»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg camelia perl6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by sorear on 25 June 2013.
00:01 dwarring left 00:14 atroxaper joined 00:19 atroxaper left 00:24 btyler joined, djanatyn left
dalek rl6-roast-data: 308115a | coke++ | / (6 files):
today (automated commit)
00:27
00:29 btyler left 00:30 btyler joined
[Coke] (mojo6) I started working on that, with the least bit of brain I could, which involved mojo-util. I finally have some cycles available, but probably need someone to work with to try to port mojo more aggressively. (otherwise, I tend to work more on shiny things or cleaning up tests, things like that.) 00:34
00:36 djanatyn joined 00:39 btyler left 00:40 SOMEWHEREOVERTHE joined, SOMEWHEREOVERTHE left
[Coke] has a very close relative whose birthday is Nov 11 00:41
00:49 Tene_ is now known as Tene 01:03 dayangkun joined 01:08 hoverboard joined 01:12 psch left 01:28 rurban2 left 01:47 hypolin joined, rurban1 joined 01:49 Sqirrel left 01:50 Sqirrel joined, FROGGS left 01:51 klapperl_ left 01:53 klapperl joined, hypolin left 01:54 Alula_ left 01:58 chenryn joined 02:02 raiph joined 02:03 pdcawley left, FROGGS joined 02:04 Alula_ joined 02:06 pdcawley joined 02:12 BenGoldberg joined 02:17 noganex_ joined 02:20 noganex left 02:26 dragon_d joined 02:29 lelf`` joined, lelf` left 02:37 rurban1 left
ssutch how does one address circular module loading in p6? 02:42
in python normally we just move the import statement local 02:43
ren1us ssutch: irclog.perlgeek.de/perl6/2014-06-12#i_8860068 (the answer I got a while back) 02:46
02:52 skids joined, rurban1 joined
ssutch ren1us: thanks! 02:57
02:59 dragon_d left 03:02 rindolf joined
ssutch ren1us: did you get around it by using need My::Class; 03:03
03:04 dragon_d joined
Mouq .tell jnthn Something is broken on the latest Moar such that htmlify.p6 no longer works, dieing with "Error while compiling op call: Local '__lowered_param_0' already declared" 03:06
yoleaux Mouq: I'll pass your message to jnthn.
Mouq .tell jnthn htmlify.p6 in perl6/doc, that is 03:07
yoleaux Mouq: I'll pass your message to jnthn.
ssutch =/ maybe i just cant have type constraints unless the two classes are in the same file 03:15
03:17 atroxaper joined
Mouq ssutch: You can't just declare a stub? 03:17
my class XYZ { } ? 03:19
.ask tadzik If there's going to be an effort to port Mojo to Perl 6, would it be better to rename your Template::Mojo as Mojo6::Template or similar? 03:21
yoleaux Mouq: I'll pass your message to tadzik.
Mouq and tadzik++
jnthn++ for that matter
coke++
e'rybody++
TimToady Mouq++ :) 03:24
03:27 atroxaper left 03:28 atroxaper joined, xenoterracide__ left
Mouq :) 03:30
03:32 atroxaper left 03:35 atroxaper joined 03:36 SevenWolf joined, lelf`` left
Mouq is equally entertained and perplexed by the Futurama quotes in the Mojolicious source 03:46
03:46 [Sno] left
ssutch hrm IO::Socket::INET.read(12) runs the cpu to 100% and blocks the OS until ubuntu kills it 03:48
on the moar backend
wireshark can see the reply from the server 03:49
TimToady the IO system still is a bit primitive compared to the rest of p6, I fear 03:51
ssutch heh 03:52
03:52 araujo left
TimToady 'course, if it was C, I'd just say "that's a library issue, not a language issue" :) 03:52
_sri Mouq: perplexed? 03:53
ssutch maybe ill come back to this tomorrow and give it a go on another backend, or try to fix the IO issue 03:57
for now though i'm bushed, g'night p6!
Mouq _sri: The first I saw was mojolicio.us/perldoc/Mojo/IOLoop.txt and I thought maybe there was a death ray hidden in the code xD
'night ssutch! 03:58
_sri but there is!
03:58 BenGoldberg left
Mouq _sri: Mercury beam? or neutron cannon? 03:59
04:00 ribasushi left 04:03 kaare_ joined 04:06 anaeem1 joined 04:07 ribasushi joined, ssutch left 04:08 Rotwang left 04:11 raiph left
_sri Mouq: mostly magic... it opens a portal to callback hell and summons a greater evil 04:15
you know... the usual async i/o stuff
04:16 chenryn left
Mouq _sri: Ohhh, yes. That is dark magic indeed, my friend 04:27
04:35 virtualsue joined 04:44 chenryn joined 04:52 kaare_ left 05:04 isBEKaml_mobile joined 05:06 gfldex joined 05:07 [Sno] joined 05:14 rurban1 left 05:21 kaare_ joined 05:24 potatogim left 05:30 dragon_d left 05:46 dayangkun left 06:02 virtualsue left 06:06 denis_boyun joined 06:15 chenryn left 06:17 denis_boyun left 06:18 chenryn joined 06:19 denis_boyun_ joined 06:23 FROGGS left 06:34 denis_boyun_ left 06:37 virtualsue joined 06:39 denis_boyun joined
tadzik Mouq: probably, yes :) 06:41
yoleaux 03:21Z <Mouq> tadzik: If there's going to be an effort to port Mojo to Perl 6, would it be better to rename your Template::Mojo as Mojo6::Template or similar?
06:42 SamuraiJack_ joined
tadzik well, actually, I think Template::Mojo is a perfect name for that 06:42
it makes sense for templating engines to be in a Template namespace, I think. Mojo::Template could be that-part-of-mojo-that-deals-with-template-engines 06:43
sergot morning o/ 06:44
06:46 Sqirrel left 06:47 darutoko joined 06:57 FROGGS joined 06:59 denis_boyun left 07:02 dayangkun joined 07:09 dalek left 07:10 dalek joined, ChanServ sets mode: +v dalek 07:14 jack_rabbit left, dayangkun left 07:28 dayangkun joined 07:38 virtualsue left 07:40 igorsutton joined 07:46 pecastro left 07:51 hoverboard left 07:56 sqirrel joined
atroxaper Hello, #perl6 ! 08:06
FROGGS hi atroxaper
atroxaper FROGGS: o/
08:06 Adriaaan joined
atroxaper How do you think what should we get if translate pod 'L<C<boom>>' to HTML? Is it <a href="boom"><code>boom</code></a>? Or the same but with href="#" ? 08:08
FROGGS I have no idea... I'm not much into Pod :/ 08:09
08:10 denis_boyun_ joined
atroxaper I almost finish my system for writing Pod To Something converters. Now I'm trying to rewrite Pod::To::HTML to my system. 08:10
FROGGS I think you have to as others, sorry 08:11
atroxaper The next plained step is Pod::To::Markdown. Because github cant parse Pod format :)
FROGGS Mouq++ and lue++ might know 08:12
atroxaper Maybe somebody will see my question :)
FROGGS: thank you
08:19 sunnavy left, sunnavy joined 08:23 fhelmberger joined
atroxaper Current Pod::To::HTML do so '<a href="<code>boom</code>"><code>boom</code></a>'. And I think it's wrong... 08:24
FROGGS yeah, I can agree on that :o) 08:31
I guess href="#" makes most sense, because you do not want to have huge anchors or even newlines in them 08:32
08:37 SigmaWP joined
atroxaper FROGGS: maybe. But as I understand from S26 <a href="boom"><code>boom</code></a> will be the best. I mean we should get bare content of L<> and put into href. 08:37
FROGGS yeah, true 08:38
08:39 SigmaWP left
dalek kudo/nom: e21b8fc | (Tobias Leich)++ | lib/Test.pm:
strip # from description in todo tests
08:48
08:49 dmol joined 09:00 denis_boyun_ left
masak good antenoon, #perl6 09:02
09:03 ivanshmakov left
atroxaper masak: hello 09:05
09:07 SevenWolf left, spider-mario joined
masak hello, and welcome to another interesting day in Perl 6's development! 09:07
09:09 xinming left
atroxaper masak: What do you mean? :) 09:10
09:11 virtualsue joined
atroxaper Btw, is there list's shift by from the last element instead of first. Or list's push to the first element? I need to write simple stack. 09:11
09:11 xinming joined
masak I mean just what I say. which part of the sentence did you find confusing? 09:12
atroxaper: .pop -- .unshift
atroxaper: prefer the .push/.pop pair to the .unshift/.shift pair for performance.
vendethiel rewrote his 100-lines C++ queue class in 20 lines of P6
atroxaper masak: the sentence isn't usual)
masak granted. 09:13
but it's not very complicated, either.
atroxaper masak: thanks!
jnthn .shift is pretty efficient these days since it just moves a start pointer. 09:16
yoleaux 03:06Z <Mouq> jnthn: Something is broken on the latest Moar such that htmlify.p6 no longer works, dieing with "Error while compiling op call: Local '__lowered_param_0' already declared"
03:07Z <Mouq> jnthn: htmlify.p6 in perl6/doc, that is
jnthn Hm, that's nothing to do with Moar, that's a Rakudo optimizer issue...
Looks like, anyways... 09:17
masak jnthn: is that to the point where .shift isn't significantly slower than .push ?
jnthn If you s/push/pop/ then yes
masak right, meant that.
jnthn unshift can be a little more costly
masak ah, thought so.
so it still makes sense to prefer .push/.pop
jnthn but if the thing is in a steady state of shift/unshift rather than continuous growth it'll come out about the same 09:18
For documentation alone, if you mean "a stack" then push/pop make sense. :)
atroxaper jnthn: yes. I want a stack. 09:19
09:21 fhelmberger left 09:28 ivanshmakov joined 09:33 pecastro joined
vendethiel .ask raiph finally coming to your gist (after an awesome week without touching a comp), I have a question on "# you don't explain why sigils are needed in general. maybe remove?" <- why don't I need to ? 09:37
yoleaux vendethiel: I'll pass your message to raiph.
09:37 virtualsue left
vendethiel also "capture" a function seems correct to me 09:39
09:44 virtualsue joined
vendethiel well, integrated everything except 2 or 3 minor thingies I rewrote differently. raiph++ 09:44
09:46 denis_boyun joined 09:51 lelf`` joined 09:56 lelf``` joined, lelf`` left 09:57 teodozjan joined 10:05 virtualsue left 10:06 virtualsue joined
vendethiel oh, looks like the mailer delivered me something ... 10:25
10:27 dayangkun left
vendethiel lizmat hey-oh 10:32
10:54 sqirrel left
dalek nqp: 8683e9c | sergot++ | src/vm/moar/QAST/QASTOperationsMAST.nqp: 10:57
nqp: nativecast operator added
nqp:
nqp: nativecast op has been added, it makes us able to cast things
nqp: conditionally. FROGGS++
10:57 dalek left 10:58 dalek joined, ChanServ sets mode: +v dalek, chenryn left
dalek kudo/nom: 25c5114 | (Tobias Leich)++ | tools/build/NQP_REVISION:
bump nqp/moar rev for nativecallcast op
10:58
10:58 vend joined
vend \o 10:59
FROGGS hi vend
vendethiel hi
sergot o/
timotimo spam mails are getting harder and harder to spot. this time i got a fake amazon order confirmation with the correct address. it said "Die Bestellung geht an: Deutschland"
11:00 Akagi201 joined
dalek volaj: d249648 | (Tobias Leich)++ | / (3 files):
Reenable usage of nqp::nativecast, and its tests

This reverts commit c0f68bca0a3d34a7d59f872fce0e7f34892cdff2.
