work on it until next year :) 22:41
22:41 skids joined
nine perlpilot: next year isn't that far away 22:41
perlpilot: I for one plan to do no hacking at all the next couple of days :) 22:42
Skarsnik and when you will get back, you will get hundred of hungry mail about precompil stuff!
perlpilot nine: in addition to holiday festivities, I'm moving my family a few hundred miles (still in Texas though :)
nine And now: sleep. Good night #perl6 :)
labster lizmat++ 22:43
Skarsnik Good night
perlpilot good night nine!
labster Good work on the DateTime mess.
22:43 pierre-vigier left 22:45 pierre-vigier joined
jdv79 FROGGS: around? 22:47
22:49 pierre-vigier left, pierre-vigier joined, pierre-vigier left
lizmat labster: thank you, it has been on my radar for a while 22:52
having had to deal a lot with epoch / dates at former $work, I would like to see them faster :-) 22:53
labster Don't remind me I need to deal with dates at current $work
I'm going to double-check that one of the changes doesn't break large years, though. 22:54
dalek kudo/nom: 18947f4 | lizmat++ | src/core/DateTime.pm:
Fix for RT #127003 , zefram++
22:54 starless joined
lizmat labster: if it did, please add tests for them :-) 22:55
perlpilot lizmat++ for DateTime, labster++ for double-checking :)
lizmat because all of my changes are spectest clean
krshn a long time perl user just getting started on perl6 22:56
what is the best resource to start out with
RabidGravy docs.perl6.org
Skarsnik perl6intro?
22:56 rindolf left
labster > DateTime.new(127317232781632218937129); 22:56
-2602446-2807173-1473579083988834741T17:38:49Z
timotimo github.com/rakudo/star/blob/master...course.pdf <- this one's pretty good 22:57
Skarsnik that an interesting date
labster The same problem I ran into: anything that stores a year needs to be an Int. 22:58
krshn thanks -- have been going through docs.perl6.org for a while. I was hoping something like Modern Perl or something
docs.perl6.org is great though
22:59 starless left
Skarsnik there is a perl6intro thingy but I can't remember the url x) 22:59
timotimo perl6intro.com
jdv79 it looks like my pause changes got reverted. *sigh*
krshn timotimo: looks perfect. thanks 23:00
timotimo have fun with it! :)
23:00 molaf left
krshn loving perl6 so far. but still start writing perl 5 at the drop of the hat 23:01
the biggest hurdle for me is to think in terms of perl 6
labster lizmat: Either we add a couple Ints, or we add a fast path for normal dates which use int, and a slow path for people who want years in the billions 23:02
krshn let me throw a random question out there 23:03
Skarsnik timotimo, any idea where this error could came from? travis-ci.org/Skarsnik/perl6-gumbo...s/98570710 the previous build (2 days) ago was working and I only changed nc lib path in this. The code pointed by the error is just given on a Int (github.com/Skarsnik/perl6-gumbo/bl...r.pm6#L63)
krshn i am a scientist and have always wanted to code only in perl.. will perl6 be my goto language for scientific computing?
lizmat labster: year is already supposed to be an Int... 23:04
krshn I am currently moving from matlab -> julia and it has been good so far .. but in my opinion, perl6 can be a very good competitor to julia for scientific computing too.. any thoughts? 23:05
Skarsnik No idea. It's not really fast and it does not have lot of testing for Math like other language could have
krshn it has multiple dispatch, native types 23:06
Skarsnik but since it's easy to create sub language in perl6 with the slang, it can probably be interesting x)
krshn so potentially, there is hope
timotimo Skarsnik: that's weird. i don't see an instance of infix:«<» in that code
hankache krshn scientific computing relies most of the time on functional programming
as a matter of fact Perl 6 can be used for functional programming 23:07
labster lizmat: here github.com/rakudo/rakudo/blob/nom/...ime.pm#L98
hankache it has lazy lists, junctions, hyperoperator, feed operator (pipe)
lizmat labster++
perlpilot hankache: heh, "can be"? that's an understatement :)
timotimo scientific computing probably also relies on a bunch of matrix manipulation and vector arithmetic and statistics on big series of numbers and such 23:08
krshn thats my point too .. perl6 has a lot of potential
hankache perlpilot: s/can be/is
krshn the way I see it, Perl 6 is a mutant, it copied every cool feature from every language out there 23:10
23:10 perlpilot left
timotimo that'd make it more of a chimera, IMO 23:10
hankache :D
krshn haha :-)
labster lizmat: $a and $c and probably $e need to be Int. Which is why I'm starting to think we need a pessimized path.
