»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or rakudo:, std:, or /msg camelia p6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by masak on 12 May 2015.
ShimmerFairy japhb: IIRC they're sitting in specs/ . I think discussion on that simply died down 00:00
japhb OK, gotcha. 00:04
00:09 xinming_ left
psch m: my $x = :{1 => 4}; say $x.perl 00:11
camelia rakudo-moar c6a50f: OUTPUT«:{1 => 4}␤»
00:12 xinming joined
psch ShimmerFairy: i'm similarly suprised 00:12
949d82cd4850e5f4d2f0911d136312d3650272b6 is "make :{} mean object hash" 00:13
...except it doesn't
...except it doesn't, if there's no % sigil
s:2nd/doesn't/does/
00:14 n0tjack joined
psch m: my %h = :{1 => 4}; %h{2} => 5; say %h.perl 00:18
camelia rakudo-moar c6a50f: OUTPUT«{"1" => 4}␤»
psch m: my %h = :{1 => 4}; %h{2} = 5; say %h.perl
camelia rakudo-moar c6a50f: OUTPUT«{"1" => 4, "2" => 5}␤»
00:19 n0tjack left
skids m: my %h := :{1 => 4}; %h{2} = 5; say %h.perl 00:20
camelia rakudo-moar c6a50f: OUTPUT«:{1 => 4, 2 => 5}␤»
skids The default hash is not an object hash. 00:21
00:21 laouji joined
ShimmerFairy which seems like a bug to me 00:21
skids No it's by design. 00:23
psch m: my %h{Any} = 1 => 2; say %h.perl
camelia rakudo-moar c6a50f: OUTPUT«(my Any %{Any} = 1 => 2)␤»
psch ehh, there was some non-binding things iirc 00:24
i might be mistaken, i've been awake a bit :)
skids That... may be a bug. I thought I remembered that working.
psch s/things/syntax/
m: my %h := :{1 => 4}; %h{2} = 5; say %h.perl 00:25
camelia rakudo-moar c6a50f: OUTPUT«:{1 => 4, 2 => 5}␤»
ShimmerFairy skids: as far as I'm concerned, it's a design bug if the methods I tried don't produce an object hash like I expect.
psch imo assigning an :{} should turn it into a Hash where the key can be Any
and i'm pretty sure that's intended with the commit i mentioned above
i mean, otherwise the syntax is nigh-useless 00:26
or rather, useless in the obvious way
(s/in/for/ maybe?)
star: my %h = :{ 1 => 2 }; say %h.perl 00:27
camelia star-m 2015.03: OUTPUT«("1" => 2).hash␤»
psch shrugs
00:27 cognominal left
psch star: my %h{Any} = { 1 => 2 }; say %h.perl 00:27
camelia star-m 2015.03: OUTPUT«Hash[Any,Any].new("1" => 2)␤»
psch *that* is extra-weird
but eh :P
skids m: my @a = array[int].new(1,2,3); @a.perl.say; array[int].new(1,2,3).perl.say; # would you epect that to do similarly?
camelia rakudo-moar c6a50f: OUTPUT«[1, 2, 3]␤array[int].new(1, 2, 3)␤»
psch i actually would, i think 00:28
ShimmerFairy yes. If I'm assigning a Positional or Associative to a @ or % declaration, I'd expect the variable to adapt, not the assigned thing
psch skids: the thing i'm wondering is, if that's not how i'm supposed to achieve that, how *am* i supposed to achieve it? 00:29
skids But, type constraints are tied to the variable.
psch skids: mind, binding is a "only if you really know what you're doing" thing
(which isn't to say that binding correctly solves it, but i'm wondering how to declare the var, for % specifically) 00:30
skids Oh I would not say that about binding. Only about sigilless variables :-).
Yeah I would have expected %h{Any} to have worked.
I mean Any,Any. Oh that was why. 00:31
ShimmerFairy I would not think of a %h{Any} declaration as the way to get an object hash. That feels really wrong to me as the only way.
00:38 laouji left, laouji joined
ShimmerFairy At least for hashes, I'd expect assigning an object hash in a declaration to be sufficient. And if not that, then at least my %h; %h{1} = 2 should _definitely_ not stringify the key I gave :) 00:41
psch design.perl6.org/S03.html#Term_precedence 00:42
00:42 laouji left, laouji joined
psch about 3/4th of a page (on my screen) down 00:42
"By default a hash forces all its keys to be strings. To compose a hash that allows arbitrary objects (such as numbers) as keys, add a colon on front:" 00:43
which means we have a bug
00:43 z448 joined
psch 'cause %h = :{1 => 2} should give an object hash 00:43
i'm not awake enough anymore to think about "what if the first key we use when assigning a value isn't a string?" 00:44
00:44 aborazmeh joined, aborazmeh left, aborazmeh joined
psch which is to say i'm not giving an opinion on "my %h; %h{1} = 2 should _definitely_ not stringify the key I gave" 00:44
ShimmerFairy If I didn't give a stringy key, the last thing I expect is for it to be treated like a stringy key :) 00:45
00:47 n0tjack joined
skids I think that part is a bug. design.perl6.org/S02.html#Hash_subs..._bare_keys 00:49
dalek kudo-star-daily: 46e96c4 | coke++ | log/ (8 files):
today (automated commit)
skids Unless someone decided the default key type is "Str(Any)" 00:50
psch m: my %h = :{ 1, 2 }; say %h.perl 00:51
camelia rakudo-moar c6a50f: OUTPUT«Odd number of elements found where hash initializer expected␤ in block <unit> at /tmp/1qQPkP2oSB:1␤␤»
00:51 n0tjack left
psch m: my %h = :{ Pair.new(:key(1), :value(2) }; say %h.perl 00:52
camelia rakudo-moar c6a50f: OUTPUT«5===SORRY!5=== Error while compiling /tmp/UQvjls8CSI␤Unable to parse expression in argument list; couldn't find final ')' ␤at /tmp/UQvjls8CSI:1␤------> 3my %h = :{ Pair.new(:key(1), :value(2) 7⏏5}; say %h.perl␤»
psch m: my %h = :{ Pair.new(:key(1), :value(2)) }; say %h.perl
camelia rakudo-moar c6a50f: OUTPUT«Odd number of elements found where hash initializer expected␤ in block <unit> at /tmp/BQuXNMZ3H6:1␤␤»
psch yeah uhhm
psch slowly steps away from the keyboard
ShimmerFairy I think the base issue might possibly be that the default Associative type has such a specific constraint (Str), where no other sigil type has as specific a default :) 00:54
(And fwiw I'm not about to suggest not making Str keys the default, since I can predict a lot of things would break for that)
skids It's an expectation born of a host of predecessr languages. 00:55
ShimmerFairy good thing P6 has a habit of breaking expectations :P
skids I think that not having string keys by default would be a hard pill for a lot of casual PHP/shell coders to swallow. 00:58
(on top of the several other paradigm shifts)
ShimmerFairy skids: I think Str keys by default does a lot of good in most cases, it just makes wanting object keys harder. It certainly seems tricky on the surface. 01:00
"It" being the question of how to make it easier, that is :) 01:01
skids I don't have warm fuzzies for that new .perl syntax for %h{Any} yet, I don't understand why it's necessary. 01:02
ShimmerFairy skids: it's so that using :{} will produce the same kind of output for .perl (and if you didn't use :{}, it tells you about a neat new syntax you can use instead!) 01:04
skids (Isn't the only reason to use %h{Any} that you want type safety on things bound to %h s nobody binds a Hash[Str,Any] or whatnot to it?) 01:05
I mean, use it as opposed to %h := :{}; 01:07
ShimmerFairy skids: if you ask for keys of type 'Any', then you can bind anything that ultimately is Any. What syntax you use (%h{Any} or :{}) shouldn't make a difference. 01:08
01:08 z448 left
ShimmerFairy Hm, where did $file.open(:w); $file.say() go? I could swear that used to exist 01:08
skids ShimmerFairy: of course it matters. One is a type constraint on a variable name, the other is a roperty of the object itself. 01:11
m: my %h := :{}; %h{1} = 2; %h.perl.say; my %b{Any} = :{}; %b{1} = 2; %b.perl.say; %h := { foo => "bar" }; %h.perl.say; %b := { foo => "bar" }
camelia rakudo-moar c6a50f: OUTPUT«:{1 => 2}␤(my Any %{Any} = 1 => 2)␤{:foo("bar")}␤Type check failed in binding; expected 'Associative[Any,Any]' but got 'Hash'␤ in block <unit> at /tmp/g_ttEHGn13:1␤␤»
ShimmerFairy o_o but... a Str hash should match an Associative[Any, Any] constraint, since Str ~~ Any 01:12
skids No, I don't think paramterized types recurse typechecks into parameters. 01:13
ShimmerFairy m: say Hash ~~ Associative[Any, Any]
camelia rakudo-moar c6a50f: OUTPUT«False␤»
ShimmerFairy o_o 01:14
psch m: my %h; %h{1} = 2; say %h.perl # this i'd also expect to type check fail, fwiw
camelia rakudo-moar c6a50f: OUTPUT«{"1" => 2}␤»
psch 'cause it's a literal Int...
vOv
Cool not withstanding
or maybe it does :/
skids psch: The last spec link I posted would seem to agree.
psch skids: yeah, i think that's what brought me there 01:15
anyay, i need some rest o/
01:16 Akagi201_ joined 01:18 Akagi201_ left 01:19 n0tjack joined 01:20 Akagi201 left 01:23 n0tjack left 01:38 n0tjack joined 01:43 n0tjack left
ShimmerFairy $ perl6 -e 'say ?run("false")' 01:49
False
$ perl6 -e 'say ?run("false", :out)'
True
That's... not good.
01:51 kaare_ joined 01:52 n0tjack joined
ShimmerFairy a closer look says that the exitcode itself is getting set wrongly. Huh. 01:54
01:56 n0tjack left
skids Ahah. design.perl6.org/S09.html#line_1182 (and it has been that way for a long while.) 01:58
ShimmerFairy rakudobugs the exitcode thing 02:00
02:08 dayangkun joined 02:09 z448 joined 02:13 z448 left 02:24 AlexDaniel left, n0tjack joined 02:29 n0tjack left
ShimmerFairy m: class Foo { has $.bar is rw }; my @a = Foo.new(bar => 1); say @a; @a.push(@a[*-1]); @a[*-1].bar = 42; say @a 02:36
camelia rakudo-moar c6a50f: OUTPUT«[Foo.new(bar => 1)]␤[Foo.new(bar => 42) Foo.new(bar => 42)]␤»
02:36 Juerd left
ShimmerFairy known bug? 02:36
m: my @a = 1; say @a; @a.push(@a[*-1]); @a[*-1] = 42; say @a # for comparison 02:37
camelia rakudo-moar c6a50f: OUTPUT«[1]␤[1 42]␤»
ShimmerFairy m: class Foo { has $.bar is rw }; my @a = Foo.new(bar => 1); say @a; @a.push(@a[*-1].clone); @a[*-1].bar = 42; say @a # workaround, fixed with adding .clone
camelia rakudo-moar c6a50f: OUTPUT«[Foo.new(bar => 1)]␤[Foo.new(bar => 1) Foo.new(bar => 42)]␤»
02:55 Juerd joined
zostay is there a doc somewhere that tells me how to translate my pre-glr stuff to glr? i've got bugs everywhere, mostly because flattening is different than before 02:56
yoleaux 1 Sep 2015 07:17Z <cdc> zostay: I think you can replace "Supply.on-demand(-> $a { ... })" construction with a supply block. gist.github.com/jnthn/a56fd4a22e7c...onstructs. I find this latter construction more readable.