11:00
volaj: 20424bd | (Tobias Leich)++ | lib/NativeCall.pm6:
bind result of nqp::nativecallcast so there is no extra container

Otherwise calling nqp::isint($result) will blow up.
volaj: 019d26e | (Tobias Leich)++ | lib/NativeCall.pm6:
box native string returned by nativecallcast when we get one
volaj: ec3f509 | (Tobias Leich)++ | lib/NativeCall.pm6:
simplify sub nativecast

We do not need to box the return value anymore since we provide the proper box target now. A call looks now like: nativecast(int32, Int, $opaque-pointer)
atroxaper How can I define a module more correctly? module Foo {} or class Foo; And what is the difference? 11:05
I have a problem... If I define "module Foo { sub bar(){} }" then i can't call Foo.bar(). But if I define "class Foo; method bar(){}" I can call Foo.bar() 11:07
I don't want to use is export.
FROGGS Foo.bar() is always a method call of an instance of a class 11:08
Foo::bar() would be a sub call in a module/class Foo, the sub must be our scoped then
m: module Foo { our sub bar() { 42 } }; say Foo::bar 11:09
camelia rakudo-moar e21b8f: OUTPUT«42␤»
FROGGS m: module Foo { our sub bar() { 42 } }; say bar # this will blow up
camelia rakudo-moar e21b8f: OUTPUT«===SORRY!=== Error while compiling /tmp/K7eWgnlvf9␤Undeclared routine:␤ bar used at line 1. Did you mean 'bag'?␤␤»
atroxaper FROGGS: Oh. Yes! I forgot about 'our' and can't imaging whay Foo::bar doesn't work)
FROGGS hope that helps :o) 11:10
11:10 virtualsue left
atroxaper FROGGS: as I understand first example from perl6advent.wordpress.com/2009/12/1...exporting/ is wrong without 'our' 11:10
FROGGS m: module Foo { sub bar() { 42 } }; say Foo::bar 11:11
camelia rakudo-moar e21b8f: OUTPUT«Could not find symbol '&bar'␤ in method <anon> at src/gen/m-CORE.setting:12877␤ in any find_method_fallback at src/gen/m-Metamodel.nqp:2604␤ in any find_method at src/gen/m-Metamodel.nqp:934␤ in block at /tmp/MNfQl2mei5:1␤␤»
11:11 kaleem joined
FROGGS yes, subs are my scoped by default 11:11
(in contrast to any kind of package, like module/class/role/...) 11:12
atroxaper FROGGS: 'our' helped me. And you too :)
FROGGS :o)
11:15 atroxaper left, vend left, atroxaper joined 11:20 atroxaper left 11:31 brrt joined 11:38 fhelmberger joined 11:53 anaeem1 left 12:04 rindolf left, shlomif joined, lelf``` left 12:05 shlomif is now known as rindolf 12:07 lelf joined 12:08 cognominal left, cognominal joined 12:11 brrt left
vendethiel then again, rakudobrew totally failed me. Guess I'll just build by source 12:14
12:15 Adriaaaan joined
vendethiel "make: write error" 12:15
12:15 lelf left
FROGGS that sounds MacOSXish 12:16
vendethiel yeah
and why has vend gone to sleep ? uh.
FROGGS that is a bug in make
vendethiel I can tell it's not sleeping ... 12:17
12:17 Ven joined 12:18 Adriaaan left, dayangkun joined
lizmat vendethiel: just ignore the "make: write error" on OS X 12:20
it's a long standing issue with make on OS X that seems to be tickled by rakudo
Ven I can't. There's no perl6 executable ..
lizmat run it again
empirically, the segfault in the make install only happens if it compiled something in nqp
if you run it again, all in nqp has already been compiled, so no problem then 12:21
vendethiel uh-oh. That indeed looks like a nasty bug ... 12:22
12:22 rurban1 joined, jnap joined 12:28 LordV joined, clkaoud left, isacloud__ left, ggherdov left 12:29 _sri left, Bucciarati left 12:30 Woodi_ joined, clkaoud joined, ivanshmakov left, isacloud___ joined, jtpalmer_ joined, ugexe_ joined, haroldwu_ joined, nwc10_ joined, go|dfish left, Woodi left, nwc10 left, Alula_ left, jtpalmer left 12:31 integral joined, integral left, integral joined, clkaoud left, clkaoud joined, isacloud___ left, isacloud___ joined, ggherdov joined, _sri joined 12:32 go|dfish joined, nwc10_ is now known as nwc10, isBEKaml_mobile left 12:34 brrt joined, Ven left 12:35 Bucciarati joined 12:36 Vlavv left, atroxaper joined 12:38 atroxaper left, atroxaper joined 12:40 Ven joined
dalek ast/S26-WHY: a2ab77a | (Rob Hoelz)++ | S26-TODO:
More things TODO for S26
12:47
12:49 Vlavv joined 12:53 virtualsue joined 12:54 Akagi201 left, Akagi201 joined 12:55 Akagi201 left 12:56 Akagi201 joined
[Coke] FROGGS: rather than strip the #'s, please escape them. 12:58
FROGGS [Coke]: how? 12:59
12:59 ivanshmakov joined
[Coke] FROGGS: excellent question. (looks at testanything.org/tap-specification.html) ... I withdraw my suggestion. :) 13:01
FROGGS++
(I think that \# works, but not worth it.)
13:07 dmol left 13:17 woolfy left 13:18 sqirrel joined 13:20 woolfy joined 13:22 raiph joined, denis_boyun left 13:23 rurban1 left, denis_boyun_ joined 13:27 guru joined, guru is now known as ajr_ 13:28 lizmat left 13:32 rurban1 joined, lizmat joined 13:33 psch joined, btyler joined
psch hi #perl6 o/ 13:34
moritz \o psch 13:35
colomon_ \o
raiph o/ 13:40
yoleaux 09:37Z <vendethiel> raiph: finally coming to your gist (after an awesome week without touching a comp), I have a question on "# you don't explain why sigils are needed in general. maybe remove?" <- why don't I need to ?
13:42 zakharyas joined
zengargoyle sergot: moritz: thanks for the bit. 13:44
raiph .tell vendethiel Has scriptogr.am/vendethiel/post/and-pe...came-forth changed? Given other langs don't need sigils, your answer to "Why is it needed?" seems unsatisfying. Maybe "When is it needed?". My issue with "capture" was that it already has a different technical meaning in P6 (actually at least 2 entirely distinct technical meanings)
. But this last is a very minor nit.
yoleaux raiph: I'll pass your message to vendethiel.
vendethiel I'm here
yoleaux 13:44Z <raiph> vendethiel: Has scriptogr.am/vendethiel/post/and-pe...came-forth changed? Given other langs don't need sigils, your answer to "Why is it needed?" seems unsatisfying. Maybe "When is it needed?". My issue with "capture" was that it already has a different technical meaning in P6 (actually at least 2 entirely distinct technical meanings)
vendethiel thanks yoleaux. 13:45
raiph: maybe "which sigil to use" ? 13:46
13:47 SamuraiJack_ left
vendethiel and wrt capture, I see, but I don't have a better term for it atm 13:47
jnthn You don't *need* all kinds of things in a langauge, but sometimes introducing them gives nice properties.
raiph vendethiel: yeah, that sort of thing (sigil)
13:48 haroldwu_ is now known as haroldwu 13:49 haroldwu left, haroldwu joined 13:50 troydm left 13:54 kaare_ left 13:57 atroxaper left 14:01 atroxaper joined, troydm joined 14:06 skids left, atroxaper left 14:07 atroxaper joined, bluescreen10 joined
zengargoyle vendethiel: in Passing parameters ... Also, in Perl 6, the last expression (or last line) of a function is automatically returned. ... might be better mentioned after a non-say function. 14:07
vendethiel not for now
14:08 sqirrel left
zengargoyle i.e. it seems to imply that 'Hello, Stephen' is the return value. 14:08
and would make more sense with something like sub add($n) { $n + 1 } 14:09
becase it's the REPL that's not printing the return value since it saw output. 14:10
14:11 atroxaper left
zengargoyle m: my $rv; sub hello(Any $name) { $rv = say "Hello, $name" }; hello("Steve"); say $rv; 14:12
camelia rakudo-moar 25c511: OUTPUT«Hello, Steve␤True␤»
psch i think that's better solved with mentioning that the REPL doesn't print the return value when the last statements is a say 14:15
in "Getting started" there's mention of how the REPL prints the retval or nothing if there wasn't one
maybe add that the return value of say gets swallowed by the REPL because it's assumed that it doesn't matter 14:16
or is it output to STDOUT in general? something to that effect
vendethiel psch: it's mentioned already 14:17
psch yeah, i just saw :) 14:18
14:18 treehug88 joined
zengargoyle i just think that if you introduce a rule, it shoule be with an example that shows that rule, or at least not with an example that demonstrates an exception to that rule (unless you point out the exception) 14:23
14:24 zakharyas left
zengargoyle then again, the REPL seems to be full of weirdness with nested scopes and such. 14:26
lizmat FROGGS: it seems like your last change to lib/Test.pm makes some "skip" tests also get the "todo" text
specifically in t/spec/S04-declarations/will.t, test #3: 14:27
ok 3 - SKIP declared variable not visible in block yet# TODO will init NYI
FROGGS ummm
lizmat this will give a TODO passed 14:28
when in fact we're skipping the test
14:29 molaf joined
lizmat somehow $todo_reason is not reset 14:29
FROGGS lizmat: that means that this todo lives too long: github.com/perl6/roast/blob/master...will.t#L25 14:30
but yeah.. why
I can't take care of it right now, I'm bisecting rakudo :/
lizmat perhaps it selects the wrong skip mmd
ok
will look at it then
FROGGS lizmat++
14:36 chenryn joined 14:38 Akagi201_ joined 14:39 Akagi201 left 14:40 raiph left, raiph joined
lizmat hmmm... it seems to generate a todo() for *every* line in the block that's being todo'd, rather than just the line with "is" 14:45
dalek ast: 5826376 | (Elizabeth Mattijsen)++ | S04-declarations/will.t:
Work around 4x todo generating strangeness
14:48
lizmat the only thing I can think of, is that it somehow gets confused by the BEGIN and INIT blocks 14:49
14:49 isBEKaml_mobile joined
lizmat this seems to fix the test error reporting problem, but does not fix the underlying fudging issue 14:49
cycling&
15:07 atroxaper joined
hoelzro morning #perl6 15:10
FROGGS hi hoelzro
15:11 cxreg2 is now known as cxreg, atroxaper left, atroxaper joined 15:13 igorsutton left, kaleem left
hoelzro ahoy FROGGS 15:14
15:16 Akagi201 joined 15:18 Akagi201_ left 15:21 ssutch joined, dayangkun left 15:26 hoverboard joined 15:27 ajr_ left 15:28 denis_boyun_ left, ajr joined 15:29 ajr is now known as Guest67321, Guest67321 is now known as ajr_ 15:30 skids joined 15:31 isBEKaml_mobile left 15:34 [Sno] left 15:35 ugexe_ is now known as ugexe, FROGGS left 15:36 ssutch left
moritz rakudo's todo() function should catch # in the TODO message 15:38
it seems to be a common-ish error
15:38 Ven left
masak +1 15:41
15:41 fhelmberger left, felipe joined 15:42 FROGGS joined
gtodd easy to use "time" to make lazy comparisons of the speed of a script/oneliner with perl6-m versus perl6-j versus perl6-p how can I (not really scientifically) compare/test the speed of p6 on the jvm factoring out the jvm load? maybe start the repl perl6-j and then eval the script ... are there newbeasy-to-use timing or benchmarking tools somewhere in the distrubtion (./t ./tools)? 15:42
granted in the flying car future jvm startup time will not be a factor .... 15:44
FROGGS jnthn: there are 44 cases when compiling the setting that where blocks were inlined... I try to figure out now which not-inlined blocks cause the REDUCE fail for perl6-p 15:45
15:45 kaleem joined
moritz gtodd: there's a perl6-bench repo where there's some logic for factoring out startup time 15:45
FROGGS gtodd: we already have an eval server
moritz no idea how well that functionaility is exposedto the outside
FROGGS our spectests use it... you'd start the server and send stuff to it that gets evaled 15:46
15:46 kaare_ joined
moritz another option is to run END { say now - BEGIN now } 15:46
it's not quite correct (it doesn't count compilation time prior to that time), but it's a good start
FROGGS I think in perl6-bench it runs an empty program ten times to measure the startup time, and this gets subtracted later 15:47
15:47 Ven joined
gtodd moritz: oh good ok 15:48
vendethiel moritz: CHECK instead of BEGIN ?