23:10 cpage_ left
lizmat labster: not before Christmas 23:12
but yeah :-) 23:13
it appears we need an extended format after all :-( #127007 23:14
labster All of these ISO edge cases make me want to restrict core DateTime to RFC 3339. 23:15
23:16 cygx joined
cygx o/ 23:16
krshn: perl6 has native arrays and ffi, so it just takes someone to write a bunch of wrappers over the appropriate fortran libs
timotimo fortran has a different order for multi-dimensional arrays than c (and perl6) do 23:17
cygx just write an AT-POS that mangles the index, and you're good to go 23:18
labster Are we frozen for v6.c, or can I continue to commit bugfixes?
cygx (assuming you fortran lib uses ISO_C_BINDING)
don't know off-hand what's required for native fortran bindings 23:19
grondilu Perl 6 for functional programming? Not really. No tail recursion.
Skarsnik spec are freezed I think, but bug fix or unspeced stuff could change I think
RabidGravy Hmm, the gdbm thing works a lot better without an error callback
krshn actually, let me ask it from the other angle.. is there anything that structurally constrains perl6 from being good at scientific computing?
RabidGravy anyway, ttfn
timotimo ttyl RabidGravy 23:20
scientific computing was always something we wanted for perl 6 to be good for
grondilu just make a R slang and be done with it
timotimo hehe.
AlexDaniel haha :) 23:21
krshn thats exactly my impression from summarily reading about perl6 features
wow.. if perl6 does scientific computing, my 10+ years wait will be over -- I started with perl5 10 years ago :-) 23:22
AlexDaniel krshn: yeah. In that sense the possibilities are endless
krshn: I think that it needs just a little bit of time for all such things to be written 23:23
krshn i meant, 10+ years of wait will be worth it
cygx interesting side note: C99 was basically the attempt to eat fortran's lunch regarding scientific computing
tgmath.h, _Complex, variable-length arrays, restrict, ...
lizmat labster: bugfix away, I know I am...
AlexDaniel m: say τ² + ½ 23:24
camelia rakudo-moar 18947f: OUTPUT«39.9784176043574␤»
lizmat labster: but aren't we at least internally working with RFC 3339 ?
23:24 RabidGravy left
AlexDaniel krshn: ↑ though it already has a lot of cool stuff :) 23:24
cygx
.oO( whatever happened to Fortress? )
 23:25
krshn yep. I have been on-and-off on perl6 over the past 10 years .. but now, I am going to make a serious attempt at writing all perl in perl6 23:26
jdv79 so, the perl field value in META6.json should be v6 or 6?
AlexDaniel krshn: great! Make sure to report all bugs that you find :)
krshn: if you stumble upon any
jdv79 since we've now concluded, it seems, that version should not have the v in string form
23:28 kid51 joined
krshn does anyone know when a camelia book will be out? 23:28
AlexDaniel grondilu: well, you can still use tail recursion. It's just that there's no tail call elimination :) 23:29
23:30 kaare_ left
lizmat $ 6 'DateTime.new(127317232781632218937129).say' 23:30
+4034522497029953-07-13T17:38:49Z # labser
labster ^^^
krshn when I try to install the latest version of perl6, I get Could not find symbol '&bool' 23:37
timotimo ah, yes. you'll have to kill your install folder to make that part go away
Skarsnik hm, you install with what?
krshn An exception occurred while evaluating a constant at /home/krshn/local/linux/share/perl6/sources/0FD28AE1C9590AB6C8A04A508BF05A7043967B19:13 Exception details: Could not find symbol '&bool' in block at /home/krshn/local/linux/share/perl6/sources/0FD28AE1C9590AB6C8A04A508BF05A7043967B19:13 23:38
jdv79 woohoo. fixed.