10 Sep 2015 15:34Z <timotimo> zostay: paul evans on twitter was complaining that p6sgi is "as disappointing as the perl5 one" and lacks support for streaming of input data and response backpressure and "..."
02:57 n0tjack joined
zostay thx for the feedback timotimo and cdc :) 02:57
02:57 BenGoldberg left
zostay lemme go dust off my twitter account 02:57
02:58 noganex joined 03:01 noganex_ left 03:02 n0tjack left
ugexe perl6 -e 'say ?run("false", :out).exitcode' 03:07
False
perl6 -e 'say run("false", :out).out.close.exitcode' 03:09
1
ShimmerFairy just noting that ?(run().exitcode) will return False on a successful exitcode :) 03:10
ugexe you have to close .out first 03:11
not that that isnt a bug of course 03:12
ShimmerFairy ugexe: I still cry foul on Proc.exitcode being wrong once you've opened streams. The equivalent bash redirects don't screw with the exit code, and I wouldn't expect redirects to mess with it in the first place. 03:14
03:22 n0tjack joined
colomon doh. what is cwd now? 03:23
ugexe $*CWD ? 03:24
colomon ugexe++ 03:25
03:27 n0tjack left 03:29 n0tjack joined 03:31 khw left 03:34 n0tjack left 03:36 khw joined 03:46 laouji left 03:47 aborazmeh left 04:02 n0tjack joined 04:06 n0tjack left
dalek c: a47ed84 | skids++ | lib/Language/subscripts.pod:
Explain convenience behaviors in associative subscripting section

  (could use some links, but too tired)
Also, provide an example of off-end access of a native array.		 04:09
04:16 aborazmeh joined, aborazmeh left, aborazmeh joined 04:25 rmgk is now known as Guest85186, rmgk_ joined, rmgk_ is now known as rmgk 04:29 Guest85186 left 04:34 n0tjack joined 04:35 khw left
skids Hrm was hoping docs would update the HTML before I konked out. Apologies for any markup errors, have to sleep. 04:36
04:38 n0tjack left 04:42 skids left 04:44 laouji joined 04:49 zhanggong joined, zhanggong left, X230-XXZX joined 04:50 X230-XXZX left 04:53 aborazmeh left
zostay anyone want to go on a wild goose chase for me? gist.github.com/zostay/13a073e62520b1041ecb 04:56
04:56 aborazmeh joined, aborazmeh left, aborazmeh joined
awwaiid hahainternet, psch: Thanks again for the reference earlier. I ended up with github.com/awwaiid/pinpoint-code-c...code-color (used 'shell', which works also but isn't mentioned in Proc) 05:01
05:06 n0tjack joined 05:10 nightfrog left, n0tjack left 05:11 baest_ is now known as baest
ugexe doc.perl6.org/routine/shell 05:12
05:23 nightfrog joined 05:34 nightfrog left 05:36 nightfrog joined 05:38 xfix joined, n0tjack joined 05:43 n0tjack left 05:49 domidumont joined 05:50 domidumont left 05:51 domidumont joined
JimmyZ zostay: how about put 'dd $_' inside for $node.nodes { 05:56
06:10 n0tjack joined 06:14 n0tjack left
[Tux] test 50000 40.466 40.386 06:16
test-t 50000 39.359 39.278
gooooood morning perl6! 06:17
so it looks like GLR is now consistently faster than nom was just before the merge
06:19 aborazmeh left 06:22 mohij joined 06:28 abraxxa joined
grondilu I don't see improvement on rakudo compile time, though. 06:31
TimToady compiler doesn't use lists much 06:32
JimmyZ compiler is nqp ... 06:33
06:33 abraxxa left
grondilu yeah good points 06:33
06:37 espadrine joined
masak good morning, #perl6 06:37
masak will be busy with $teaching for the rest of the day, but just wanted to say hi 06:38
ShimmerFairy masak: $teaching.WHAT :P 06:40
06:43 n0tjack joined 06:47 abraxxa joined, n0tjack left
TimToady jnthn: on shaped list assignment, we should probably require the input to be strucured; flat input can always be reshaped based on the arrays's shape, or using nested .rotor calls 06:53
map should try to produce something of the same shape as the input, even if map returns some number of arguments that is not 1, I suspect 06:54
.keys should likely produce tuples that can be used as .[||tuple] to retrieve an individual leaf 06:57
(you can get the top-level keys with .[*]:k anyway, so we don't have to put that behavior into .keys)
rotate and reverse make some sense on slices, but probably not on the whole array, unless we want some APLish trick of treating a shaped array as flat when convenient 07:00
but offhand I think we probably want to something explicit to open up a flat view, either for reading or writing 07:01
*to require
by and large, my bias is to treat the shaped array as a complete object, rather than trying to pretend it's an array of arrays 07:03
hence a preference for iterators to visit leaves, not just the top level
the shape being a convenient barrier to say what should be considered the leaves 07:04
07:06 FROGGS left, rurban joined 07:08 thou left 07:11 kjs_ joined 07:15 n0tjack joined 07:18 Ven joined, mohij left 07:19 FROGGS joined 07:20 lizmat joined, n0tjack left
timotimo TimToady: do you have a comment about "my %foo = :{ 1 => 1 }" giving you a non-object-hash? a justification for asking the user to use := instead of = 07:21
07:22 gfldex joined
FROGGS m: my Int %foo = :{ 1 => 1 }; say %foo.keys[0].WHAT # hmmm... 07:23
camelia rakudo-moar c6a50f: OUTPUT«Type check failed in binding key; expected 'Str' but got 'Int'␤ in block <unit> at /tmp/e7UF2oH5Ww:1␤␤»
TimToady I think we could issue a "Useless use of :{}" there 07:24
likewise even without the : 07:25
timotimo yes, we used to give a deprecation notice about "itemized hash assigned to hash variable" 07:26
07:27 zakharyas joined
TimToady that seems saner than making hashes an exception to one-arg or iteration 07:27
timotimo i think so, too
TimToady but nothing seems quite sane when I'm up this late...
07:28 darutoko joined
TimToady been hacking a version of rakudo that separates .cache from a (non-caching) .list, and almost have it working 07:28
we have a heck of a lot of places that call .list just to get a list currently, that accidentally also cache, so I think this is a good thing to clarify 07:29
still have a couple three bugs in the test suite, though, so probably can't finish tonight 07:30
timotimo are you going to push the WIP branch?
branches*
TimToady I suppose I could, if someone's interested in playing with it 07:32
timotimo i'm not guaranteeing it 07:36
07:38 Ven left
dalek kudo/list-v-cache: b410729 | TimToady++ | src/ (13 files):
early hackage to distinguish .cache from .list
 07:38
07:40 kjs_ left
dalek kudo/list-v-cache: 8911bc9 | jnthn++ | src/core/List.pm:
No, .infinite did not survive the GLR.
 07:42
rakudo/list-v-cache: 2af8cb4 | jnthn++ | src/Perl6/Optimizer.nqp:
rakudo/list-v-cache: Make optimizer aware of types of nqp:: ops.
07:42 dalek left 07:43 dalek joined, ChanServ sets mode: +v dalek, FROGGS left
moritz TimToady: thanks; the .list vs. .List distinction in Seq seemed a bit dubious to me :-) 07:44
TimToady oh, that's a bit of a fossil, could just use .list now
not related, but there seems to be one failing test due to eqv calling .perl on something with no .perl method, so we'll have to address that somehow 07:45
TimToady is going to bed now, however... 07:46
ShimmerFairy Don't most things have a .perl? O_o
♞ TimToady o/
07:46 FROGGS joined 07:48 n0tjack joined
timotimo some things just say "NameOfClass<$!WHICH>" 07:49
ShimmerFairy I could've sworn .perl was something defined on Mu, and thus quite hard to make disappear :) 07:50
timotimo yeah
m: Mu.perl
camelia ( no output )
07:53 n0tjack left, kjs_ joined
TimToady m: say $~MAIN.perl 07:53
camelia rakudo-moar c6a50f: OUTPUT«Method 'perl' not found for invocant of class 'Perl6::Grammar'␤ in block <unit> at /tmp/UVxqWLSUXb:1␤␤»
TimToady rafk & 07:54
08:00 lizmat left 08:01 mohij joined
ShimmerFairy oh yeah, NQP objects are how you make things like .perl disappear :P 08:02
08:03 kjs_ left 08:04 thou joined 08:08 thou left 08:09 cognominal joined 08:12 duncan_dmg joined 08:19 rhr joined 08:20 n0tjack joined 08:24 YouthEnter2 left, n0tjack left, rindolf joined 08:33 espadrine left
moritz fall back to === semantics for NQP objects? 08:33
ShimmerFairy I've run into problems of this nature with NQP objects before, so I think the better long-term solution would be to either replace NQP objects that can be exposed with P6 versions, or to make the appropriate NQP objects more P6-like (I'm thinking of things like .gist and so on) 08:38
08:38 kjs_ joined
timotimo we can just give the nqp objects the methods, that'd be enough i believe 08:39
hllize will take care of the rest
08:39 mohij left
ShimmerFairy m: say Cursor.^attributes # would that fix stuff like this too? 08:41
camelia rakudo-moar c6a50f: OUTPUT«Method 'gist' not found for invocant of class 'NQPAttribute'␤ in block <unit> at /tmp/y3xo3Vqmgt:1␤␤»
08:41 dakkar joined
timotimo probably 08:42
ShimmerFairy timotimo: I don't suppose we could redeclare bootstrapped private attributes in src/core, could we? For the bug I just gave, that would seem to be an alternate solution if it were possible :) 08:44
timotimo oof. 08:45
perhaps we can do monkey typing?
timotimo is ankle deep in JIT right now :)
ShimmerFairy Of course, letting NQPAttribute and the like behave more P6-y would make it a bit nicer, even if we'd like to not have NQP-level types like BOOTSTRAPATTR exposed :P 08:46
that Cursor example wasn't the best, come to think of it, because it's inheriting from an NQP class. The bootstrap issue will whine about BOOTSTRAPATTR instead :) 08:47
08:48 laouji left 08:49 laouji joined 08:50 leont joined 08:51 mohij joined 08:52 n0tjack joined
jnthn morning, #perl6 08:52
yoleaux 10 Sep 2015 21:45Z <FROGGS> jnthn: here is the current state of perl6-j, if you are curious: gist.github.com/FROGGS/3598992b921b34a1c6d5
jnthn
.oO( In the "what happens if I snort powdered naga jolokia" sense of curious maybe... )
 08:55
08:56 n0tjack left
FROGGS O.o 08:57
jnthn .ask FROGGS Don't suppose a $self := nqp::decont($self); at the start of that method helps? Dunno why it'd only be needed in some cases and not others though...
yoleaux jnthn: I'll pass your message to FROGGS.
FROGGS jnthn: I tried a nqp::decont($attr) in the bindattr_s call that fails 08:58
yoleaux 08:57Z <jnthn> FROGGS: Don't suppose a $self := nqp::decont($self); at the start of that method helps? Dunno why it'd only be needed in some cases and not others though...
FROGGS jnthn: (without luck)
jnthn Or rather, why only on moar but not JVM
Ah, OK :(
What's really werid is that the first bind works o.O 08:59
First bindattr, I mean
FROGGS it doesnt
jnthn Oh wait
Yeah, the first debug line is over the create
FROGGS aye
create works, bind doesnt
though funnily the bind does work while building the setting 09:00
it just fails afterwards, in user code or when precompiling nativecall
jnthn What the heck...
FROGGS I tell ya
09:01 FROGGS[mobile] joined
FROGGS[mobile] jnthn: sorry, have to eat cake now :/ 09:01
09:02 FROGGS[mobile]2 joined, lizmat joined
jnthn Best excuse ever :) 09:03
Enjoye!