15:48 Akagi201 left
FROGGS do we have CHECK? 15:48
moritz vendethiel: then it won't count compile time at all
vendethiel oh, you want to count it
gtodd FROGGS: yeah I thought of just timing the naked startup and subtracting :) 15:49
15:49 Akagi201 joined
gtodd FROGGS: I didn't really look into setting up my own eval server ... if it is not too complicated that is what I am going to do :) 15:50
FROGGS gtodd: it sits just there in the root dir of rakudo 15:51
grep for the script's filename in t/, and you'll know how to invoke it
gtodd likes the idea of perl6 running all the time ...
15:51 Akagi201_ joined
FROGGS well, that kinda happens on my box anyway :o) 15:52
15:52 brrt left
FROGGS jnthn: these are the 44 poisened blocks btw: gist.github.com/FROGGS/15dc422d8cd27849b9a0 15:54
15:55 Akagi201 left 15:59 carlin joined
[Coke] Ugh. looks like I need to add a ulimited perl6 runner to the daily star runs. 16:01
(looks like modules/file-find might be timing out.) 16:02
er, hanging.
16:03 rurban1 left 16:04 lyda left
colomon_ [Coke]: I ran into a bit of that last week. 16:04
[Coke] yah, I wasn't really looking at things when on vaca last two weeks, just pushing the results I did get. 16:06
Easy enough to steal from the roast-data project which is already doing that. 16:07
anyone mind if I rip out the Order::Decrease tests from roast? 16:09
gtodd ok sorry to blather on .... just so I 'm getting this straight ... I just did time perl6-j -e 'END { say now - BEGIN now }' which printed out 0.2249989 and "20.04s user 1.34s system 174% cpu 12.266 total" so this tells me it seems to take about 12 seconds to start the jvm (a mechanical clock on my desk roughly confirms this) the 20s user being something to do with threads/multicore or in any case not for my brain to de
[Coke] (given that rakudo doesn't have them, and anyone else is dead, dying, or hasn't started yet?)
gtodd: "for my brain to de" 16:10
16:10 btyler_ joined 16:11 isBEKaml joined
FROGGS [Coke]: +1 16:11
gtodd eeply consider ... trying that .... :)
16:12 rurban1 joined
FROGGS gtodd: you see that it uses 174% cpu 16:12
it has a concurrent GC
colomon_ [Coke]: wait, what's that about Order::Decrease?
gtodd yes :-)
16:12 btyler left
gtodd FROGGS: what you just said == "deep" for me :) 16:13
FROGGS colomon_: it will go more or less
16:13 slavik left
colomon_ Should be Order::Less now? 16:14
m: say "a" cmp "b"
camelia rakudo-moar 25c511: OUTPUT«Less␤»
16:14 dmol joined
gtodd if I remove that time from perl6-j it comes second to perl6-m moar and ahead of perl6-p parrot which I gather is expected behavior these days ... 16:15
16:17 _Slade joined
colomon_ [Coke], FROGGS: should the Order::Decrease tests be removed or renovated? 16:17
gtodd so the goal is parity re "correctness" vis-à-vis the SPEC but for now ..... moar = fastest parrot = most "correct" jvm == "safest" or something like that?
FROGGS colomon_: as I understand it there are tests for less/more, and the others are kept as regression tests, which does not make much sense 16:18
gtodd: jvm is not really safe
gtodd: parrot was reliable until like a week ago
(we broke it)
and yes, moar is the fastest 16:19
colomon_ FROGGS, [Coke]: Huh. Well, I guess I'm against deleting things that we don't understand why they are still there at all. Maybe give a day or two for someone to protest to their proposed deletion?
16:20 Akagi201_ left
gtodd FROGGS: ok ... I'm sure parrot is just "resting" :-) 16:20
FROGGS colomon_: that already happened, TimToady also said something about that
gtodd: I am about to fix the regression btw
colomon_ FROGGS: Ah, well then, burn it.
jnthn JVM is safest when it comes to concurrency stuff :) 16:21
gtodd FROGGS: possibly jvm == most useful for getting people to try perl6 ... or maybe perlito ...
jnthn gtodd: Most people will get bored waiting for it to start up... ;) Unless they have a specific interest in JVM, Moar is the best bet for new users, I think. 16:22
FROGGS yeah, I think so too
"here, install that" ... and three minutes later they can play
gtodd yeah ... 16:23
but hmm eval server + web-ish front end .... (maybe perlito too) and a way to save snippets somewhere ...
16:23 raiph left
FROGGS gtodd: yes, that idea flies around from time to time :o) 16:26
gtodd maybe that would be a perlito front end specific sort of thing for $FUTURE with a web repl (a la Juila) and ways to save edit code in some fancy HTML5 Web30 thingy
could work for perl5 too :-D
16:27 Alula_ joined, slavik joined
gtodd or just use emacs ;-) 16:27
jnthn: is moar the sort of thing that will build/link/run just about anywhere eventually? 16:28
FROGGS jnthn: yes, that's the plan 16:29
16:29 khisanth_ is now known as Khisanth
jnthn gtodd: Well, anywhere its dependencies build, at least... But those are relatively portable. 16:30
16:30 denis_boyun_ joined
gtodd here it seems to be lnked to libc and libmoar ... 16:30
oh and libkvm libm and libthr :-P 16:31
16:32 atroxaper left
gtodd jnthn: anyway pretty cool stuff :-) yay for you 16:32
16:32 atroxaper joined
jnthn :) 16:32
Time to cook some dinner, methinks... bbiab.
gtodd and your helpers of course
jnthn yes, helpers++ :) 16:33
16:34 virtualsue left 16:35 denis_boyun_ left 16:37 atroxaper left, kurahaupo left, Ven left, virtualsue joined, chenryn left, kaleem left 16:39 Rotwang joined 16:54 Akagi201 joined, Akagi201_ joined 16:56 pdcawley left 16:58 Akagi201 left, Akagi201_ left 16:59 denis_boyun_ joined, pdcawley joined 17:04 treehug88 left
[Coke] colomon_: the tests were added when we switched from decrease to less. 17:04
IMO, only makes sense to have stuff like that in roast across perl 6 spec release boundaries, and we haven't crossed one of those yet. 17:05
TimToady tends to agree
17:05 teodozjan left
[Coke] will remove them later tonight, methinks. 17:05
TimToady deprecated deprecations, yum! 17:06
[Coke] is also considering adding all test files to rakudo's spectest.data and fudging them appropriately, or removing the corresponding roast files because they date from pugs.0
that involves work, though, so that won't happen so quickly. :) 17:07
TimToady wonders how hard it would be to write an input method that greps character names and gives you a list that matches to pick from
of course, then we'll discover that nobody can remember what « and » are called 17:08
or how to spell DIAERESIS... 17:09
timotimo isn't that some kind of disease?
17:10 virtualsue left
TimToady most endemic to europe 17:10
*mostly
timotimo sounds dangerous
isBEKaml timotimo: not at all, it's just a tick. 17:12
timotimo: or tic
timotimo :D
17:16 itz_ joined
[Coke] TimToady: you might want to review the unicode modules on the most wanted list to see if those make sense in our new world order. 17:16
(you may have already done some of the work they're asking for.)
17:25 denis_boyun_ left 17:27 denis_boyun_ joined 17:33 atroxaper joined 17:37 pecastro left, atroxaper left 17:39 lue left
zengargoyle should probably start a web page somewhere for "How do I type «»?" 17:47
TimToady how would one search for it? :P
17:47 pdcawley left
timotimo a single serving site? 17:47
zengargoyle cut-n-paste
vendethiel TimToady: symbolhound :-)
[Coke] hang it off perl6.org somewhere, with all the (optional) unicode chars you might have to type, sure. 17:48
TimToady be sure to put the phrase "funny french quotes" :)
geekosaur how do you type those quotes to find out how to type them?
:p
colomon_ geekosaur: first, travel to Paris...
17:49 denis_boyun_ left
timotimo what's a ¢? 17:49
17:49 pdcawley joined
TimToady takes two of 'em to have an opinion 17:49
zengargoyle i moderated a japanese language learning forum a long time ago and the whole "How do I type things not on my keyboard" question was sorta complicated to answer.