Skarsnik timotimo, How I can track down commit? (to find the culprit of my infix < error)
timotimo "git bisect" helps with that
23:38 AndyDee joined
jdv79 that's the fastest pause fix ever. < 1h. 23:38
dalek ast: 38e9881 | lizmat++ | S32-temporal/DateTime.t:
Test for RT #127003
ast: 0fa2c59 | lizmat++ | S32-temporal/Date (2 files):
Seems we need an extended format after all
krshn checkout out rakudo from git
nom branch
timotimo don't forget to always checkout the tools/build/NQP_REVISION and tools/build/MOAR_REVISION inside nqp 23:39
krshn doesn't this do it for me? perl Configure.pl --gen-moar --gen-nqp --backends=moar --prefix=$HOME/local/linux
Skarsnik it should work 23:40
you just did it? no previous checkout?
23:40 Ben_Goldberg joined
krshn so I am getting this error for the first time.. installed it this way 5-6 times in the past 23:40
timotimo krshn: yeah, files left over are causing this problem, i believe 23:41
Skarsnik yes, but is that a clear rakudo directory?
timotimo i personally have the install folder for perl6 only so i can wipe it if necessary
Skarsnik because for now the whole stuff does not like old stuff x)
krshn Skarsnik: yes.. my rakudo is clean
dalek kudo/nom: 8955469 | lizmat++ | src/core/Date (3 files):
Fixes for RT #127005 and RT #127007, zefram++

Plus additional large year value fixed prompted by labster++
krshn but as timotimo said, it could be left over from previously
is there a way to delete it ?
timotimo nothing like "make uninstall"; only rm can do it for you 23:42
krshn I tried deleting only that dir but got a bunch of errors after that
timotimo which dir exactly?
$HOME/local/linux ?
krshn No.. deleted /home/krshn/local/linux/share/perl6/sources/
atweiden timotimo: in JSON::Fast source, lib/JSON/Fast.pm:54 isn't compiling on rakudo 2015.11 23:43
Unable to parse expression in argument list; couldn't find final ')'
lizmat oops s/127005/127004/
cygx krshn: get rid of the whole perl6/ thing
timotimo atweiden: let me have a look
krshn ok.. I was hesitant doing that :-(
timotimo atweiden: is that the version that's inside panda? 23:44
atweiden this is git head
but i believe it's the same
timotimo can you tell me your exact rakudo version? 23:45
also, does "STATEMENT_LIST('hi')" work for you?
like, in a perl6 -e ?
atweiden it's the 2015.11 release, compiled as github.com/atweiden/pkgbuilds/blob...o/PKGBUILD
timotimo oh!
you said 2015.11
sorry about that
yeah, that's incompatible 23:46
atweiden i figured
timotimo you'll need an earlier JSON::Fast for that
atweiden so it works on nom?
krshn cygx: that worked. thanks.
cygx krshn: when in doubt, go with the nuclear option ;) 23:47
krshn anyone one know happened .. why the complaints?
hehe :-)
cygx something-something change of version scheme
Skarsnik lot of change in precompiled stuff + change everywhere
krshn is that a recommended workflow when upgrading perl6 -- or this is transient because of so many things changing? 23:48
timotimo it does work on nom, yes
krshn I mean the nuclear option :-)
Skarsnik nuke option is the way x)
timotimo in 2015.11 you'll probably just have to replace STATEMENT_LIST with SEQ
can you try that, atweiden? 23:49
cygx krshn: it's supposed to be a one-time thing... until the next time it becomes necessary ;)
atweiden timotimo 1 sec
krshn got it :-)
dalek ast: 2763fa2 | lizmat++ | S32-temporal/DateTime.t:
Add tests for RT #127004
krshn how does something like this work -- perl6 -e 'm: say pi²' 23:50
Skarsnik timotimo, I mean, is there an easy way to deal with the triple npq/rakudo/moar to check old version ? x)
krshn is pi a predefined variable?