09:03 domidumont left
jnthn [Tux]: Nice the performance is going in the right direction :) 09:04
[Tux] indeed \o/
and all tests still pass
jnthn :) 09:05
09:05 domidumont joined 09:06 FROGGS[mobile] left 09:07 andreoss joined
andreoss can I use NQP as Perl6 slang? 09:08
use Slang::NQP or something like that
moritz don't think so. 09:09
dalek kudo/nom: 524a763 | lizmat++ | src/core/traits.pm:
"cached" is one of the possible Routine traits
 09:11
lizmat m: class cached {}; sub a($a) is cached {} # weird 09:13
camelia rakudo-moar c6a50f: OUTPUT«5===SORRY!5=== Error while compiling /tmp/bFwHl1eyqJ␤Can't use unknown trait 'is ' in a sub declaration.␤at /tmp/bFwHl1eyqJ:1␤ expecting any of:␤ rw parcel hidden-from-backtrace hidden-from-USAGE␤ pure default DEPRECATED inli…»
jnthn lizmat: If you name a type in a trait, its type object is passed, not a named arg
lizmat maybe we should use an "isa" trait to mark inheritance?
jnthn For the 100000000th time... 09:14
...no.
lizmat okokok
must have missed the discussion :-(
jnthn Sorry, I just saw it too many times
lizmat I forgot why not, or never knew it
is it performance of the grammar ? 09:15
jnthn Even if we do that, you'd *still* have this situation.
Well, it's more that it doesn't especially fix anything, and would come at the cost of breaking a huge amount.
timotimo yikes, so many failures
lizmat if we use "isa" only for type objects, and "is" for all the other traits, how would we *still* have this situation? 09:16
sorry if I'm asking for the obvious, but I'm really in the dark here wrt that
jnthn Because you'd still want "is" to pass the role when there is one
09:17 mohij left
lizmat ?? 09:17
09:17 espadrine joined
jnthn role foo { }; sub bar is foo { } 09:17
lizmat roles are set with "does" ?
ah, punned roles
09:17 ab5tract_ joined
jnthn no no, it's on a sub 09:17
That's actually going to call trait_mod:<is>(&bar, foo)
While if you just say 09:18
sub bar is notatype { }
You get
trait_mod:<is>(&bar, :notatype)
The named arg syntax is convenient and OK in the small
But we really should encourage the type-based approach for userland modules
As it's a good bit more robust 09:19
Probably especially in the face of having multiple versions of modules loading.
lizmat ah, so you're saying that things like "is cached" where cached is *not* a type, is really the problem?
moritz didn't even know about the type-based approach 09:20
jnthn :)
ab5tract_ We have types now? ;)
moritz idly roasts ab5tract_
jnthn lizmat: I'm saying it's useful to have and probably OK for some uses, but we probably want to guide module authors towards the type-based approach. 09:21
lizmat: Or not, and just see if it emerges as a best practice anyway ;)
jnthn has a fix for the S17-lowlevel/threads.t regression he caused 09:23
lizmat jnthn++
09:24 n0tjack joined
lizmat so suppose I have code that has an "is foo" trait on a sub 09:25
later, the system adds a "foo" class
all of a sudden my code will be broken after that update
jnthn Aye, though given (except native types) class names culturally start uppercase, it's not likely 09:27
The convention hinted at in S14 is that lowercase role names are typically for use as traits
lizmat ok, well... I need to hit the road again to make it in time to Lyon 09:29
moritz
.oO( into the Lyon's den )
09:29 n0tjack left
jnthn lizmat: Safe travels 09:29
lizmat thank you 09:30
I'll get back to this... :-)
jnthn :)
lizmat commute&
09:30 lizmat left
jnthn wonders how on earth S15-nfg/many-threads.t got busted 09:31
It's a reliable SEGV here
Wow, MVM_CROSS_THREAD_WRITE_LOG is pretty loud on this one. 09:35
And all inside of HYPER, matching the analysis in the backlog. :) 09:36
09:40 zakharyas left 09:42 brrt joined 09:43 zakharyas joined 09:46 FROGGS[mobile]2 left, laouji left, laouji joined
timotimo there's some "todo passed" in the spec tests, but also a whole bunch of failing tests :\ 09:48
jnthn "whole bunch"? 09:49
timotimo a screenful 09:50
the previous results have disappeared into the scroll buffer, but i'm running a new set right now
gather.t seems to fail, for example
jnthn Odd...no sign of that here 09:51
timotimo how about t/spec/S02-literals/adverbs.rakudo.moar ....................... Dubious, test returned 255 (wstat 65280, 0xff00)
t/spec/S02-literals/numeric.t ................................. Dubious, test returned 4 (wstat 1024, 0x400)
t/spec/S02-types/WHICH.t ...................................... Dubious, test returned 1 (wstat 256, 0x100)
jnthn Only the first of those, which looks like an NQP object leaking into Perl 6 space 09:52
09:52 thou joined
timotimo pairs-as-lvalues? 09:52
failed 2 of 5
jnthn No
timotimo weird! 09:53
i'll double-check for unclean checkouts
oh
am i supposed to be on the master branch of glr?
most probably >_>
moritz "master branch of glr"? 09:54
timotimo er
nine adverbs has been failing here for a day or two
timotimo the master branch of roast
nine yes
FROGGS timotimo: also keep in mind that a "screenful" differs in meaning here :o) 09:55
timotimo :) 09:56
09:56 n0tjack joined, thou left
timotimo i meant to give a rough estimate 09:56
nine A screenful here would be 100 files ;) 09:57
FROGGS files?
timotimo wow
nine lines
timotimo ah
FROGGS I have about 64 lines 09:58
timotimo one error gets reported over like 3 or 4 lines
FROGGS is jnthn still at 80x25? :P
timotimo one failing file, more correctly
nine FROGGS: AFAIK even the windows terminal can do more lines, just not more columns ;)
psch FROGGS: the "bind_attribute_native" call that dies in your gist works if it gets a hint for the correct field...
FROGGS: i have no idea if that's actually useful to know, though :) 09:59
FROGGS: mind, that's also in jdb, not with any kind of patch that figures out the right hint...
FROGGS psch: I think it very much is
psch: still I dont know how to debug bind_attribute_native with st.NO_HINT :/ 10:00
I like to add debugging statements or some such, but that's problematic how bind_attribute_native is implemented
10:01 n0tjack left
FROGGS psch: that it works with a hint explains why the decont hack does not work out which jnthn++ mentioned 10:01
psch FROGGS: what i've done is 1) uncomment the bytecode dumping, 2) extend RuntimeException with SpecialRuntimeException in P6OpaqueBaseInstance, 3) throw that instead of RuntimeException in resolveAttribute, 4) breakpoint on that exception type, 5) read __P6Opaque__68.class bytecode 10:02
s/i\'ve done/i'm doing/
FROGGS hmm
I should probably learn how to jdb
psch FROGGS: i suppose compiling nqp with -g to javac allows jdb to read that source
FROGGS: 'cause it keeps telling me to do that when i want to "list" generated bytecode... 10:03
FROGGS: although we'd have to inject that into the classfilewriter from ASM...
(fsvo "inject")
FROGGS >.<
timotimo only three failing test files now 10:10
psch eh, actually that's bogus anyway
timotimo adverbs, many-threads and i had to kill start
psch i mean, "javac -g" adds debugging annotations that come from the code that's compiled 10:11
FROGGS timotimo: that sounds about correct
psch ...but we don't actually have code that's compiled except what's in P6Opaque{...}.java
FROGGS hmmm
psch and that code's not particularly useful 'cause it's mostly throwing exceptions :P
10:12 TEttinger left 10:13 domidumont left
psch hmm, apparently we can use visitLabel and visitLineNumber to attach line numbers to bytecode segments... 10:14
FROGGS listens carefully 10:19
psch FROGGS: mail-archive.ow2.org/asm/2012-02/msg00003.html is what made me think that 10:21
FROGGS: i'm not sure how useful it is for us, because we still don't actually have source... and some things you're allowed to do in jvm-bytecode don't map to java code either
FROGGS nods 10:22
psch huh 10:24
gist.github.com/peschwa/9072d94296c74114aa7c
that's the first few lines of __P6Opaque_68.bind_attribute_native
__P6Opaque__68 is Attribute 10:25
FROGGS psch: what's the 'huh'? I don't spot it
psch FROGGS: the "huh" is in jdb, actually...
FROGGS bah! :P
psch sorry :s
FROGGS *g*
psch revision 2 of the gist has the jdb bit i'm looking at 10:26
(i wanted to add a file but UI confusion...)
basically, .delegate (which is .field_10) is null 10:27
and we pass STable.NO_HINT
which is -1
so we aren't actually looking for the attribute "$!name", we already know we can't find it 10:28
because we get bad info, somehow
10:28 laouji left
psch i'm thinking .delegate probably shouldn't be null... 10:28
oh, one error in there 10:29
.field_10 isn't .delegate
10:29 n0tjack joined, laouji joined
Ulti m: say $*IN.get; 10:29
camelia rakudo-moar 524a76: OUTPUT«Céad slán ag sléibhte maorga Chontae Dhún na nGall␤»
psch .delegate just is the 10th field in __P6Opaque__68 10:30
FROGGS psch: none of the 15 fields are .delegate, right?
psch FROGGS: right, the field_* thingies are P6-level attributes
FROGGS aye
psch (or nqp-and-higher rather)
FROGGS from BOOTSTRAP.nqp line 107 to 121
they are: str int int int int Mu Mu Mu Mu Mu int int int int Mu 10:31
fits very well the null/0
psch FROGGS: yeah, it should :)
FROGGS: cause that's what defines those fields
10:32 laouji left
andreoss moritz: would it be a sane idea to have something like Slang::NQP or Inline::NQP or I'm missing something here? 10:32
10:32 laouji joined
FROGGS andreoss: what do you need t for? 10:32
it*
andreoss If I have NQP code that runs much faster than Perl6's one
FROGGS hmmm
10:33 n0tjack left
FROGGS I guess a Slang::NQP should be doable with some hackory 10:33
psch but i'd only be faster if you pass it to NQP::Compiler instead of Perl6::Compiler, no? 10:34
andreoss or maybe i have a chunk of code that operates only on natives and I don't want unnecessary object wrapping to occur
FROGGS psch: so we want to find out why .delegate is null if it is null?