[Coke] timotimo: american cent sign. 100¢ = 1 USD 17:50
timotimo [Coke]: wasn't looking for a serious answer
vendethiel on windows, it's especially hard :(
timotimo vendethiel: no, on windows you have alt+numbers 17:51
TimToady yeah, windows doesn't make cents
vendethiel exactly why it's hard.
17:51 ajr_ left
vendethiel I remember the combinations for é, è, ï ... that's a terrible failure 17:51
isBEKaml yeah, windows doesn't make sense.
17:51 denis_boyun joined
timotimo quite the contrary 17:51
windows has brought loads and loads of people closer to the topic of computing 17:52
that is a sensible thing to do
vendethiel ... and is totally irrelevant to typing not-on-my-keyboard characters ?
timotimo before windows, people were entering their machine code by flipping switches and hitting the "store byte and advance" button
17:52 lue joined
zengargoyle i'm hoping things have gotten better with Win7+ 17:52
jnthn Every other Windows is non-terrible.
timotimo zengargoyle: as always, skip every second one
jnthn 8 is a terrible one.
isBEKaml jnthn: yeah, before that it was vista 17:53
vendethiel doesn't think there's any difference between 8 and 7 ...
jnthn Right.
isBEKaml and Me before it.
jnthn Me sucked.
vendethiel W Millenium!
isBEKaml jnthn: there were 2 releases that flopped successively. '98 and 'Me. 17:54
timotimo jnthn: it should be "I sucked", though
TimToady is glad they haven't named an element Millenium...
isBEKaml jnthn: '95 and ten years later, it was XP.
TimToady XP gave everyone experience points 17:55
[Coke] I have that will smith album. it's pretty good.
zengargoyle i'm happy to have pretty much never used Windows much at all
but sad that it's a total Mystery OS to me 17:56
17:56 hoverboard is now known as hovercraft
colomon_ hated Vista so much he will probably not be coming back to Windows for a long, long time. 17:59
geekosaur win7 was actually a pretty good OS 18:00
timotimo colomon_: to be fair, win7 is pretty solid
geekosaur win8, the less said, the better
isBEKaml zengargoyle: Don't fret too much. If you read Neal Stephenson's "In the beginning was the command line", I'm sure much of it will make sense. :-)
PerlJam But Perl 6 is awesome!
:-)
zengargoyle oh for sure. my path has been Apple II, Amiga, Solaris, Linux 18:01
geekosaur (in win8.1 you can at least partially deactivate metro) 18:02
vendethiel sometimes doesn't understand that hate for vista
might be because I got it pretty late, with one or two service packs, but it never crashed on me ..
geekosaur that would be why
vista sp2 was fairly solid
earlier versions, forget it
TimToady my wife has been running Vista for a long time, and it never gave her any trouble as a user 18:03
timotimo my cousin's computer bluescreened when we installed a driver on it ...
i laughed pretty hard
18:04 Possum left
colomon_ TimToady: my Vista complaints were as a programmer. I think my final straw was the fact that "patch" didn't work anymore. 18:05
18:06 [Sno] joined
TimToady nobody in their right mind does *programming* on Windows... <whistles> 18:06
vendethiel sighs 18:07
which reminds me -- farewell, 2014.03-parrot :D !
TimToady :D
isBEKaml TimToady: that Petzold book - it still sells. :D 18:08
TimToady: First it was "Programming Windows" with C++ and then it was with C#.
TimToady
.oO("it's not that the bear dances well, but that it dances at all")
18:09
18:21 molaf left 18:25 guru joined 18:26 guru is now known as Guest37603, Guest37603 is now known as ajr_ 18:27 spider-mario left 18:28 jeffreykegler joined 18:30 jeffreykegler left, _Slade left 18:31 _slade joined, Possum joined
Woodi_ evening everyones :) 18:32
I'm a bit missing new R*... is one planned soon or maybe getting monthly R is new way to go ? 18:33
18:37 atroxaper joined, darutoko left
[Coke] R* is not guaranteed every month. 18:37
though we do try.
dalek rl6-roast-data: ab8bc02 | coke++ | / (5 files):
today (automated commit)
18:38
isBEKaml I don't recall R* ever being monthly - although, it was announced to be a semi-annual/annual thing.
18:41 hovercraft left, atroxaper left
betterworld m: my $x = set(); $x (|)= [2014] 18:44
camelia rakudo-moar 25c511: OUTPUT«===SORRY!=== Error while compiling /tmp/o6cg8jCBCD␤Two terms in a row␤at /tmp/o6cg8jCBCD:1␤------> my $x = set(); $x ⏏(|)= [2014]␤ expecting any of:␤ postfix␤ infix stopper␤ infix…»
betterworld should this work?
FROGGS m: my $x = set(); $x (|=) [2014]
camelia rakudo-moar 25c511: OUTPUT«===SORRY!=== Error while compiling /tmp/izJENSejSI␤Two terms in a row␤at /tmp/izJENSejSI:1␤------> my $x = set(); $x ⏏(|=) [2014]␤ expecting any of:␤ postfix␤ infix stopper␤ infix…»
FROGGS std: my $x = set(); $x (|)= [2014] 18:45
camelia std 0f2049c: OUTPUT«===SORRY!===␤Confused at /tmp/oKalT3UYuy line 1:␤------> my $x = set(); $x ⏏(|)= [2014]␤ expecting any of:␤ feed_separator␤ infix or meta-infix␤ infixed function␤ statement modifier loop␤Parse failed␤FAI…»
FROGGS std: my $x = set(); $x (|=) [2014]
camelia std 0f2049c: OUTPUT«===SORRY!===␤Confused at /tmp/QZgII15ldU line 1:␤------> my $x = set(); $x ⏏(|=) [2014]␤ expecting any of:␤ feed_separator␤ infix or meta-infix␤ infixed function␤ statement modifier loop␤Parse failed␤FAI…»
FROGGS betterworld: does not look like
I'd at least expect that std can parse it...
std: my $x = set(); $x (&)= [2014]
camelia std 0f2049c: OUTPUT«===SORRY!===␤Confused at /tmp/0TTj5Gm2MD line 1:␤------> my $x = set(); $x ⏏(&)= [2014]␤ expecting any of:␤ feed_separator␤ infix or meta-infix␤ infixed function␤ statement modifier loop␤Parse failed␤FAI…» 18:46
betterworld would be neat. 18:47
FROGGS .u union 18:50
yoleaux U+222A UNION [Sm] (∪)
masak I don't see offhand why that oughtn't work.
FROGGS m: my $x = set(); $x ∪= [2014]
camelia rakudo-moar 25c511: OUTPUT«===SORRY!=== Error while compiling /tmp/ZTako5zctC␤Two terms in a row␤at /tmp/ZTako5zctC:1␤------> my $x = set(); $x ⏏∪= [2014]␤ expecting any of:␤ postfix␤ infix stopper␤ infix…»
[Coke] anyway, if you want at least monthly updates, it's worth checking out the compiler releases and doing a panda update every month. You could go crazy and switch to github and install nom branch whenever, but that may be too much work. (or you can just ping us, and someone will do a release sooner than later.)
FROGGS m: my $x = set(); $x ∪ [2014]
camelia ( no output )
FROGGS std: my $x = set(); $x ∪= [2014] 18:51
camelia std 0f2049c: OUTPUT«===SORRY!===␤Confused at /tmp/LhncuwjFCG line 1:␤------> my $x = set(); $x ⏏∪= [2014]␤ expecting any of:␤ feed_separator␤ infix or meta-infix␤ infixed function␤ statement modifier loop␤Parse failed␤FAI…»
FROGGS TimToady: ---^ ?
18:54 treehug88 joined
TimToady STD doesn't know the set operators yet 18:54
and you can't modify a set because it's immutable
masak um, but isn't that just modifying a container? 18:55
jnthn You'd be putting a new resulting set into the contariner there, though...
FROGGS ohh, that's set, err, sad
TimToady yes, we could fake it with the container, as we do with strings
FROGGS I always that the ??= works with every infix
thought* 18:56
TimToady rakudo doesn't parse op= properly yet
masak TimToady: and numbers. += doesn't modify the original number either.
TimToady: in fact, I can't think of any op= that goes in and destroys the original value.
18:57 pecastro joined
TimToady agrees it should work in this case 18:57
betterworld I think the uppercase Set is mutuble, but that doesn't work either with (|)= 18:58
TimToady p6: my @a = 1,2,3; @a [X*]= 4; say @a 18:59
camelia rakudo-jvm 25c511: OUTPUT«(timeout)»
..niecza v24-109-g48a8de3: OUTPUT«4 8 12␤»
..rakudo-{parrot,moar} 25c511: OUTPUT«===SORRY!=== Error while compiling /tmp/tmpfile␤Preceding context expects a term, but found infix = instead␤at /tmp/tmpfile:1␤------> my @a = 1,2,3; @a [X*]=⏏ 4; say @a␤»
TimToady niecza++
19:00 isBEKaml left
TimToady Set is not mutable either; set() is just Set.new in disguise 19:00
jnthn Rakudo's parsing for infix= looks very much like STDs, so I@m not sure what's going on.
oh, wait...not quite
TimToady std: my @a = 1,2,3; @a [X*]= 4; say @a
camelia std 0f2049c: OUTPUT«ok 00:01 126m␤»
jnthn I can see why the latter doesn't work 19:01
I can't see why the set ones don't.
19:02 Rotwang left
jnthn STD has an awful extensibility-killing hack, alas... 19:02
[ <?before '='> <?{ $infix = $<infix>; }> <infix_postfix_meta_operator($infix)>
So infix postfix meta-operators can only start with = :P
TimToady yes, '=' is a bit of a special case
jnthn OK, we're missing that special-casing in Rakudo...what is actually demanding the before, and not letting it fall out of LTM?
oh, the need for the block maybe... 19:03
TimToady I don't remember why, offhand
jnthn Anyways, seems our infixish could do with a little re-org...
19:03 treehug88 left
TimToady could've been some limitation of the P5 support for doing fancy stuff like passing the infix match to the subrule 19:06
or maybe it was somehow related to making sure that the LTM didn't suffer an exponential explosion somewhere 19:07
as it is, that one case pretty much doubles the number of infixes
if you try to include the = in the LTM, which I don't think it tries to anymore 19:08
so it might be a fossil
zengargoyle i sorta thought SetHash was the mutable version of Set 19:09
jnthn k
TimToady zengargoyle: it is
jnthn Still got $dayjob things to tend to for now, so can't look at it for the moment.
TimToady you can do $set<foo> = True if $set is a SetHash, but not if it's a Set
hoelzro let's say I have a role that uses some helper classes, and now I want to expose those classes for extension to the outside world. My::Role::Inner doesn't seem to work; should it? Or should I just convert the whole thing into a module? 19:10
TimToady and the SetHash could implement the op= directly, instead of relying on a container 19:11
jnthn hoelzro: You can't sanely put non-lexical things inside of a role, due to the genericity of a role.