timotimo m: say pi
camelia rakudo-moar 18947f: OUTPUT«3.14159265358979␤»
Skarsnik pi is a constant
nebuchadnezzar m: my $s = "A\c[COMBINING DOT ABOVE]\c[COMBINING DOT BELOW]"; $s.NFD.map(&uniname).map(&say);
camelia rakudo-moar 18947f: OUTPUT«LATIN CAPITAL LETTER A␤COMBINING DOT BELOW␤COMBINING DOT ABOVE␤»
timotimo m: pi = "hi"
camelia rakudo-moar 18947f: OUTPUT«Cannot modify an immutable Num␤ in block <unit> at /tmp/g6Z_B9QU_a:1␤␤»
timotimo Skarsnik: delete the nqp and moar installations and use --gen-moar --gen-nqp 23:51
labster m: say π;
camelia rakudo-moar 18947f: OUTPUT«3.14159265358979␤»
Skarsnik the gen try to match the rakudo version?
krshn m: pi = "hi"
camelia rakudo-moar 895546: OUTPUT«Cannot modify an immutable Num␤ in block <unit> at /tmp/U86QMCcP6G:1␤␤»
timotimo Skarsnik: yes
m: say e
camelia rakudo-moar 895546: OUTPUT«2.71828182845905␤»
timotimo m: say i
camelia rakudo-moar 895546: OUTPUT«0+1i␤»
krshn m: my pi = "hi"
camelia rakudo-moar 895546: OUTPUT«5===SORRY!5=== Error while compiling /tmp/TyetOZDbok␤Malformed my␤at /tmp/TyetOZDbok:1␤------> 3my pi7⏏5 = "hi"␤»
nebuchadnezzar interesting, in REPL I got “(True True True)” at the end of my chained map 23:52
Skarsnik m: my constant pi = "hi";
camelia ( no output )
23:52 smls_ left
krshn so you cannot use pi as a variable in code? 23:53
timotimo m: my \pi = 100
camelia ( no output )
timotimo you can
m: my \pi = my $ = 4; pi++; say pi 23:54
camelia rakudo-moar 895546: OUTPUT«5␤»
Skarsnik most variable have sigil so..
timotimo if you just have "my \pi = 4", you'll end up with something write-once. if you put a scalar in there, however, it just means you can't replace the scalar with another scalar. but the value inside is changable
hankache krshn but pi is already defined
krshn m: my $pi=3; say $pi; 23:55
camelia rakudo-moar 895546: OUTPUT«3␤»
krshn duh..
:-)
what does \pi do?
hankache m: say pi;
camelia rakudo-moar 895546: OUTPUT«3.14159265358979␤»
timotimo that's how you define a sigil-less variable
krshn sigil-less variables. hmm interesting.. 23:56
dalek ast: 06bacaa | lizmat++ | S32-temporal/DateTime.t:
Add test for very large epochs, labster++
timotimo yeah, they are a SSA kind of thing - unless the thing you put into them is a scalar container like i showed above 23:57
krshn m: my \hello = "world"; say hello ~ " world";
camelia rakudo-moar 895546: OUTPUT«world world␤»
krshn m: my \hello = "world"; hello ~= " world"; say hello; 23:58
camelia rakudo-moar 895546: OUTPUT«Cannot modify an immutable Str␤ in block <unit> at /tmp/8W_daerOWu:1␤␤»
lizmat will look at the rest of zefram++ 's DateTime related tickets tomorrow
for now, good night #perl6!
krshn m: my \hello = "world"; \hello ~= " world"; say hello;
camelia rakudo-moar 895546: OUTPUT«Cannot assign to an immutable value␤ in block <unit> at /tmp/C4_V7y_nMR:1␤␤»
krshn so sigil-less are always immutable? 23:59

Powered by Cro and the Raku® Programming Language.

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

Thank you!