10:34 brrt left
psch FROGGS: yeah, i'm pretty sure .delegate being null there is why we don't get anything useful out of bind_attribute_native 10:34
FROGGS k
10:34 pat_js joined
psch FROGGS: i suspect something somewhere in the nqp::create() call... 10:34
FROGGS psch: the create call just allocates the object 10:36
Ulti do you not get $*IN when you do perl6 -e '' ?
and if not why not
FROGGS Ulti: you probably need to pass it a filehandle 10:37
psch oh, right, the "huh" 10:38
Ulti notices the double quotes in his one liner ;___; BASH mad $* an empty string and Rakudo complained about IN
timotimo andreoss: we're supposed to let you work with natives sufficiently well so that the perl6 code reaches the speed of the nqp code ... at least somewhat
10:38 telex left
psch FROGGS: the "huh" was: "hints are for fast field access" but the decompiled bytecode says "check the delegate, if that's null check the hint" 10:38
Ulti might be a good place for a more awesome error message though
psch FROGGS: i assume the object should be allocated with a delegate in place, or we need hints
jnthn iirc, the delegate is used for mixin related bits and maybe for deserialized objects, but not for other cases 10:39
10:39 mprelude joined
FROGGS P6opaque.change_type looks interesting 10:40
10:40 telex joined
psch jnthn: yeah, that's what the comments say as well, which means the codegen is probably wrong 10:42
jnthn psch: It should be something like "if there's a delegate then use it, otherwise use the hint, and if there's not one then fall back to the name-based lookup" 10:43
psch jnthn: oh, of course... which means i'll have to pull apart the name based lookup in the byte code :s 10:44
jnthn psch: I think the name-based lookup is implemented in the P6opaque instance base class
Rather than code-gen'd
psch jnthn: yeah, it's resolveAttribute... which is what throws
ohh 10:45
the nameToHintMap is wrong 10:46
FROGGS how so?
psch hold on 10:47
i had the contents a bit ago, but i forgot how :P
hm, no, it has "$!name", must've missed that last time 10:49
10:50 Ven joined
psch hm 10:50
we have on classHandle in the REPRData 10:51
with an ST with the same ID as the classHandle we're looking for
but they have different SCs
FROGGS O.O 10:52
psch s/on/one/
gist.github.com/peschwa/9072d94296c74114aa7c
plus this: 10:53
main[1] print rd.classHandles[0].st == classHandle.st rd.classHandles[0].st == classHandle.st = true
missing a \n between "ndle.st" and "rd.class"...
wonky paste :s
FROGGS I can read it anyway :o) 10:54
psch and that's why we don't find nameToHintMap
which is why we don't find the attribute
10:55 kjs_ left
psch i guess i'll blindly try only comparing the STables 10:56
because i don't really have a grasp on what serialization contexts do... :)
FROGGS this is scary
10:57 AlexDaniel joined
FROGGS dont several classHandles share a STable? 10:57
AlexDaniel not sure why this RT was not deleted for 12 hours… 10:58
FROGGS AlexDaniel: hi, what RT are talking about?
timotimo STable combines a type object and a REPR, dunnit?
AlexDaniel FROGGS: rt.perl.org/Public/Bug/Display.html?id=126035
psch FROGGS: uh, i'm not sure. i was assuming what timotimo says is true, which means every ST is unique to one combination 10:59
timotimo oh
one per HOW + REPR pairing
FROGGS psch: yes, but if you have several object of a kind, they share an STable
AlexDaniel :)
timotimo yeah, HOW instance
FROGGS objects* 11:00
timotimo which means it'd be per class
FROGGS hmmm
psch FROGGS: well, they also need the same hints then
FROGGS should that be good enough here?
AlexDaniel: I marked it as spam...
AlexDaniel FROGGS: yeah, I see :) 11:01
psch FROGGS: i think it should be, we're only looking for "how can we access the value", not "what is the value"
AlexDaniel thanks
11:01 n0tjack joined
psch still curious how this suddenly come up... 11:01
the code in questions is almost 2.5 years old :S 11:02
FROGGS psch: yeah, I cannot imagine how the Attribute belongs to another SC after building the setting
timotimo AlexDaniel: she says she still believes in signs. does that mean she's fond of languages with sigils?
FROGGS timotimo: or signs as in prefix ops? 11:03
timotimo could be
AlexDaniel timotimo: yeah, we can probably send some Larry's pics
timotimo www.expeditedssl.com/aws-in-plain-english - this is partially amusing 11:04
FROGGS I mean, it is good that we can lexically scope here belief :o)
timotimo especially the description of Direct Connect
11:05 n0tjack left
psch building setting with the patch now 11:06
11:07 itz joined 11:08 domidumont joined
psch well, it builds NativeCall at least :) 11:10
actually it even finished the build!
FROGGS psch: please run 'make j-test'
AlexDaniel It's like: Stacking cash on the sidewalk and lighting it on fire – haha 11:11
FROGGS psch: do you still have a modified 'for flat nqp::hllize' in there?
11:11 domidumont left
psch FROGGS: yeah 11:12
11:12 domidumont joined
psch FROGGS: also, NativeCall builds but apparently fails literally every test 11:12
oh, wait
t/04-nativecall/16-rt125729.t ........ ok
every other fail is all fails, all with "Bad plan" 11:13
FROGGS that tests typed CArrays IIRC
psch (Wstat: 0 Tests: 0 Failed: 0)
11:13 ibo2perl6 joined
timotimo o_O 11:13
how'd that happen
psch oh
FROGGS: actually, the traits.pm bit looks as before 11:14
FROGGS psch: can you commit or gist your patch?
psch FROGGS: yeah, let me clean up a bit first 11:15
ibo2perl6 m: my $bh = «a a a b b c».BagHash; $bh.perl.say; for $bh.kv -> $k, $v { say "$k => $v" } # this breaks one of my scripts
camelia rakudo-moar 524a76: OUTPUT«("a"=>3,"c"=>1,"b"=>2).BagHash␤a 3 => c 1␤Too few positionals passed; expected 2 arguments but got 1␤ in block <unit> at /tmp/UQsMFDWol9:1␤␤»
FROGGS psch: I'd like to try that then on the rakudo cu_load branch 11:16
psch: because then loading the setting and other stuff does not happen through CompUnitRepo anymore
ibo2perl6 m: my $bh = «a a a b c».BagHash;for $bh.kv -> $k, $v { say "$k => $v" } 11:17
camelia rakudo-moar 524a76: OUTPUT«a 3 => c 1␤Too few positionals passed; expected 2 arguments but got 1␤ in block <unit> at /tmp/5WqW0URw5O:1␤␤»
FROGGS psch: low level stuff is loaded via Perl6/ModuleLoader.nqp, and perl 6 modules a loaded via CompUnit(Repo)
dalek p: 52e409e | peschwa++ | src/vm/jvm/runtime/org/perl6/nqp/sixmodel/reprs/P6OpaqueBaseInstance.java:
Only compare STables when looking for the nameToHintMap.

For some reason the SC of a class apparently changes, which makes us unable to find the type hint needed to access attributes. This patch hacks around that.
FROGGS m: my $bh = «a a a b c».BagHash;for $bh.kv -> $k, $v? { say "$k => $v" }
camelia rakudo-moar 524a76: OUTPUT«a 3 => c 1␤Use of uninitialized value $v of type Mu in string context in block at /tmp/JYUrweds3T:1␤b 1 => ␤»
FROGGS m: my $bh = «a a a b c».BagHash;for $bh.kv -> $k, $v? { say "$k => $v".perl } 11:18
camelia rakudo-moar 524a76: OUTPUT«"a 3 => c 1"␤Use of uninitialized value $v of type Mu in string context in block at /tmp/p0XdQrtjxm:1␤"b 1 => "␤»
FROGGS m: my $bh = «a a a b c».BagHash;for flat $bh.kv -> $k, $v { say "$k => $v".perl }
camelia rakudo-moar 524a76: OUTPUT«"a => 3"␤"c => 1"␤"b => 1"␤»
FROGGS ibo2perl6: ^^
ibo2perl6 does that mean I should now flat it myself?
FROGGS ibo2perl6: aye 11:19
ibo2perl6 FROGGS++
FROGGS ibo2perl6: now less stuff gets flattened surpi^Wautomatically
11:19 amaliapomian_ joined
ibo2perl6 :D 11:20
psch FROGGS: patch is only in nqp, as dalek mentions
i'm afk for a bit o/
11:23 AlexDaniel left
FROGGS psch++ 11:23
tadzik does anyone run rakudobrew on windows? 11:26
FROGGS never tried 11:27
timotimo never tired
tadzik I've been told it works :) 11:29
timotimo that's good :)
11:29 Psyche^_ joined
[Tux] m: [[^2]].perl.say 11:30
camelia rakudo-moar 524a76: OUTPUT«5===SORRY!5=== Error while compiling /tmp/WasS9DTD5Q␤Unable to parse expression in bracketed infix; couldn't find final ']' ␤at /tmp/WasS9DTD5Q:1␤------> 3[[^7⏏052]].perl.say␤»
[Tux] is that intended?
m: [[^2],[^2]].perl.say
camelia rakudo-moar 524a76: OUTPUT«5===SORRY!5=== Error while compiling /tmp/60QNUllJJB␤Unable to parse expression in bracketed infix; couldn't find final ']' ␤at /tmp/60QNUllJJB:1␤------> 3[[^7⏏052],[^2]].perl.say␤»
[Tux] m: [[1,2]].perl.say 11:31
camelia rakudo-moar 524a76: OUTPUT«[1, 2]␤»
FROGGS [Tux]: seems to be an LTM fail
11:31 Ven left
FROGGS m: say [[^]] 11:31
camelia rakudo-moar 524a76: OUTPUT«one()␤»
[Tux] I have to leave in a sec. feel free to RT that yourself :P 11:33
11:33 Psyche^ left, n0tjack joined 11:34 kaare_ left 11:35 kaare_ joined 11:36 ilmari joined
ilmari trying to install panda with rakudobrew failed in t/tester.t 11:37
yoleaux 25 Sep 2014 20:07Z <nwc10> ilmari: lunch
ilmari hahahaaaa
yes, I have had lunch since then
11:38 n0tjack left
[Tux] panda pased this morning still 11:38
ilmari freshly brewed moar rakudo
oh, it passed on the second go 11:39
[Tux] I'm off. bbl
ilmari is here a preferred pastebot? 11:40
s/bot/bin/
11:40 thou joined
ilmari fpaste.scsys.co.uk/498559 11:40
tadzik :D 11:41
ilmari: we usually use github gist
ilmari: tester is a funny one since it simulates failing tests inside it, but doesn't surpress console output
(because back in the days there wasn't a way to do that, and it's quite possible that it still isn't :)) 11:42
so it always looks confusing
but now it seems like it failed for real
ilmari but it definitely failed in that run (no panda installed)
tadzik very interesting 11:43
but the second try worked?
I blame sunrays
ilmari yes
itz did it fail at the reverse dependency building step?
ilmari itz: see the paste
itz no 11:44
ilmari no?
tadzik revdeps are ded
and thankfully
11:45 thou left
itz oh thats good news 11:45
tadzik I thought it'd be a controversial decision but everyone seems to like it
weird :)
does anyone remember what IRC nickname github:patzim uses? 11:47
I remember it starting with miz... I think
ilmari: well, please let me know if you see it happen again, especially if reproducible
11:49 tadzik joined
ilmari tadzik: got it fail once more running prove -e perl6 -lv t/tester.t 11:51
tadzik oh, interesting
I get a full pass 11:52
on freshly built rakudo and panda
ilmari it only fails very occasionally
tadzik planned two tests but ran one, like it crashed midway between lives-ok() and dies-ok() 11:53
very weird
moritz that's why test plans are still useful 11:54
tadzik indeed
ilmari done_testing() and the non-zero exit would have caught that too
11:54 kjs_ joined
ilmari gist.github.com/ilmari/b6c28f43569f5d6d8a98 11:54
tadzik aha! 11:55
moritz what it does not catch is some refactoring not running all of your tests (but still exiting cleanly)
tadzik ilmari: are you using some new-ish prove perchance? 11:56
ilmari I'm not saying test plans are never useful, but it wouldn't make any difference in this case
TAP::Harness v3.35 and Perl v5.22.0
tadzik it looks like prove sees the result that the inner-prove produced and thinks it's the outer one
it's quite nasty all in all
ilmari tadzik: but only sometimes?