TimToady much less GC load to avoid creating new set objects continually
hoelzro jnthn: ah, ok
19:12 spider-mario joined
jnthn hoelzro: We should probably make it an error to do so. 19:12
Rather than them just vanish into the ether...
hoelzro heh
jnthn Even then it's all rather suspect to nest things inside the role.
hoelzro I'm beginning to see that
=)
jnthn Because we don't make all the inner things generic either.
TimToady we work very hard to keep our intentional genericity separate from our accidental genericity in p6
jnthn So much can go wrong. 19:13
TimToady p5 was so accidentially generic
hoelzro I found some interesting behavior where it worked fine when the methods were in the "default" consumer class of a role, but when I moved the methods into the role, my script broke
in an unpredictable way
and although I will probably move past this design and this sort of thing should be discouraged, I was thinking on mentioning it as a possible rakudo bug 19:14
jnthn Thing is, the classes inside the role will end up closing over a single copy of the type variable - the latest one...
And any other role params.
hoelzro oh, that's interesting 19:15
jnthn Which is not a semantic I like.
hoelzro now, if my outer role is not parameterized...should it matter?
jnthn Well, all roles are parameterized
19:15 hoverboard joined
TimToady the apocalpyses wanted Forests of Frogs, but we never quite got there, since it entails virtual class names 19:15
jnthn They all get an implicit first parameter which is ::?CLASS
Which is used for the type of the invocant. 19:16
hoelzro oh, duh
jnthn Otherwise, composing a role into a few classes and expecting multi-dispatch to do the right thing wouldn't work out well
(That was actually broken in an earlier impl of roles, 4 years ago.)
hoelzro well, lemme see if I can code up a small example 19:17
my module has gotten quite big
jnthn: if you're curious: gist.github.com/hoelzro/e03fc6ab275563643d65 19:19
don't ask me what madness started this =)
I'll be doing some refactoring after work to fix this 19:20
when you move BUILD into DefaultOptionCanonicalizer, it works fine
19:22 telex left 19:24 isBEKaml joined, telex joined 19:25 hoverboard is now known as hovercraft 19:30 hovercraft is now known as moistcherry 19:32 LordV left 19:37 atroxaper joined 19:41 isBEKaml left 19:42 atroxaper left 19:46 moistcherry is now known as hoverboard 19:47 itz_ left 19:54 skids left
[Coke] colomon_: ping 19:55
colomon_ [Coke]: pochi
er, pong
[Coke] mmhehe
colomon_ is not sure how that happened
19:55 _slade left
[Coke] colomon_: see S32-list/sort.t - there's a commented out test there with your name on it. If that's an invalid test, can we just remove it? 19:55
guesses you typed P O TAB
colomon_ probably 19:56
leftover text editor (TextMate) impulses, cause surely someone already type pong here, right?
[Coke] in fact, looks like that whole block might be wonky. 19:57
ah, no, just the commented out stuff and the _2 class. 19:58
removing it.
colomon_ I dunno.
it's all on the weird side
For sure defining multi sub cmp(RT71258_2 $a, RT71258_2 $b) shouldn't affect sort 19:59
dalek ast: 89d6c1b | coke++ | S32-list/sort.t:
Remove commented out invalid test/unused class
20:00
[Coke] colomon++
colomon_ on the flip side, it would be nice if multi sub infix:<cmp> *did* affect sort. but it doesn't, does it? 20:01
lue .tell atroxaper assuming L<foo> is a synonym for L<doc:foo> (S26 doesn't say anything about this explicitly), I could see a case for L<C<foo>> being equivalent to L<doc:foo> rather than L<doc:C<foo>>. However, I'd also say L<C<foo>> is wrong, and you deserve whatever broken output results :) You really should do either C<L<foo>> or L<C<foo>|foo>
yoleaux lue: I'll pass your message to atroxaper.
carlin there's an irssi script that lets you do tab-completion on words using /usr/share/dict/words 20:02
[Coke] r: my %h; $line = $?LINE; %h = { a => 1 }; %h = { b => 2 }; 20:03
lue would argue that links should be treated as verbatim even when there's no alternate presentation present.
camelia rakudo-{parrot,jvm,moar} 25c511: OUTPUT«===SORRY!=== Error while compiling /tmp/tmpfile␤Variable '$line' is not declared␤at /tmp/tmpfile:1␤------> my %h; $line⏏ = $?LINE; %h = { a => 1 }; %h = { b => ␤ expecting any of:␤ pos…»
[Coke] r: my %h; %h = { a => 1 }; %h = { b => 2 };
camelia rakudo-{parrot,jvm,moar} 25c511: OUTPUT«Saw 1 call to deprecated code during execution.␤================================================================================␤%h = { ... } called at:␤ /tmp/tmpfile, line 1␤Please use %h = ... instead.␤----------------------------…»
20:04 skids joined
[Coke] 6^H 20:05
rindolf Hi all. 20:08
PerlJam greetings rindolf. How are you this fine day?
rindolf TimToady: are you on Reddit? If not - you should be.
PerlJam: I'm fine - philosophical.
PerlJam: I socialised with some dog owners/etc. today. 20:09
hoverboard fwilson suck it
rindolf A 12 years old boy enjoyed my new Emma Watson factoid.
timotimo what was that?
hoverboard ezech_ centos
TimToady hoverboard: you seem to be on the wrong channel 20:10
20:11 kaare_ left
rindolf PerlJam: are you on reddit? 20:11
TimToady: meow!
hoverboard TimToady, apologies, was trying /amsg without knowing what it did 20:12
TimToady hugme: hug hoverboard 20:13
hugme hugs hoverboard
PerlJam rindolf: I have gotten on reddit. I'm not on it right now.
hoverboard ):
rindolf PerlJam: yes, I meant if you have an active reddit account. 20:14
PerlJam aye. Idon't reddit much though
It turns out that I'm not very social :)
TimToady needs to be on writit, not reddit
rindolf PerlJam: heh, not true.
TimToady: heh.
TimToady or maybe wrottit 20:15
dalek kudo/nom: c55942f | (Elizabeth Mattijsen)++ | src/core/CompUnitRepo/Local/File.pm:
Make CURL::File faster

This gets us back the time we lost in the spectest recently. Part of the speedup is attained by only going through the directory of the CompUnitRepo only once: this behaviour may need tweaking in certain situations.
rindolf TimToady: reminds me of groups.yahoo.com/neo/groups/fc-sol...sages/1301 20:16
TimToady: reddit is kinda like a new (and in most respects - better) Usenet
TimToady: Usenet has become unusable now. 20:17
It was good up to the end of Web 1.0 or so.
[Coke] reads blogs.perl.org/users/chad_exodist_g...oming.html and wonders if we have any fallout. 20:18
rindolf TimToady: you should totally do an "I am Larry Wall, the creator of patch, rn, perl and Perl - ask me anything."
20:20 skids left
FROGGS jnthn: I think I'll know what block triggers the parrot fail in the next hours... 20:22
rindolf Arguably both patch and rn alone were more important than perl and Perl. 20:23
[Coke] I think we're missing a dep in the moarvm rakudo build.
(with -j9, it tries to copy perl6-m into place before the setting is built.
rindolf TimToady: I'd love to see you on ##philosophy too.
TimToady rindolf: self-promotion is not really very high on my todo list these days, nor is writing my memoirs... 20:24
FROGGS [Coke]: ahh, that kind of dep
dalek ast: ccc4e24 | coke++ | S02-types/deprecations.t:
Remove tests for old version of Order

There was never a "released" perl6 spec for this, so we don't have to test/support the old values - they've already been removed from rakudo.
20:25
rindolf And maybe it's me, but I think Perl 6 was good while it lasted and we should now rethink our strategy. It's far too perfectionist and 2nd system effect.
20:25 erkan joined
zengargoyle i think there are already plenty of TimToady answering questions and rambling alredy on YouTube. :P 20:25
TimToady rindolf: it's just you :P
rindolf TimToady: OK, you can still be on reddit, and answer questions or post interesting stuff.
TimToady: I kinda have the Cassandra syndrome. 20:26
[Coke] rindolf: I imagine you'll not get a lot of traction with that train of thought unless you have specific ideas about what else to do besides work on Perl 6.
FROGGS rindolf: it really is just you :o)
TimToady every time I answer a question on reddit, I forget a Japanese verb :)
rindolf TimToady: or www.shlomifish.org/humour/fortunes/...ybody-else
TimToady: heh, believe in yourself.
TimToady I'd rather believe in using the time I have left usefully. 20:27
rindolf TimToady: "every time you masturbate god kills a kitten", right?
TimToady: ah, you can become rich easily enough.
And then have MOAR time.
TimToady: charge people for public speeches.
lizmat wonders about the future of rindolf 20:28
20:28 beastd joined
TimToady is not the least bit interested in making a lot of money 20:28
rindolf TimToady: or collect donations after that - stand-up philosophy.
TimToady: don't you want more time?
FROGGS rindolf: you can't buy time basically
rindolf lizmat: well, tomorrow i may get hit by a bus.
TimToady money and time are only slightly interfungible
rindolf FROGGS: well, time >>> money.
FROGGS rindolf: only sort of 20:29
[Coke] multi-dim arrays are NYI, correct?
rindolf I have a lot of time because i'm happily unemployed, but i need money.
lizmat [Coke]: yes
rindolf My parents mostly support me.
TimToady my most precious resource is my energy, and I have only so much of that to spend each day
rindolf and also national insurance.
TimToady: I have more and more energy.
And i'm 37.
And don't exercise a lot.
PerlJam rindolf: this isn't a self-help group ;) 20:30
TimToady wanders off to get more energy from lunch...
dalek ast: ef44f78 | coke++ | S02-types/array.t:
use standard "NYI" text
lizmat TimToady++ rindolf--
rindolf TimToady: om nom nom.
lizmat: :-)
lizmat++ # Balance in the force.
lizmat: maybe TimToady should work half-time. 20:31
PerlJam lizmat++ TimToady++ (because there still isn't enough karma for some people)
FROGGS rindolf: perhaps it is better to end that discussion
rindolf PerlJam++
FROGGS: sure, sorry.
lizmat is reminded of Jar Jar Binks
rindolf IS there a #TimToady ?
So ... how's the weather? 20:32
lizmat should there be a #rindolf ?
FROGGS rindolf: I'm not sure he'd enjoy that
lizmat news at eleven
rindolf lizmat: it was sometimes created ad-hoc.
lizmat: there's #perlcafe and stuff.
Everyone is welcome.