tadzik ilmari: ...apparently
I'm on 3.30 11:57
let me update and see if it happens to me then, sometimes
11:58 dayangkun left
tadzik nope, runs like a charm 11:58
ilmari it's only failed twice for me so far, out of tens of runs
running it in a loop now
failed again just now 12:00
while prove -e perl6 -lrv t/tester.t ; do :; done 12:01
tadzik it consistently never fails for me :/
nine ibo2perl6: for $bh.kv -> ($k, $v) { say "$k => $v".perl } is nicer I think 12:02
ilmari is there a readline-enabled repl for perl6? 12:04
nine ilmari: install Linenoise 12:05
ilmari ah, _obviously_ 12:06
12:06 n0tjack joined
ilmari searched for readline on modules.perl6.org and installed Readline, but that didn't help 12:06
tadzik heh 12:07
oh, I was to write m-b for linenoise
timotimo afk for ~4 hours 12:08
awwaiid is pleased with their first non-toy perl6 grammar, github.com/awwaiid/pinpoint-code-c...code-color 12:09
ilmari nine: wow, that makes the repl take a _lot_ longer to start up 12:10
from .2s to 2.2s
12:10 n0tjack left
tadzik :o 12:11
ilmari and I thought Devel::REPL was bad
12:11 n0tjack joined
tadzik module loading is very constly :/ 12:11
costly, even
jnthn Well, it is if you don't pre-compile... 12:12
tadzik yep
ilmari panda install doesn't?
tadzik I guess we should do something about it...
ilmari: it used to, but doesn't anymore
let me point you to the reason why
ilmari: www.nntp.perl.org/group/perl.perl6....g2216.html 12:13
ilmari tadzik: I see 12:14
12:14 rurban left 12:17 Mhoram joined, Mhoram left, n0tjack left
FROGGS tadzik: please wait for the cu_load branch to land before doing any precomp work... 12:18
tadzik okay
12:22 skids joined 12:24 skids left 12:27 ibo2perl6 left 12:31 garryBarlowsLeft left
[Coke] (80x25) of course jnthn would be. I also am, but i'm an old fart at this point. "I remember using a physical terminal to access the MTS at college..." 12:31
12:33 rurban joined
[Coke] (126035) it's spam. to deal with spam, don't delete, click the big red S in the menu, upper right hand side. 12:33
jnthn My eyes aren't (haven't ever been, won't ever be) good enough to do much more than 80x25 :) 12:34
[Coke] (well, it's red after you click on it) 12:35
12:35 brrt joined, Alina-malina left
FROGGS jnthn: there might come a time where you plug in cameras into your left ear, and attach them to your forehead :D 12:35
dalek Heuristic branch merge: pushed 24 commits to rakudo/cu_load by FROGGS 12:37
jnthn is a little curious if google glass might be interesting for him
12:38 n0tjack joined
jnthn "Show me a zoom on that bit of the real world" :) 12:38
FROGGS I've not yet seen one
[Coke] maybe we can get Perl 6 added to www.fileformat.info/'s "how to print"
FROGGS [Coke]: we can ask them
jdv79 what is cu_load for? 12:39
[Coke] m: say "\x[1f595]"
camelia rakudo-moar 524a76: OUTPUT«🖕␤»
FROGGS jdv79: separating Perl 6 level module loading from nqp level
jdv79 o
12:41 laouji left 12:42 n0tjack left
TimToady jnthn: see irclog.perlgeek.de/perl6/2015-09-11#i_11201188 for my nsa musings 12:46
jnthn TimToady: I saw them, thanks. :) I need to ponder them a bit, but a nasty Moar race condition stole my attention this morning. 12:47
TimToady yeah, and I'm .caching madly enough that I only slept 3 hours 12:48
psch FROGGS: are you looking at cu_load currently? 12:49
FROGGS psch: I was hoping it helps loading the setting more cleanly, but the opposite effect happened
jnthn TimToady: That is...not much sleep 12:50
[Coke] spw hasn't forgotten about the youtube video of the q&a, roman is working on it. 12:51
psch FROGGS: hmm, yeah i'm seeing another SOE, with similar code path to what we had with the nom jvm build...
FROGGS :/ 12:52
timotimo looking forward to that recording
12:53 sufrostico joined
TimToady jnthn: it's just jetlag, mostly, and I can take naps when I need 'em now that I'm home 12:54
jnthn ah, jetlag...
jnthn still feels like he's recovering from his last batch of travel 12:55
12:58 abaugher joined, Woodi left 13:00 Woodi joined
itz has anyone integrated rakudobrew with PS1 to display which version is in use? 13:00
[Coke] does panda not work when using a relatively-path'd perl6 to bootstrap? 13:03
tadzik hmm
yes
moritz [Coke]: did you 'make install'?
[Coke] ... yup, that seems to be a problem. (or maybe a perl6 not in your path)
moritz: yup.
trying to run the inline::perl5 tests, it's kind of a pain with a local dev copy. 13:04
ARGLEBARGLE. also, panda still not working with http proxy. 13:05
I think that's a Christmas blocker. 13:06
tadzik it is
if you have HTTP::UserAgent installed
[Coke] adding it to the list. :)
tadzik I think I even documented that :)
FROGGS WRTFM?
[Coke] gist.github.com/coke/38a589cec239409a4212 13:07
tadzik: how can I install HTTP::UserAgent ?
... with panda?
tadzik :}
jdv79 is Digest GLR'ed yet?
[Coke] right. I think that needs to be bundled, then.
jdv79 that blocked me from using IO::Socket::SSL recently
dakkar surprise of the day: @().elems == 1 13:08
jdv79 which is what H::UA uses
tadzik [Coke]: to my defense, you can tell panda to use http all the time if it's git being blocked you're after
jdv79 looks like not
[Coke] tadzik: still needs a proxy.
moritz m: say @().perl
camelia rakudo-moar 524a76: OUTPUT«(Nil)␤»
dakkar moritz: is that expected? 13:09
[Coke] is camelia updated past glr yet? checking...
looks like. 13:10
dakkar: what version of perl6 do you have?
13:10 n0tjack joined
dakkar [Coke]: 5746d 13:10
built a few days ago
jdv79 carving 6.0 out of roast sounds like a lot of work 13:11
[Coke] perl6 -version might be easier to figure out...
jdv79: yup
jdv79 that only a select few can even do
dakkar This is perl6 version 2015.07.1-776-g5746df2 built on MoarVM version 2015.07-108-g7e9f29e
[Coke]: but yes, past glr
[Coke] RT: 1,041 tickets, 14 [GLR], 11 testneeded 13:13
13:15 n0tjack left
[Coke] trying to see if I can find anything on RT#77644. this segfaults: ./perl6-valgrind-m -e 'EVAL q[] while 1' 13:16
(whereas ./perl6 just chews memory) 13:17
TimToady NB: after we merge list-v-cache, some subset of the GLRization will have to be revisited--anywhere we added a .list, that will need to change to .cache (unless it's before a .elems, which will .cache now, so .list.elems can just be .elems) 13:20
13:24 khw joined 13:26 pmurias joined
jnthn TimToady: I assume you updated the binder to use .cache instead of .list? 13:28
13:28 thou joined
TimToady yes, I even updated the jvm binder :) 13:29
(without testing it)
jnthn Nice 13:30
TimToady down to 1 test failure (that is my fault :)
well, I guess the .perl one is also my fault :)
13:32 aborazmeh joined, aborazmeh left, aborazmeh joined 13:33 thou left, hernanGOA joined
itz did the old [1, 2]<> have a name? 13:35
jdv79 isn't that the zen slice 13:36
moritz it is 13:37
TimToady it's one form of it 13:38
13:38 thou joined
TimToady .[] and .{} are also Zen slices 13:38
itz perl6advent.wordpress.com/2014/12/...zen-slice/ has {} as well
13:39 n0tjack joined 13:41 amaliapomian_ left 13:48 n0tjack left
dalek kudo/nom: 8dd779a | hoelzro++ | src/core/ (2 files):
Remove initialization of removed dynamic vars
 14:03
kudo/nom: 13fda77 | hoelzro++ | src/core/Exception.pm:
Fix recommendation for $^X
kudo/nom: 8004846 | hoelzro++ | src/Perl6/Grammar.nqp:
Error out if we see semicolon form without unit
14:06 lichtkind joined
colomon *** Error in `/home/colomon/.rakudobrew/moar-nom/install/bin/moar': double free or corruption (!prev): 0x00007fce30089ae0 *** 14:06
generated during the smoke test, but I don’t know by what. :(
14:07 n0tjack joined
daxim in p6 string model, do I still have the distinction between lax "UTF8" (possibly 72 bits) and strict "UTF-8"? it was occasionally useful to use unassigned codepoints beyond 1ffff. 14:07
14:07 brrt left
TimToady no, we don't use utf-8 internally anymore, and graphemes are (by default) stored in 32-bit ints 14:09
utf-8 is just an encoding now 14:10
jnthn The utf-8 decoder in MoarVM follows Unicode Corrigendum 9
14:11 echo`division joined
moritz m: say chr(0x1ffff + 1) 14:11
camelia rakudo-moar 800484: OUTPUT«𠀀␤»
14:12 n0tjack left 14:13 kjs_ left
TimToady
.oO(non corrigendum illegitimi)
 14:14
daxim I didn't know about that text: www.unicode.org/versions/corrigendum9.html I'm relieved that I can pass around non-characters for interchange
colomon X::AdHoc.new(payload => "Could not submit test report: Unexpected named parameter 'pretty' passed") 14:15
tadzik: ???
14:15 steffi joined
TimToady m: say (^100000000).EXISTS-POS(10000000) 14:15
camelia rakudo-moar 800484: OUTPUT«True␤» 14:16
tadzik :S
TimToady m: say (^100000000).EXISTS-POS(100000000)
tadzik colomon: oh hm
camelia rakudo-moar 800484: OUTPUT«(timeout)»
TimToady I have a fix for that too
tadzik colomon: update JSON::Fast
colomon tries panda reboostrap 14:17
jnthn TimToady: At least it can reify 10 million elements before timing out :)
colomon … which doesn’t exist
even when spelled correctly 14:18
moritz colomon: when you need a rebootstrap, you usually can't call panda anymore; which is why rebootstrap.p6 is a separate script (in the panda source dir)
colomon just asked rakudobrew to build panda again, looked like it brought in a bunch of changs for JSON::Fast
tadzik yeah 14:20
FROGGS jdv79: I just fixed Digest 14:23
grondilu: ^^ 14:25
14:28 aborazmeh left
colomon tadzik: that doesn’t seem to have cleared things up. :\ 14:29
tadzik :/
is smoker using JSON::Fast?
or ::TIny
14:30 n0tjack joined 14:31 cognominal left
colomon tadzik: neither 14:31
grondilu FROGGS++
dalek rl6-roast-data: 22e34e6 | coke++ | / (9 files):
today (automated commit)
colomon I think I’ve got emmentaler just using Rakudo’s built-in JSON
tadzik :o
colomon though of course it calls panda,
and uses whatever panda uses
there
colomon ponders completely blowing away his rakudo/panda setup and starting from scratch 14:32
dalek kudo/list-v-cache: 524a763 | lizmat++ | src/core/traits.pm:
"cached" is one of the possible Routine traits
 14:33
kudo/list-v-cache: 8bc2c40 | TimToady++ | src/core/traits.pm:
Merge branch 'nom' into list-v-cache
rakudo/list-v-cache: d46a494 | TimToady++ | src/core/Seq.pm:
14:33 dalek left
grondilu so many C<flat> :/ Will be always have to use this that often or is it just temporary? 14:34
14:34 dalek joined, ChanServ sets mode: +v dalek
grondilu *we 14:34
FROGGS grondilu: I think most of them will stay... 14:35
grondilu :/
[Coke] Only use it if you need to be flat. :)
14:35 n0tjack left
FROGGS grondilu: for example when you use meta ops (X, Z), they produce lists of lists 14:35
moritz had to add surprisingly few flat() calls in his modules 14:36
FROGGS grondilu: and when you subscript, it retains the requested structure... so you want to flat that out
grondilu ok
FROGGS m: my @a = ^10; say @a[^5,5..9]
camelia rakudo-moar 800484: OUTPUT«((0 1 2 3 4) (5 6 7 8 9))␤»
14:36 n0tjack joined
FROGGS stuff like that 14:37
colomon ’s biggest GLR issue so far has been Set.list no longer being the same as Set.keys 14:38
14:39 muraiki joined
TimToady yeah, I can argue that one both ways, but something to be said for consistency with bags and mixes 14:40
muraiki if I want to specify a type such as a Set of Str in a sub's signature, what's the correct syntax to do so?
colomon TimToady: I don’t mind the change, but I used Set.list a lot where today it needs to be Set.keys 14:41
14:41 n0tjack left
colomon the resulting failure states are pretty amusing. :) 14:41
14:41 n0tjack joined
TimToady jnthn: list-v-cache is ready to merge, no failures except known ones in adverb.t and litvar.t 14:43
shall we just bite the bullet, since it's likely to have some performance improvement?
plust the failure mode of using .list instead of .cache is fairly obvious 14:44
[Coke] sooner the better if we want it in the september release. 14:45
jnthn TimToady: litvar.t?
Odd
adverb.t isn't yours, I don't think
TimToady: Anyway, we can merge
TimToady well, except it was caused by my .perl change to eqv
jnthn Oh 14:46
:)
TimToady but that's independent
the litvar might possibly be my problem, but I think we can proceed anyway
jnthn The S15-nfg/many-threads.t one is because better code-gen made a certain race much more likely
14:46 n0tjack left
jnthn Yeah, I'd go for it 14:46
moritz jnthn: that's one of the best reasons for a regression I've heard of :-) 14:47
dalek kudo/nom: b410729 | TimToady++ | src/ (13 files):
early hackage to distinguish .cache from .list
rakudo/nom: 834c805 | TimToady++ | / (8 files):
rakudo/nom: Merge branch 'nom' into list-v-cache
14:47 dalek left
TimToady killed dalek :) 14:48
14:48 dalek joined, ChanServ sets mode: +v dalek
moritz in #perl6, you kill dalek :-) 14:48
TimToady anyway, it's merged
moritz TimToady++
TimToady: so the implication is that if you write function that's supposed to receive the iterator, you don't use a @-sigiled parameter anymore 14:49
right?
TimToady yes, that was already in GLR 14:50
jdv79 FROGGS: thanks
TimToady it just now calls .cache internally rather than .list
pmurias hoelzro: once I get @array.push working anything else will be stopping the merge? 14:52
vendethiel moritz: you mean you exterminate it ;-) 14:53
moritz vendethiel: I guess I do :-) 14:59
ugexe the new nom merge appears to break something for me but only on windows 15:00
hoelzro pmurias: I would like to see if I can get around the optimization removal in the MAST compiler; if I can't get it in the next few days, let's just merge, perhaps on Sunday?
15:02 n0tjack joined 15:04 domidumont left
pmurias hoelzro: it's currently only diabled when we are cross compiling 15:04
* disabled
dalek p/js-merge-wip: a7d006d | (Pawel Murias)++ | src/vm/js/QAST/Compiler.nqp:
Minor cleanup.
 15:05
hoelzro I know; I just would rather remove that "if cross compiling" check
call me particular =)
TimToady litvar.t passes after the merge, somehow, and jvm compiles \o/ 15:07
pmurias hoelzro: copy & paste would avoid that, but I would rather avoid that
15:07 n0tjack left
itz is there a version of dd which returns the type like this "dd $a" -> Str $a = "A" 15:09
15:09 dalek left 15:10 dalek joined, ChanServ sets mode: +v dalek 15:11 rurban left 15:12 laouji joined
FROGGS TimToady: jvm compiles? 15:13
TimToady: also NativeCall.pm?
but it will surely fail the tests from t/04-nativecall 15:14
[Coke] t/spec/S17-procasync/no-runaway-file-limit.t is running away 15:16
FROGGS O.o 15:17
ahh, is that the one where jnthn said the proper fix will be interesting?
15:17 captain-adequate joined
[Coke] gist.github.com/coke/4be19f57f3845d3713f0 - roast failures. 15:18
jnthn t/spec/S17-lowlevel/thread.t is fixed in MoarVM, but didn't bump yet 15:20
ugexe rakudobrew is broke on travis and windows now :( 15:22
15:24 TEttinger joined
[Coke] t/spec/S17-procasync/no-runaway-file-limit.t in make spectest hung. running by hand, get: 15:24
moar(85050,0x11812b000) malloc: *** error for object 0x7fa1eeecbdc0: pointer being freed was not allocated
15:28 FROGGS left, araujo left 15:29 cognominal joined 15:34 n0tjack joined 15:38 n0tjack left
dalek kudo/gmr: 278a719 | (Stefan Seifert)++ | src/Perl6/ (2 files):
Fix return 1, :a(2)

Though return is implemented as a function call, we may not promote pairs to named arguments, because we cannot demote them afterwards.
Fixes RT #75118		 15:45
nine The gmr branch is now free of regressions and actually fixes two RT tickets :) 15:46
jnthn nine++
TimToady ugexe: if it's reporting a Seq error, then probably you need to change .list to .cache somewhere
15:46 echo`division left 15:48 steffi left 15:49 domidumont joined, abraxxa left
ugexe TimToady: the program just abruptly stops with no error or anything. after running it more its not always, but random, although its always at the same spot, at something like: sub work(&code) { await start { &code } }; my $foo := work { gather for 1..3 { <some work here> } }; for $foo.list ... 15:50
TimToady well, could be threads, or it could be .cache; try changing the .list there to .cache maybe? 15:53
15:54 skids joined 15:57 kjs__ joined 16:03 zakharyas left 16:06 n0tjack joined
dalek p/js-merge-wip: 1b3369e | (Pawel Murias)++ | / (5 files):
Support calling methods on arrays.

Do this by wrapping them in special objects for method calls.		 16:07
16:07 mprelude left 16:10 n0tjack left
hoelzro pmurias: I'll see if I can fix the underlying issue; I'll take the condition check over copy/paste 16:19
itz m: [[1,]] 16:20
camelia ( no output )
itz m: dd [[1,]] 16:21
camelia rakudo-moar 0c79be: OUTPUT«[1]␤»
itz m: dd [[1],]
camelia rakudo-moar 0c79be: OUTPUT«[[1],]␤»
itz is that the single arg rule?
jnthn itz: yes 16:22
16:23 n0tjack joined 16:24 tokuhiro_ left 16:25 laouji left 16:27 n0tjack left 16:31 ab5tract_ left
pmurias hoelzro: keep in mind that serialize_sc must be also overridable, as we use that to avoid serializing twice while cross compiling 16:34
hoelzro pmurias: yes, that's fine 16:35
pmurias just returning an ast fragment instead of a null would remove the if 16:36
wouldn't be more elegant :( 16:38
16:39 n0tjack joined, andreoss left 16:43 n0tjack left
itz stevemynott.blogspot.co.uk/2015/09/...actor.html 16:44
moritz "LIST NOW IMMUTABLE" the example is a bad one, since pre-GLR died the same way 16:51
it's just that push, pop, shift, unshift are gone from List 16:52
itz ok I'll change that .. probably later since I'll batch up changes 16:53
ilmari Linenoise gets confused about the cursor position with multibyte utf8 characters
16:53 tadzik left 16:54 tadzik joined
ilmari it uses bytes when calculating the cursor position 16:54
not just the cursor position. it seems to operate entirely on bytes 16:55
16:55 dakkar left
ilmari if I type å backspace, it deletes the second byte of the character 16:57
and if I then hit enter, perl6 dies with ===SORRY!=== Malformed termination of UTF-8 string 16:58
ah: github.com/antirez/linenoise/issues/25
17:04 itz left
jnthn dinner & 17:05
[Coke] jnthn: I am planning a 20 mile hike tomorrow ending in a curry. Thought you might appreciate that. 17:08
17:10 kjs__ left 17:11 n0tjack joined, anaeem1_ joined 17:12 rurban joined, anaeem1_ left
flussence p6doc is blowing up in a test at lib/Perl6/TypeGraph.pm:89... I can get the failing test to pass via s:g/\./»./, but I don't know if that's the right way to go about it. 17:14
17:15 n0tjack left
moritz flussence: uhm, what kind of method calls need the >> now? 17:16
ilmari Method 'super' not found for invocant of class 'Array'
17:17 duncan_dmg left
moritz ilmari: callwith, callsame etc. to the rescue 17:17
ilmari (that's the error from lib/Perl6/TypeGraph.pm:89)
moritz huh 17:18
I thought it's a p5ism
17:24 n0tjack joined 17:25 tokuhiro_ joined 17:29 n0tjack left, tokuhiro_ left, brrt joined
nine _itz: you might want to mention, that a Slip slips into a list ;) my @a = 1, (2, 3).Slip, 4; # @a is now [1, 2, 3, 4] 17:29
brrt hmm 17:30
m: my @list = 1, (2, 3), 4; dd @list.flat.list;
camelia rakudo-moar 0c79be: OUTPUT«(1, $(2, 3), 4)␤»
brrt m: my @list := 1, (2, 3), 4; say @list.flat.list; 17:31
camelia rakudo-moar 0c79be: OUTPUT«(1 2 3 4)␤»
brrt why?
nine brrt: Array itemizes its contents
brrt is there any way to unitemize it 17:32
nine brrt: and itemization prevents flattening
brrt hmm
nine m: my @a = 1, (2, 3), 4; say @a.perl; my @b = @a.List.flat; say @b.perl; say @a.map(*.list).flat.list.perl; 17:33
camelia rakudo-moar 0c79be: OUTPUT«[1, (2, 3), 4]␤[1, 2, 3, 4]␤(1, 2, 3, 4)␤»
brrt ok 17:34
so .List makes it a list again, and it can flatten
.list also does this
nine m: my @a = 1, (2, 3), 4; say @a.perl; my @b = @a.list.flat; 17:35
camelia rakudo-moar 0c79be: OUTPUT«[1, (2, 3), 4]␤»
nine brrt: no, there's a difference between .list and .List ^^^
brrt yeah, but not in this aspect
m: my @a = 1, (2, 3), 4; say @a.list.flat.list; 17:36
camelia rakudo-moar 0c79be: OUTPUT«(1 (2 3) 4)␤»
brrt NO
thats not fair
17:36 telex left
nine brrt: .list on an Array really doesn't do anything. It just returns self, since Arrays are also Lists 17:36
brrt: .List gives you a real, pristine List
brrt perl6 has an unmatched ability to make sane programmers go crazy 17:37
17:37 n0tjack joined 17:38 telex joined
brrt in a good way? 17:39
nine Granted this .list vs. .List distinction is not by design. I just added .List because I got tired of writing the same code again and again and I really think, internals should return Lists as far as possible. That their implementation uses an Array to gather the results should be a detail that's best hidden. 17:42
brrt yeah, that's probably best 17:43
_itz nine: thanks I'll add that 17:44
17:44 n0tjack left 17:46 mprelude joined
brrt actually, it works out pretty well 17:46
17:56 brrt left 17:57 Alina-malina joined 17:58 brrt joined 18:00 espadrine left 18:05 n0tjack joined 18:09 n0tjack left
skids flussence: I had slready got past that and started on working on the next problem -- if you want I can commit the fixes so far. 18:11
flussence ahh, carry on. I can read the pod just fine for now :)
dalek c: e232576 | skids++ | lib/Perl6/TypeGraph (2 files):
Partway fix doc build after a GLR-related something or other
 18:13
18:22 FROGGS joined 18:26 tokuhiro_ joined
FROGGS o/ 18:27
muraiki \o
FROGGS what do I do now? I don't want to work on the jvm today :S 18:28
muraiki the lisp machine port 18:29
FROGGS O.o
muraiki I have used a bunch of languages on the jvm, but never actually learned java 18:30
I feel like I should learn it as the lingua franca of most business development, but it's so rarely used in the type of stuff I do (non-enterprise web apps) that I haven't really felt the need to
18:30 tokuhiro_ left
muraiki but I look forward to being able to use java libraries from p6 instead of from scala (too much time converting types around) or clojure (depending on how enterprisey the java code is, can be a nightmare) 18:31
psch FROGGS: you could try figuring out why my portaudio bindings segfault when trying to use jackd as audio driver, but don't when using alsa :P 18:32
FROGGS: although last i tried was about a week or two ago, so maybe something changed already
FROGGS psch: where is the code?
japhb FROGGS: I vote for the rakudo/cpp3 merge. ;-) 18:33
psch FROGGS: github.com/peschwa/Audio-PortAudio
FROGGS japhb: that's why I was working on the jvm...
japhb DANGIT.