[Coke] r: our sub eval_born { 5 }; OUR::eval_born()
rindolf Everyone is welcome to join #perlcafe .
camelia rakudo-jvm 25c511: OUTPUT«No such method 'Any' for invocant of type 'Parcel'␤ in any at gen/jvm/BOOTSTRAP.nqp:1661␤ in block at /tmp/tmpfile:1␤␤» 20:33
..rakudo-parrot 25c511: OUTPUT«No such method 'Any' for invocant of type 'Parcel'␤ in any at gen/parrot/BOOTSTRAP.nqp:1673␤ in block at /tmp/tmpfile:1␤␤»
..rakudo-moar 25c511: OUTPUT«Cannot find method 'Any'␤ in block at /tmp/tmpfile:1␤␤»
lizmat decides to forget Jar Jar Binks
rindolf kinda hates speaking in riddles.
lizmat: who can forget him?He's awesome.
jnthn kinda hates rindolf speaking...
rindolf lizmat: www.debunking911.com/questions.htm - see this - he's running for Senate. 20:34
jnthn: sorry.
[Coke] r: say &OUR::bar() # RT #120930
synopsebot Link: rt.perl.org/rt3//Public/Bug/Displa...?id=120930
camelia rakudo-jvm 25c511: OUTPUT«No such method 'Any' for invocant of type 'Parcel'␤ in any at gen/jvm/BOOTSTRAP.nqp:1661␤ in block at /tmp/tmpfile:1␤␤»
..rakudo-parrot 25c511: OUTPUT«No such method 'Any' for invocant of type 'Parcel'␤ in any at gen/parrot/BOOTSTRAP.nqp:1673␤ in block at /tmp/tmpfile:1␤␤»
..rakudo-moar 25c511: OUTPUT«Cannot find method 'Any'␤ in block at /tmp/tmpfile:1␤␤»
rindolf jnthn: anytihng you're working on?
FROGGS m: our sub eval_born { 5 }; say OUR:: 20:35
camelia rakudo-moar 25c511: OUTPUT«("\&eval_born" => sub eval_born () { #`(Sub|140451773726160) ... }).hash␤»
20:35 skids joined
FROGGS m: our sub eval_born { 5 }; say OUR::eval_born 20:35
camelia rakudo-moar 25c511: OUTPUT«(Any)␤»
FROGGS m: our sub eval_born { 5 }; say OUR::eval_born()
camelia rakudo-moar 25c511: OUTPUT«Cannot find method 'Any'␤ in block at /tmp/u6UhQiyEs7:1␤␤»
FROGGS m: our sub eval_born { 5 }; say OUR::<&eval_born()>
camelia rakudo-moar 25c511: OUTPUT«(Any)␤»
FROGGS m: our sub eval_born { 5 }; say &OUR::eval_born()
camelia rakudo-moar 25c511: OUTPUT«5␤»
rindolf ok , maybe I Should leave and chat to TimToady on PM. 20:36
20:36 dmol left
PerlJam rindolf: or let him eat his lunch in peace? 20:36
[Coke] use Test; eval_lives_ok 'our sub eval_born { 5 }', 'can define a sub in eval'; is &OUR::eval_born(), 5
rindolf Carpe Diem! "He who saves one soul has saved the world entire" -- the JEwish Oral Torah.
[Coke] r: use Test; eval_lives_ok 'our sub eval_born { 5 }', 'can define a sub in eval'; is &OUR::eval_born(), 5 20:37
camelia rakudo-moar 25c511: OUTPUT«ok 1 - can define a sub in eval␤Cannot find method 'Any'␤ in block at /tmp/tmpfile:1␤␤»
..rakudo-jvm 25c511: OUTPUT«ok 1 - can define a sub in eval␤No such method 'Any' for invocant of type 'Parcel'␤ in any at gen/jvm/BOOTSTRAP.nqp:1661␤ in block at /tmp/tmpfile:1␤␤»
..rakudo-parrot 25c511: OUTPUT«ok 1 - can define a sub in eval␤No such method 'Any' for invocant of type 'Parcel'␤ in any at gen/parrot/BOOTSTRAP.nqp:1673␤ in block at /tmp/tmpfile:1␤␤»
rindolf PerlJam: sure, we'll talk when he returns.
PerlJam: I'm still here for the time being.
20:38 araujo joined, araujo left, araujo joined
[Coke] m: EVAL "our sub eval_born {5}"; eval_born.say 20:38
camelia rakudo-moar 25c511: OUTPUT«===SORRY!=== Error while compiling /tmp/or5aRkpvmW␤Undeclared routine:␤ eval_born used at line 1␤␤»
[Coke] m: EVAL "our sub eval_born {5}"; eval_born().say
camelia rakudo-moar 25c511: OUTPUT«===SORRY!=== Error while compiling /tmp/ZJ_7FnJ1fP␤Undeclared routine:␤ eval_born used at line 1␤␤»
[Coke] so, this is related to the LTA message, but not quite the same. 20:39
FROGGS [Coke]: you should use single quotes for one thing
jnthn It doesn't actually get as far as the eval 20:40
*EVAL
[Coke] m: EVAL 'our sub eval_born {5}'; eval_born().say
camelia rakudo-moar 25c511: OUTPUT«===SORRY!=== Error while compiling /tmp/yShBdJhkAy␤Undeclared routine:␤ eval_born used at line 1␤␤»
FROGGS m: EVAL 'our sub eval_born {5}'; &::('eval_born)().say
camelia rakudo-moar 25c511: OUTPUT«===SORRY!=== Error while compiling /tmp/Lct11FY1Lr␤Unable to parse expression in single quotes; couldn't find final "'" ␤at /tmp/Lct11FY1Lr:1␤------> ub eval_born {5}'; &::('eval_born)().say⏏<EOL>␤ expectin…»
[Coke] jnthn: looking at S02-names/our.t
FROGGS m: EVAL 'our sub eval_born {5}'; &::('eval_born')().say
camelia rakudo-moar 25c511: OUTPUT«5␤»
[Coke] first test thinks you can do OUR::Eval_born() after that EVAL 20:41
FROGGS [Coke]: only at runtime
m: EVAL 'our sub eval_born {5}'; &::('OUR::eval_born')().say
camelia rakudo-moar 25c511: OUTPUT«5␤»
[Coke] shall I update the test to use &:: ?
FROGGS either that or do not use EVAL 20:42
[Coke] part of the point is to use EVAL.
"call to eval-born our sub via OUR works" 20:43
FROGGS yeah, then you have to invoke it indirectly...
err
OUR:: is still compile time
jnthn m: EVAL 'our sub eval_born {5}'; say OUR::eval_born()
camelia rakudo-moar 25c511: OUTPUT«Cannot find method 'Any'␤ in block at /tmp/dBoipup99e:1␤␤»
jnthn ugh
That *is* an LTA error. 20:44
FROGGS or should OUR:: be runtime?
[Coke] there is already a ticket for the LTA error.
jnthn OK
[Coke] rt.perl.org/Ticket/Display.html?id=120930
jnthn Sorry, only half able to follow
rindolf Bye all, this channel has too low Signal to noise ratio.
20:44 rindolf left
FROGGS O.o 20:44
[Coke] yah, shame we talk so much about getting perl 6 done. 20:45
<eyeroll>
lizmat *phew*
FROGGS ahh
m: EVAL 'our sub eval_born {5}'; say &OUR::eval_born()
camelia rakudo-moar 25c511: OUTPUT«5␤»
FROGGS now I get it
jnthn oh!
Well, at least that works, yes...
[Coke] odd.
FROGGS 5, yeah :o) 20:46
jnthn I think it's compiling th eother one into a coercion.
masak does rindolf have a known problem of some kind? just curious.
[Coke] m: use Test; eval_lives_ok 'our sub eval_born {5}' ; is &OUR::eval_born(), 5
camelia rakudo-moar 25c511: OUTPUT«ok 1 - ␤Cannot find method 'Any'␤ in block at /tmp/wxx1hRFDi2:1␤␤»
FROGGS (roughly five steps until I know what explodes in perl6-p)
20:46 FOAD left
[Coke] ^^ why does that work in FROGG's standalone, but not in the testS? 20:47
20:47 skids left
FROGGS [Coke]: the test is ok 20:47
but there is no desription
description, even
m: use Test; eval_lives_ok 'our sub eval_born {5}' ; is &OUR::eval_born(), 5, 'heureka'
camelia rakudo-moar 25c511: OUTPUT«ok 1 - ␤Cannot find method 'Any'␤ in block at /tmp/PsHy4IGHVR:1␤␤»
[Coke] m: use Test; eval_lives_ok 'our sub eval_born {5}', "two" ; is &OUR::eval_born(), 5, "two"
camelia rakudo-moar 25c511: OUTPUT«ok 1 - two␤Cannot find method 'Any'␤ in block at /tmp/2XqdNmncZ6:1␤␤»
FROGGS ummm
[Coke] m: use Test; eval_lives_ok 'our sub eval_born {5}', "er, one" ; is &OUR::eval_born(), 5, "two"
camelia rakudo-moar 25c511: OUTPUT«ok 1 - er, one␤Cannot find method 'Any'␤ in block at /tmp/aRXOIGg6GR:1␤␤»
[Coke] ^^ nope. 20:48
FROGGS m: EVAL 'our sub eval_born {5}'; say &OUR::eval_born() == 5
camelia rakudo-moar 25c511: OUTPUT«True␤»
lizmat masak: www.shlomifish.org/meta/FAQ/#mental_illness
masak lizmat: thank you, that was exactly what I was looking for.
PerlJam lizmat++ I hadn't realized rindolf put that stuff in a FAQ
[Coke] m: use Test; EVAL 'our sub eval_born {5}'; is &OUR::eval_born(), 5 20:50
camelia rakudo-moar 25c511: OUTPUT«ok 1 - ␤»
[Coke] m: use Test; eval_lives_ok 'our sub eval_born {5}'; is &OUR::eval_born(), 5
camelia rakudo-moar 25c511: OUTPUT«ok 1 - ␤Cannot find method 'Any'␤ in block at /tmp/6PoLkNCcU7:1␤␤»
[Coke] so, EVAL works, but eval_lives_ok barfs it?
FROGGS [Coke]: I know it... 20:51
eval_lives_ok introduces a new scope
[Coke] so, have to rewrite that test to use try EVAL ?
jnthn Well, this is about our scoping, and eval_lives_ok is in the Test module 20:52
[Coke] (and then an OK if no $!, with a note about why?
jnthn: would you say the fix is to change eval_lives_ok? 20:53
jnthn tbh, I'd probably just lives_ok { EVAL '...' } 20:54
[Coke] danke.
20:55 FOAD joined 20:56 denis_boyun left
dalek ast: 188a225 | coke++ | S02-names/our.t:
Avoid introducing extra scope; hides our sub

  jnthn++ FROGGS++
20:58
20:58 denis_boyun_ joined
lizmat FROGGS: in CompUnitRepo there is a method "files" 20:59
FROGGS lizmat: yes
[Coke] there, only 991 rakudo directives that don't have an RT or a NYI associated with them. :|
lizmat it seems to be the same as candidates
I'm trying to grok what you meant by it
is this some attempt at allowing %?RESOURCE ?