FROGGS japhb: I need to port stuff to jvm, and thus need a working perl6-j
japhb: exactly
psch: do I get a commit bit? 18:34
psch FROGGS: sure, hold on
japhb is now considerably more annoyed at the JVM
psch FROGGS: done 18:35
FROGGS psch: how do I select the jackd backend? 18:36
psch FROGGS: you need to have jackd running, which might need some alsa configuration as well
FROGGS ohh
flussence doesn't jackd have some handwavey-autostart-magic feature like pulseaudio these days? 18:37
18:37 n0tjack joined
psch FROGGS: but in general it's "$ /usr/bin/jackd -R -P89 -dalsa -dhw:0 -r48000 -p256 -n3" 18:37
where -r -p and -dhw depend on your soundcard
wiki.archlinux.org/index.php/Pro_Audio#JACK 18:38
that heading should be enough to get it to a working-enough-for-trying-if-it-works state
or rather the paragraph i guess
flussence: i'm not aware of any such future 18:39
flussence oh, maybe the programs I was using did it internally...
18:39 Alina-malina left
psch flussence: might also be a distribution-thing... 18:40
18:41 darutoko left 18:43 yqt joined
FROGGS $ perl6 -Ilib test.pl 18:43
init: Host error.
Cannot locate symbol 'Pa_GetDefaultOutputDevice' in native library 'libportaudio.so'
18:43 yqt left 18:44 yqt joined
psch 19_20140130-1 is the portaudio version i was running against 18:44
18:45 Fleurety joined, n0tjack left, araujo joined, araujo left, araujo joined
FROGGS portaudio.h,v 1.5 2002/03/26 18:04:22 18:46
18:47 yqt left
FROGGS ohh, hold on 18:48
timotimo 2002? %-) 18:49
FROGGS I've also got a libportaudio.so.2
which is libportaudio2 19+svn20140130-1 18:50
psch .so{.\d+,} jungle :/
well, it's not quite a dll-hell, is it vOv 18:51
FROGGS *g*
18:51 sufrostico left
FROGGS psch: how can I tell that it is using jackd? gist.github.com/FROGGS/9f182b90d616e936a349 18:52
psch FROGGS: looks to me like it can't get any device 18:53
FROGGS: you might have to "pulseaudio -k" in case that's running
(and possibly disable to restarting daemon...) 18:54
FROGGS: i'd advice trying first without jackd whether it does anything then
FROGGS: you should then hear occassional crackles, because of lots of underruns...
skids Yes jackd put in autostart I think in a kneejerk reaction to "hey why won't anyone come over here and work on our code" when pulseaudio strated gaining steam. 18:56
Then they let the systwide daemon bitrot
Then pulseaudio turned around and said, hey, we really need a systemwide daemon for this.
And the sage never ends 18:57
*saga
pyrimidine we used to have sage in our backyard growing like crazy, so you could say our sage never ended 19:00
psch FROGGS: i've also made the experience that jackd when running doesn't appear as the default device to portaudio 19:01
19:01 yqt joined
psch FROGGS: which is why you'll probably have check with enum-devices.pl which device id the jackd client sits on 19:01
skids These days I uninstall everything and just use alsa dmix; my needs are not complicated and my patience is limited. 19:02
psch FROGGS: and replace :device(Pa_GetDefaultOutputDevice) with :device($id-you-found) in line 33
FROGGS psch: I think we need to fix that one too: gist.github.com/FROGGS/9f182b90d61...e1-txt-L17
psch FROGGS: right, iirc that's line 70 in test.pl 19:03
FROGGS: the assignment works via an intermediate variable, but not directly
FROGGS weird, but true 19:04
19:05 sufrostico joined 19:06 sufrostico left
FROGGS it seems to have connected to jackd now 19:06
but jackd gives: JackAudioDriver::ProcessGraphAsyncMaster: Process error
I don't hear cracks though or anything else 19:07
psch curious 19:13
it always either dump its core or segfaulted for me
[Coke] wonders again why we need str, gist, perl, dd (and now ddt) 19:14
psch do you have continuous output in the terminal?
i'm still expecting underruns, fwiw
oh but process error
FROGGS yeah, no idea what that means 19:15
_itz ddt is for debugging :P 19:16
psch FROGGS: did you adjust the constants to what you gave jackd?
SAMPLE_RATE and FRAMES_PER_BUFFER that is
FROGGS ummm
19:17 n0tjack joined
FROGGS now I did, but still get the process error 19:18
[Coke] _itz: maybe it should be part of the debugger, than. 19:19
*then.
psch FROGGS: did you check the device id and give that to Pa_StreamParameters? 19:20
_itz that was a jokey reference to the banned 1950s DDT .. dd is in core
psch FROGGS: aside from that i'd say "check with another jack-capable program", but "process error" sounds like it's our fault... 19:21
FROGGS psch: why do I have to do such much manually?
jercos audio is hard, let's go shopping?
psch FROGGS: because the bindings are 1/4th finished and i wanted to have a working base first :/ 19:22
FROGGS *g*
moritz doesn't like shopping
19:22 n0tjack left
psch also what jercos++ says 19:22
timotimo doesn't like shooping either
moritz in fact, I mostly prefer compiler hacking over shopping 19:23
psch FROGGS: thing is, from the list of available device the user has to chose one if the default isn't the right one
FROGGS psch: what device do I chose? default, pulse, system, hdmi or intel?
psch FROGGS: and with jack it never seems to be the right one... vOv
moritz (except when it's in an electronic or hardware store, and I have unlimited budget :-)
psch FROGGS: that depends on which device jackd says it has, but if you have pulse that might muck it up anyway... :/
19:24 n0tjack joined
FROGGS psch: jackd prints 'Acquire audio card Audio0', so, device 0? 19:24
psch: now I get a lot of output underflow and then it exits gracefully 19:25
psch FROGGS: but jackd says it got a connection? 19:26
FROGGS no, doesnt say anything
skids jackd never says anything useful except when you poke it with specially designed cattle prods (included). 19:27
FROGGS device=4 hangs, and when I ctrl+c, it prints something 19:28
ohh!
device=6 makes some peeping noise, and then segfaults :D
psch \o/
FROGGS and jackd prints process error :o) 19:29
19:29 n0tjack left
psch i have no idea how to reliably figure out the right device, fwiw 19:29
i've been guessing locally and got something similar :P
anyway, i'm pretty sure the program talking to jackd shouldnt be driven to segfault if it continuously underruns 19:30
moritz panda help; # resolve stage failed for help: Project help not found in the ecosystem
panda # without args: no output at all 19:31
19:31 espadrine joined
tadzik --help is there 19:31
that's what MAIN does by default
FROGGS psch: do you insist on pulseaudio? or do you just want to get sound running?
moritz not intellligent enough
tadzik the move to install being the default action is kind of like that :/
moritz: I think both of those cases should do help, and can be done to do so
psch FROGGS: portaudio. because low-latency and portable, and yes, i need the low latency at least
FROGGS adding a positional 'help' candidate won't hurt 19:32
19:32 n0tjack joined
FROGGS psch: have you played with libSDL_audio? 19:32
psch FROGGS: i haven't tried using it for what i wanted to use the portaudio bindings for, no 19:34
dalek c: ceed374 | moritz++ | lib/Pod/Convenience.pm6:
Fix flattening in pod-with-title
 19:35
moritz has anybody tried htmlify.p6 in perl6/doc post-glr?
19:36 n0tjack left
FROGGS psch: I think one problem is that moar tries to manage the mem of the output buffer in the callback 19:41
timotimo so it ends up calling free() on it?
19:41 n0tjack joined
FROGGS seems so 19:42
and before I changed the param to a CArray directly (used to be a Pointer we casted to CArray), it realloced the CArray I think
timotimo ugh 19:43
FROGGS I know we mark strings in callbacks, but I dunno yet if we mark other stuff as well 19:44
19:44 espadrine left
FROGGS now it segfaults during a GC run 19:45
19:45 domidumont left 19:46 n0tjack left 19:51 n0tjack joined
moritz m: Pod::Heading.new(:level(1), :contents['a', 'b']) 19:54
camelia ( no output )
moritz m: Pod::Heading.new(:level(1), :contents['a', 'b']).contents.^name.say
camelia rakudo-moar 0c79be: OUTPUT«Array␤»
19:56 n0tjack left 19:57 ggoebel joined 20:02 n0tjack joined
moritz m: say 'a b c'.subst(:g, /\s+/, '_') 20:05
camelia rakudo-moar 0c79be: OUTPUT«a_b_c␤»
20:07 n0tjack left
moritz I have a very weird thing here while debugging htmlify.p6 20:07
it dies with "Type check failed in binding <anon>; expected 'Positional' but got 'Str'"
the line it points it $someString.subst(:g, /\s+/, '_') 20:08
*it points to is
and before this I've added lines with warn(), say() and die() calls 20:09
and none of them seem to have any effect, except that they change the line number of the error
but it's clearly a run-time error, since other run-time statements before it are also executed
skids was beating his head on that earlier, too. 20:11
You have to carefully watch the line number, since there are probably multiple problem points in the code. 20:12
I found any kind of .map or such where the .subst is causes the issue.