FROGGS lizmat: candidates is about namespaces, .files is about resources 21:00
resources also mean binaries btw
because you query for CUR.files('bin/prove')
lizmat is that being done anywhere yet ? 21:01
FROGGS and that already works, and is used by the wrappers that are put in the bin folder
[Coke] m: say Junction; say junction;
camelia rakudo-moar 25c511: OUTPUT«===SORRY!=== Error while compiling /tmp/lhACmcY4tX␤Undeclared routine:␤ junction used at line 1␤␤»
[Coke] m: say Junction;
camelia rakudo-moar 25c511: OUTPUT«(Junction)␤»
FROGGS lizmat: in case you would install a binary into an inst://, yes
lizmat maybe we should spec a %?BIN then ? 21:02
FROGGS lizmat: would that help?
lizmat that would map to 'bin' in the distribution
well, we need to have the same filename mangling properties as we do on resourced and modules 21:03
*resources
FROGGS hmmm no, not really
when you create a dist that provide binary scripts, then these must be latin1, otherwise many ppl won't be able to execute them 21:04
timotimo ... binary scripts ... in latin1?
lizmat the names
FROGGS timotimo: the name you **** :P
*g* 21:05
lizmat I'm not sure that's a valid assumption or a needed one
but your point of not needing to mangle the filename is clear
timotimo ah, now i know what you mean
FROGGS lizmat: well, we can/must mangle it, sure
lizmat but then how would one know the name to type in a CLI ? 21:06
FROGGS but the authors must also keep in mind that a punycode encoded filename of a script is not ideal
timotimo i'm not sure if i asked this before, but do we have something like "give me the folder where my stuff is stored" where stuff refers to ... stuff
like icons/pictures, sound files, assets of that kind
FROGGS the author must name it sanely, that's the point
timotimo static files for a web site
stuff like that
lizmat timotimo: %?RESOURCE 21:07
FROGGS timotimo: I think so, yes
timotimo oh, that's neat
how customizable is that? can a distro (debian/ubuntu/fedora) say "the assets have to go into this folder, please" and it'll work?
lizmat S22:366 and S22:633
timotimo synopsebot!
Mouq atroxaper: I agree that L<C<boom>> should really DWIM
timotimo do you know "physicsfs"? 21:08
FROGGS timotimo: depends in what comp unit repo you install to
synopsebot Link: perlcabal.org/syn/S22.html#line_366
timotimo you can have folders and zipfiles and stuff and "layer" them over each other
lizmat timotimo: I'm still in 2 minds about that
timotimo so that one has a higher priority than the other and you can make patches and mods that way
lizmat on the one hand, I would like to see one "index" in each CURL::Installation that is read once and can be inspected at compile/run time 21:09
but that sort of works against package distributors
timotimo hmm. it's also kinda not so awesome that you have to supply the paths to all the files
lizmat so I was thinking maybe we need a CURL::RPM
that would have an "index" for each installed module 21:10
FROGGS lizmat: that's not against packagers I think... the vendor provides a CUR managed by apt for example, but I as a user install into a CUR in my home
that's how I see things
or I install into site, which is still separate from the vendor one
lizmat yes, but a vendor wants to be able to uninstalle and update 21:11
this is not about where things get installed, but *how*
FROGGS lizmat: sure, the vendor can do that with the vendor repo
lizmat and potentially create their own module for that, indeed
FROGGS lizmat: yes, the vendor can provide its own CUR class
21:11 denis_boyun_ left
lizmat still it would be nice if vendors would not have to do that and be able to use CURL::Installation 21:12
even if that is in a separate dir
Mouq .tell atroxaper github.com/perl6/Pod-To-HTML/commi...07f013a947
yoleaux Mouq: I'll pass your message to atroxaper.
FROGGS yes, we'll see when we are there and talk to ppl like [Sno]++ or dod++
lizmat s/even/but/
21:12 denis_boyun joined
lizmat I discussed this a lot with [Sno] and todd rinaldo 21:13
21:13 gfldex left
FROGGS lizmat: they need to play with it I think 21:13
lizmat indeed, but first I need to make something for them to play with :-)
FROGGS I think there is still a regression in panda due to @*INC refactorings... 21:16
lizmat please let me know what I broke
FROGGS I think there was something about the ordering of lib dirs... which is kinda tricky in panda rebootstrap process 21:17
timotimo *sigh*
FROGGS I have to dig again into that, it is hard to understand and thus hard to explain
21:17 treehug88 joined, denis_boyun left
lizmat FROGGS: also I'm still not sure whether it makes sense to have multiiple CUR objects in a single @*INC entry 21:18
it's going to make it hard to explain
FROGGS okay... at least I found the block in rakudo that triggers the perl6-p blowup
lizmat in my view the first CUR in @*INC that returns any candidate, is the one that will be loaded 21:19
FROGGS lizmat: we need a way to treat a bunch of CURs equally, so we need a way of grouping
lizmat (if there is only 1, if there are more than 1, then it will die telling it can't disambiguate)
why do we need that ?
FROGGS lizmat: then we don't need CUR at all and can switch back to the P5 way
lizmat huh? 21:20
21:20 raiph joined
lizmat could you elaborate (not necessarily now) 21:20
21:20 denis_boyun joined 21:21 rjbs left
FROGGS one major problem of the P5 way is that you get the first dist found in @INC, whether there is a *better* dist in dirs after that 21:21
so you have to "uninstall" stuff from your home or site when you want the better one from vendor or perl 21:22
lizmat no, you need to be more specific in your -use- statement
21:22 dmol joined
FROGGS and there is no way of telling: I have ten CURs, five on my disk, three in the web and two on another box, and I want to pick the *best* version from any of these 21:23
lizmat this is part of the change in thinking we need there
I don't think you need to think about having the system selecting the "best" version of a module
FROGGS lizmat: when I say: use Foo:from<lizmat>, I still get v1.1 from home when there is a v1.2 in site
[Coke] m: my Mu $j; $j = 1|2; say WHAT($j).gist
camelia rakudo-moar 25c511: OUTPUT«any((Int), (Int))␤»
lizmat as a developer, you need to think about the version of a module that you *need* 21:24
but do you *need* 1.2 in your code?
[Coke] ^^ there is a test that assumes that should return (Junction)... but I think the is is autothreading it.
lizmat if so, you should specify that with use Foo:ver(1.2)
ChoHag Developers never need the toys they say.
21:25 kurahaupo joined
lizmat I would even argue that production code should *never* contain a bare "use Foo" in the future 21:25
FROGGS lizmat: what if that is not code I maintain? what if someone releases a hotfix for Foo?
lizmat for that we have "supersedes"
FROGGS allowing to specify version ranges is to allow such scenarios 21:26
jnthn [Coke]: ooh, an LHF just in time for be getting done with $dayjob bits
[Coke] thinks we can remove the comment at the head of Test.pm now.
jnthn [Coke]: Yeah, I see what's up :)
21:26 denis_boyun left
[Coke] test is in S03-junctions/misc.t 21:26
FROGGS lizmat: I think you are forcing restrictions that will not work out
lizmat perhaps, but I don't see that as a restriction 21:27
FROGGS for one because authors release stuff and then just vanish, and also not always think about versioning
lizmat if you specify "use Foo" in production code, you really don't care which version/author you get
FROGGS I see not allowing grouping as a restriction
lizmat but grouping is only needed to resolve version conflicts, allowing the system to select the highest version 21:28
within that group
FROGGS lizmat: correct
it is not about conflicts, it is about crossing CUR boundaries, but yes 21:29
[Coke] jnthn: is the problem with the is or the .gist in Junction?
lizmat my argument is that the system shouldn't be doing this
FROGGS it is about picking the best for your search criteria, and that not locked down to sequential lookup
[Coke] (or the test? :)
lizmat but how can you guarantee that a higher version means better ?
jnthn [Coke]: Problem is with the WHAT actually
FROGGS lizmat: it is not the system, it is me allowing this by configuring groups 21:30
[Coke] dammit. :)
FROGGS lizmat: that is how versions are meant to be
jnthn > my Mu $j; $j = 1|2; say WHAT($j).gist
(Junction)
> my Mu $j; $j = 1|2; say WHAT($j)
(Junction)
(locally)
[Coke] jnthn++ # that was fast.
-> commuting. 21:31
lizmat FROGGS: reading up on S11 again 21:33
21:34 firefish5000 joined
FROGGS lizmat: sorry about my rising temper... perhaps I'd be able to accept it tomorrow :o) 21:34
I see that having groups in @*INC are odd... maybe a CUR::Group object would be cleaner, I dunno 21:35
jnthn: the block of this while relies on inlining: github.com/rakudo/rakudo/blob/nom/...ps.pm#L107 21:37
and it seems not to get inlined because of p6bindsig 21:39
or something else, dunno 21:40
lizmat FROGGS: let me know when we can talk about this tomorrow
FROGGS sure
21:42 bluescreen10 left
jnthn FROGGS: Ahh...that could explain it. 21:44
Well, explain the lack of inlining
Not the dependency on it 21:45
FROGGS yeah, I do not even try to understand parrot in that case anymore...
s/parrot/perl6-p/ 21:46
but yeah, I bisected all blocks that were sorted out, and now I only optimize this block (it is the 311'd block), and the others that match the criteria of your latest patch 21:47
jnthn OK. I'm looking at the optimizer and pondering
I think it should still inline that one
dalek kudo/nom: 049b796 | (Elizabeth Mattijsen)++ | src/core/CompUnitRepo.pm:
CUR.files should call CURx.files, not .candidates
21:52
lue Gah, syngen got up to 67% of my RAM and was about to crash my computer for the third time :( I really need to change the process from "for each step, go through all the files" to "for each file, go through all the steps"
FROGGS ohh yes, that seemed to be a copy&pasto
jnthn FROGGS: gist.github.com/jnthn/0d83712e55339f1361c2 21:54
timotimo lue: if only that were easier :( 21:55
jnthn FROGGS: I'm gonna spectest that here on r-m
dalek kudo/nom: d24fe59 | jnthn++ | src/core/operators.pm:
Fix signature of WHAT and VAR.
jnthn FROGGS: Hopefully it might allow that immediate inlining again and thus help
lue timotimo: shouldn't be too bad in my case. I think the problem is that currently I'm trying to keep 37 synopses (parsed into a simplistic structure) in memory, and by the time I reach #16 it's a bit taxing on the RAM :) 21:56
timotimo is that the only remaining block that is problematic with the inlining optimization b0rkage from recently?
lizmat lue: I have that problem all the time 21:57
timotimo ouch :)
lizmat :-)
lue It wasn't always a problem, it's just happened recently. But I need to make the process more multithreadable anyway :)
masak why didn't we choose to design one of those small, cute languages instead? :P 21:58
'night, #perl6
lizmat gnight masak
FROGGS masak: we can still switch to php6 :o)
it is just... 21:59
not small and not cute
21:59 rjbs joined
lue It would also be really nice if I could use the standard grammar's grammar rules for stuff, but that requires wrapping our heads around "slangs", and all the headaches that come along with it. 22:00
FROGGS yes, hard work that is
lizmat FROGGS: thinking about CUR some more 22:01
FROGGS jnthn: it seems to help \o/ (running spectests now)
lizmat: yeah?