Delete that .subst, and the error goes to another pat of the code. 20:13
moritz that is... decidedly ungood
skids Split the params of the subst up on different lines, and it ends up on the matcher/1st arg.
moritz plus ungood, even
skids And --ll-backtrace just tells you it happenned in bind, which we already knew. 20:14
What I can't figure out is how the error can sometimes report on the "once {" above which would seem to bear no resemblance at all. 20:17
moritz skids: do you get the same phenomenon that debug output before the error location doesn't seem to work? 20:18
20:19 nightfrog left
skids Yeah well, I threw a "say" in the concat and it did nothing. 20:21
20:22 nightfrog joined 20:23 n0tjack joined
moritz skids: it seems the error message points us to a totally wrong line 20:25
for @definitions -> [$sk, $name] {
@definitions.perl says ["Reference", "creation"] 20:26
so there's a level of nesting missing
skids Huh, I get mine on htmlify line 424ish
moritz skids: that's what the error message says, but that's not where the error occurs 20:27
20:28 n0tjack left, tokuhiro_ joined
jdv79 is there really a dd and ddt? 20:28
moritz m: dd [1, 2]
camelia rakudo-moar 0c79be: OUTPUT«[1, 2]␤»
moritz m: ddt [1, 2]
camelia rakudo-moar 0c79be: OUTPUT«5===SORRY!5=== Error while compiling /tmp/TNqaPHGd66␤Undeclared routine:␤ ddt used at line 1. Did you mean 'dd'?␤␤»
moritz seems there's a dd, no ddt
jdv79 ok
[Coke]: what are you talking about then? 20:29
brrt larry wall yapc::eu keynote: youtu.be/RvCkvXvqi3U 20:31
skids moritz: indeed, fixing the single-arg issue there lets it run much further.
psch github.com/rakudo/rakudo/pull/529 has ddt
20:32 tokuhiro_ left, brrt left 20:33 xfix left
dalek c/htmlify-glr: d1363d6 | moritz++ | htmlify.p6:
Fix fallout from the single-arg rule
 20:36
skids moritz: that push made a new branch... intentionally? 20:44
20:44 dha joined
moritz skids: yes; I'm afraid it (or the next commit) will break the build under pre-glr 20:44
skids OK, just checking
moritz m: my @a = Pod::FormattingCode.new(type => "X", meta => [["declarator", "sub"],], config => {}, contents => ["Subroutines"]).meta[0]; say @d.perl 20:45
camelia rakudo-moar 0c79be: OUTPUT«5===SORRY!5=== Error while compiling /tmp/95e0BmUaYs␤Variable '@d' is not declared␤at /tmp/95e0BmUaYs:1␤------> 3ntents => ["Subroutines"]).meta[0]; say 7⏏5@d.perl␤»
moritz m: my @a = Pod::FormattingCode.new(type => "X", meta => [["declarator", "sub"],], config => {}, contents => ["Subroutines"]).meta[0]; say @a.perl
camelia rakudo-moar 0c79be: OUTPUT«[["declarator", "sub"],]␤»
moritz why? 20:46
m: say Pod::FormattingCode.new(type => "X", meta => [["declarator", "sub"],], config => {}, contents => ["Subroutines"]).meta[0].perl
camelia rakudo-moar 0c79be: OUTPUT«["declarator", "sub"]␤»
moritz why doesn't the single-arg rule flatten that out when assigning it to an array? 20:47
moritz notices he still doesn't have a reliable mental of post-glr (non-)flattening
skids Is .meta a scalar maybe? 20:48
moritz if it makes a difference, .perl should inform me about it, no? 20:49
skids m: my $a = [1,2]; my @b = $a; $a.perl.say; @b.perl.say; 20:50
camelia rakudo-moar 0c79be: OUTPUT«[1, 2]␤[[1, 2],]␤»
skids m: my $a := [1,2]; my @b = $a; $a.perl.say; @b.perl.say;
camelia rakudo-moar 0c79be: OUTPUT«[1, 2]␤[1, 2]␤»
skids The first $a.perl could perhaps be $[] 20:51
The breaking of the transparency of Scalar was a necessary consequence of GLRR, I don;t now how far we want to let that bleed though. 20:52
dalek c/htmlify-glr: 6770eb8 | moritz++ | htmlify.p6:
Another GLR fix
 20:53
moritz enough hacking for tonight 20:54
good * everybody&
FROGGS gnight moritz 20:55
skids And for that matter, even as a scalar it should have let the .[] through to what it contained.
o/
20:55 n0tjack joined
skids m: class A { has @.a is rw }; my $f = A.new(a => [[1,2],]); $f.a[0].say; $f.a[0].WHAT.say; $f.a[0].VAR.WHAT.say; my @b = $f.a[0]; @b.perl.say; 20:59
camelia rakudo-moar 0c79be: OUTPUT«[1 2]␤(Array)␤(Scalar)␤[[1, 2],]␤»
21:00 n0tjack left, n0tjack joined 21:01 rindolf left, skids left 21:05 n0tjack left 21:06 n0tjack joined 21:07 cognominal left 21:11 n0tjack left 21:15 kinslayer left, n0tjack joined 21:19 n0tjack left 21:20 dha left 21:21 AlexDaniel joined
dalek kudo-star-daily: 9715a26 | coke++ | log/ (9 files):
today (automated commit)
 21:26
21:27 dha joined 21:28 n0tjack joined 21:33 n0tjack left
dha Would it be wrong to say that one can think of state variables in this way: They are lexically scoped in terms of access, but the contents of the variable are maintained between calls as they would be if the variable had been declared globally? 21:38
I understand that saying that initialization occurs only once is probably the more accurate way to put it, but I find that could be confusing, depending on what one thinks "initialization" is. 21:40
jercos I would say that's correct in so far as my understanding of their relation to C static variables. 21:43
It may not be a technically accuracte description of their function, but I don't see a way in which that understanding would cause unintended behavior. 21:44
japhb dha: You need to be clear about the interaction of state variables and closure clones. Talking about global declarations gives people a different sense of how sharing works (or doesn't). 21:59
21:59 yqt left
[Coke] ddt - there was a pull request 22:00
22:00 n0tjack joined 22:03 Lou_ joined 22:04 Lou_ left 22:05 n0tjack left 22:07 muraiki left
dha Ok, what do I need to know about closure clones? 22:09
And, just to be clear, that was a "you can think of it in these terms, but it's not necessarily accurate" 22:10
22:17 rurban left 22:23 pmurias left, tokuhiro_ joined 22:24 justcheckingin joined 22:25 justcheckingin left 22:27 tokuhiro_ left 22:28 atweiden joined 22:33 n0tjack joined 22:35 leont left 22:37 n0tjack left 22:38 hernanGOA left 22:49 dha left 22:53 sufrostico joined 23:04 krakan joined
Timbus m: if my $things := (1,2,3,4).grep(*>2) { say $things>>.Str; } 23:05
camelia rakudo-moar 0c79be: OUTPUT«This Seq has already been iterated, and its values consumed␤ in block <unit> at /tmp/X1g5zFXR6z:1␤␤»
Timbus m: my $things := (1,2,3,4).grep(*>2); say $things>>.Str;
camelia rakudo-moar 0c79be: OUTPUT«(3 4)␤»
23:05 stux|RC-only joined, n0tjack joined
Timbus whyfor :/ 23:06
psch m: with (1,2,3,4).grep( * > 2 ) -> $things { $things.say }
camelia rakudo-moar 0c79be: OUTPUT«(3 4)␤»
psch m: if (1,2,3,4).grep( * > 2 ) -> $things { $things.say } # although if works here as well
camelia rakudo-moar 0c79be: OUTPUT«(3 4)␤»
Timbus where the heck did 'with' come from 23:07
thats cool
psch m: if (^3).grep( * < 1 ) -> $things { $things.say } # doesn't here
camelia rakudo-moar 0c79be: OUTPUT«(0)␤»
psch oh
list with one elem
nvm
m: if 0 -> $f { say "huh" }; with 0 -> $f { say "hah!" }
camelia rakudo-moar 0c79be: OUTPUT«hah!␤»
psch that's the difference
23:08 atweiden left
Timbus duuude, thats way better than makling all our stuff return 0 but True 23:08
psch Timbus: i think that's probably pretty close to the reason why it got implemented :)
Timbus that's cool. but that's still odd re: list reiteration 23:10
23:10 n0tjack left 23:11 isacloud joined 23:12 lizmat joined
psch Timbus: yeah, the error is LTA 23:13
Timbus: but afaiu this kind of construct (i.e. "if <assignment or binding> { }") is somewhere in the area of DIHWIDT 23:14
23:14 skids joined
lizmat computer, messages ? 23:16
guess not
23:16 mprelude left
lizmat waves from Lyon 23:16
psch hi lizmat o/
lizmat psch o/
Timbus well, it's fairly common to assign in an if, though.. 23:17
at least, the perl5 code at my work is littered with it. some of which I put there. 23:18
psch m: if my $x = 5 { }; say $x
camelia rakudo-moar 0c79be: OUTPUT«5␤»
Timbus aaaa 23:19
aaah no
psch Timbus: P5 special-cases it so that that doesn't happen :)
Timbus so thats a thing, huh
psch ...maybe not really special-casing, i guess
lizmat dd my $i = 42
psch P6 scoping is a bit different, i'm confident in that statement
lizmat m: dd my $i = 42 23:20
camelia rakudo-moar 0c79be: OUTPUT«$i = 42␤»
Timbus well, to be fair i use -> instead of assignment, but its the same idea
23:20 sufrostico left
psch m: if 5 -> $x { }; say $x 23:20
camelia rakudo-moar 0c79be: OUTPUT«5===SORRY!5=== Error while compiling /tmp/73l27rh5oo␤Variable '$x' is not declared␤at /tmp/73l27rh5oo:1␤------> 3if 5 -> $x { }; say 7⏏5$x␤»
Timbus yeah. so that's ok 23:21
23:21 vendethiel left
psch in contrast: 23:22
perl -E'if( my $x = 5 ) { }; say $x, "or else"'
or else
Timbus yeah, the scoping is different, that's fine. my initial example only used := instead of -> to make it more clear the 'if' block is doing something special. the arrow makes it less clear. 23:24
psch m: say (^5).grep(*>2).WHAT 23:25
camelia rakudo-moar 0c79be: OUTPUT«(Seq)␤»
psch m: my $x := (^5).grep(*>2); say $x>>.Str; say $x>>.Str
camelia rakudo-moar 0c79be: OUTPUT«(3 4)␤This Seq has already been iterated, and its values consumed␤ in block <unit> at /tmp/y9PZ6pcQrm:1␤␤»
psch Timbus: the if block isn't doing anything special 23:26
Timbus oh..
psch Timbus: Seq is an ephemeral class, after evaluation it returns something Associative (i think..?) and breaks
Timbus: but you bound the Seq to a variable 23:27
and it's already broken
which is why i'm saying the error is somewhat LTA :) 23:28
23:28 telex left
Timbus what makes it broken straight away? oh, the evaluation of its truthiness? 23:28
psch yeah 23:29
Timbus mmh. either way, will use 'with' and probably switch to @'s, cheers
psch Timbus: the sigil isn't the problem, the binding is
...i thought, but that's apparently wrong 23:30
it is the sigil
23:30 telex joined
Timbus :) 23:30
psch those post-glr semantics...
i really should get around to reading what changed :/ 23:31
23:32 n0tjack joined
Timbus feels a little magical atm.. i guess it must just be sufficiently advanced 23:32
psch i *think* the rule in general is "don't secretly flatten"
but that rule has consequences that seem unintuitive... 23:33
m: say [[1,]]; say [[1],]
camelia rakudo-moar 0c79be: OUTPUT«[1]␤[[1]]␤»
dalek kudo/nom: 49e9745 | lizmat++ | src/core/Any.pm:
Usurp PR #529 into dd, itz++

No point keeping a separate dd and ddt in my view.
lizmat in about 7 hours there's the patch -p3 hackathon: patch.pm 23:37
better get some sleep&
dalek kudo/cu_load: b410729 | TimToady++ | src/ (13 files):
early hackage to distinguish .cache from .list
 23:39
rakudo/cu_load: 834c805 | TimToady++ | / (8 files):
rakudo/cu_load: Merge branch 'nom' into list-v-cache
23:39 dalek left, dalek joined, ChanServ sets mode: +v dalek 23:43 raiph joined 23:48 n0tjack left
psch Timbus: the alternative to assigning a Seq to an @-sigilled variable is calling .cache on the Seq, fyi 23:52
m: my $x = (^5).grep(*>2); say $x>>.Str; say $x.cache>>.Str;
camelia rakudo-moar 49e974: OUTPUT«(3 4)␤This Seq has already been iterated, and its values consumed␤ in block <unit> at /tmp/D1imEvAmwu:1␤␤»
psch m: my $x = (^5).grep(*>2); say $x.cache>>.Str; say $x.cache>>.Str;
camelia rakudo-moar 49e974: OUTPUT«(3 4)␤(3 4)␤»
Timbus i see. kinda 23:53
psch ...i'm still not 100% sure on the meaning of some of those things, though :)
m: my $x = (^5).grep(*>2).cache; say $x>>.Str; say $x>>.Str; 23:54
camelia rakudo-moar 49e974: OUTPUT«(3 4)␤(3 4)␤»
23:56 n0tjack joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!