lizmat maybe I should just make a separate CURL::Installation2 that I think matches the spec
FROGGS lizmat: what is wrong with the current?
lizmat it's not up to spec ? 22:02
22:02 virtualsue joined
FROGGS lizmat: is what way? 22:02
what exactly is not up to spec?
lue I'm still mulling over what an AST manipulation language would look like. :)
lizmat not doing smart matching on the authority or version
not mangling file names 22:03
FROGGS lizmat: the smart matching should be fixable without any regression
lizmat: what files do you want to mangle? 22:04
file names*
lizmat all file names
22:04 beastd left
FROGGS lizmat: we do not store files to disk, except for binaries... 22:04
and to mangle these, we'd need a way to mangle them, like by implementing punycode 22:06
punycode has the charm that it is still kinda readable and there is no loss of information 22:07
jnthn FROGGS: No spectest concerns on r-m 22:10
FROGGS jnthn: I'm at S04, one aborting test file, and two other failings test files 22:11
jnthn FROGGS: Is that an improvement?
lizmat jnthn: I see 2 todo's passing: 22:12
jnthn lizmat: Yes, me too :)
lizmat m: WHAT(1|2).gist
camelia ( no output )
lizmat m: WHAT(1|2).gist.say
camelia rakudo-moar c55942: OUTPUT«any((Int), (Int))␤»
jnthn lizmat: I think those are the tests taht [Coke]++ was talking about that led me to make the fix :)
FROGGS jnthn: it feels identical to this hack: gist.github.com/FROGGS/0c85c2e1363abb4a45c1
jnthn: so, yes :o)
lizmat 6 'WHAT(1|2).gist.say'
(Junction)
22:13 huf left
dalek ast: 23f4832 | (Elizabeth Mattijsen)++ | S03-junctions/misc.t:
Unfudge now passing tests
22:14
FROGGS jnthn: feel free to push, can't get worse than this anymore: github.com/coke/perl6-roast-data/b....out#L3258
dalek kudo/nom: fef7933 | jnthn++ | src/Perl6/Optimizer.nqp:
p6bindsig should not prevent immediate inlining

While on MoarVM we can lower signatures and thus optimize this out, on other backends we don't, but should not lose the immediate inline optimization on those. FROGGS++ also tracked down a behavior change that resulted from disabling this optimization on r-p; it should not make a difference, but this does indeed fix the r-p breakage.
jnthn lizmat: Thanks for unfudging. 22:15
lizmat spectesting r-p
jnthn FROGGS: There you go. Thanks a lot for hunting it down sufficiently well that the patch was fairly clear to see. 22:16
FROGGS jnthn: that means we are back to about 20-30 fails, instead of 1800 :o)
timotimo \o/
Mouq lue: Is there anything major left TODO with your Pod::Walker to begin seeing if Pod::To::HTML can't be refactored to use it?
jnthn FROGGS: OK, yay. :) 22:17
FROGGS jnthn: yeah, but it still bothers me that you can provide a patch in seven minutes, when I look at it for days :o)
jnthn FROGGS: Well, debugging is always the time consuming bit for me. 22:18
in 95%+ of the cases, anyway.
FROGGS yeah
lue Mouq: no, the only thing was that I didn't want to publicize it much until I wrote a converter using it. I don't plan on doing that any time soon though, so feel free to try it out! :)
FROGGS perhaps I have a chance some day to get into the optimizer a bit... but that day does not seem near 22:19
jnthn A bunch of the Moar deopt patches were a few lines, but took O(hours) to work out
Mouq's overight bug report is my next victim :) 22:20
lue By the way, I like how "Pod Walker" sounds like a reference to some piece of sci-fi, yet I couldn't possibly guess which one (Star Wars is my best guess, for some reason) :P
timotimo oh no!
FROGGS same for v5... and there are a few nice things still waiting
timotimo monday has come and gone and saw no weekly :(
lizmat :-(
jnthn åhnejs!
timotimo åhne.js? 22:21
FROGGS timotimo: please mention the OpaquePointer and nativecast work of sergot and me :o)
timotimo yes!
jnthn timotimo: *lol*, I have to write that library...
FROGGS timotimo: sergot also connected to his https machine as the first result of openssl test bindings today 22:22
dalek kudo/nom: 056aaf9 | (Elizabeth Mattijsen)++ | docs/ChangeLog:
ChangeLog entry
22:23
FROGGS okay, since perl6-p is in a good shape again I can peek into the JSON::RPC bug and in my Archive::Tar problem...
dalek kudo/nom: e77c12b | jnthn++ | src/Perl6/Actions.nqp:
Ensure lowered_param names are unique.

Not just unique per block, but per file; it seems the optimizer is now smart enough to flatten/inline and thus cause collisions.
22:24
jnthn Mouq: Please try ^^ to see if it fixes the bug you reported earlier; it should. If not, I can take a deeper look.
timotimo ouch %)
lizmat jnthn: are you sure there is no missing $i in there ? 22:25
or was that the point ?
22:25 treehug88 left
FROGGS QAST::Node.unique does that for you 22:26
jnthn lizmat: Well, could do that too...
lizmat: But feels like overkill to do both
lizmat ok, just checking if I understood it correctly :-) 22:27
FROGGS m: use QAST:from<NQP>; say QAST::Node.unique('foo')
camelia rakudo-moar c55942: OUTPUT«foo_1␤»
FROGGS m: { use QAST:from<NQP>; say QAST::Node.unique('foo') }; { use QAST:from<NQP>; say QAST::Node.unique('foo') }
camelia rakudo-moar c55942: OUTPUT«foo_1␤foo_2␤»
Mouq jnthn++ won round 1 22:31
Round 2 appears to be "
Error while compiling op call: Required positionals must come before all optional positionals"
jnthn Mouq: oh, ouch 22:32
Mouq running with --ll-exception
gist.github.com/Mouq/0eaaffdd956075f291ca 22:33
22:34 jnap1 joined
jnthn OK...round 2... :) 22:35
Mouq Seriously, though, jnthn+++ for the quick solution to the first bug :)
22:36 jnap left
jnthn Mouq: I'm currently playing with this patch: gist.github.com/jnthn/eedc2eff5cbaeab83c06 22:37
22:38 cognominal left, cognominal joined
jnthn m: while 1 -> $x { say 42 } 22:39
camelia rakudo-moar d24fe5: OUTPUT«(timeout)42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42␤42…»
jnthn hmm
22:44 spider-mario left
carlin m: say DateTime.new("2014-06-26T22:52:13+12:00") 22:44
camelia rakudo-moar d24fe5: OUTPUT«Invalid DateTime string '2014-06-26T22:52:13+12:00'; use an ISO 8601 timestamp (yyyy-mm-ddThh::mm::ssZ or yyyy-mm-ddThh::mm::ss+0100) instead␤ in method new at src/gen/m-CORE.setting:15422␤ in block at /tmp/hCLW8Cbci0:1␤␤»
22:45 btyler_ left
carlin the colon in the offset is acceptable in ISO 8601 22:45
lue Apparently there's a deprecated thing going on in Pod::To::HTML (that recent %a = {...} change), it's ruining my carefully crafted cursor manipulations! :)
carlin pouts
FROGGS carlin: please file a bug report (and/or provide a PR) 22:46
carlin: here is its code: github.com/rakudo/rakudo/blob/nom/...al.pm#L253 22:47
jnthn Mouq: argh, the patch is bad actually 22:50
It'll bust Parrot again
FROGGS ohh dear
jnthn ugh, Pod::To::HTML has test fails too 22:55
Also the patch doesn't actually fix it 22:56
22:59 SevenWolf joined 23:01 virtualsue left, ajr_ left 23:05 dmol left
jnthn Mouq: Got htmlify.p6 running without that error now :) 23:05
Mouq: Just gotta spectest my fix. 23:06
FROGGS jnthn: I've provided a workaround for the JSON::RPC issues on MoarVM: rt.perl.org/Ticket/Display.html?id...xn-1300057 23:10
lizmat lue: wrt %a = {...} change: would you have rather had it die ? 23:11
I mean, if deprecating it breaks more than just letting it die...
23:12 rurban1 left
lizmat also: it should only mention that after the process is done: so not sure how it could affect cursor manipulation? 23:12
lue lizmat: death wouldn't affect me here, I'm using the module in a shell() call. If I was actually upset about how it ruins everything I'd just add 2>/dev/null :) 23:13
lizmat that's Pod::To::HTML ?
lue Yes.
lizmat must have missed that when going through core modules
lue shell("$*EXECUTABLE_NAME --doc=HTML $_ > $outfile") is what I'm doing, if you're curious. 23:14
jnthn Pod::To::HTML is ecosystem
dalek kudo/nom: b014db2 | jnthn++ | src/Perl6/Optimizer.nqp:
Don't inline if/while pointy blocks.

This should unbreak htmlify.p6 from p6doc, fixing a bug reported by
  Mouq++.
lizmat ah, Pod::To::HTML is not core, that explains 23:15
timotimo aye, as is the UI for p6debug 23:17
jnthn If you drag that in, you'll end up dragging in Term::ANSIColor with it, and so on... 23:19
And have to put them into the build. 23:20
lue Part of me thinks I need to throw out all of syngen's code and start again. Maaaaaaybe write (parts of) it in nqp, depending on how badly I want some of the Perl6::Grammar rules :) 23:21
23:22 yeltzooo4 is now known as yeltzooo, hoverboard left
lizmat lue: could you gist me the deprecation message? 23:22
I can't find what in Pod::To::HTML could cause it 23:23
23:23 rurban1 joined
jnthn %head = { ... } called at: lib\Pod\To\HTML.pm, line 155 23:24
lizmat: ^^
lue lizmat: gist.github.com/lue/ed527858c5102fe26b10 (once I saw it talk about line 156 too)
lizmat my %head := $p.value; # should fix it 23:25
23:25 rurban2 joined
lue I thought perl6-j gave me backtraces on "uninitialized value of type Any" (or UVOTA for short :P). Guess it's still just perl6-p, which I haven't built in quite a while now. 23:26
23:27 rurban1 left
jnthn FROGGS: Hmm...not sure why the workaround helps, but good to know. I'd like to fix the underlying bug really... 23:29
lizmat pushed a fix to Pod::To::HTML
lue ^^^ 23:30
lue thanks lizmat++ :)
jnthn Time for some rest...'night, #perl6
lue jnthn o/
lizmat gnight jnthn
lizmat also calls it a night 23:32
sleep&
lue lizmat o/ 23:33
23:40 atroxaper joined 23:45 atroxaper left, xragnar_ joined, xragnar left, xragnar_ is now known as xragnar