»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.
TimToady r: say 1 + + 00:00
camelia rakudo 61580f: OUTPUT«===SORRY!=== Error while compiling /tmp/qlPxFjHzNi␤Two terms in a row␤at /tmp/qlPxFjHzNi:1␤------> say 1 + +⏏<EOL>␤ expecting any of:␤ postfix␤ infix stopper␤ infix or meta-infix␤ prefix or ter…»
TimToady I guess it figures if there's a prefix, there must be a term there that it can't see :)
r: say 1 +
camelia rakudo 61580f: OUTPUT«===SORRY!=== Error while compiling /tmp/9fEbUlQIUZ␤Two terms in a row␤at /tmp/9fEbUlQIUZ:1␤------> say 1 +⏏<EOL>␤ expecting any of:␤ postfix␤ infix stopper␤ infix or meta-infix␤»
TimToady doesn't even need the prefix
TimToady seems to recall this was recently rakudobugged already 00:01
00:12 Psyche^_ joined 00:16 Psyche^ left 00:21 ajr_ left 00:26 raiph left, raiph joined
japhb__ r: class X::Attribute::Required is Exception { has $.attribute; has $.class; method message() { "{$.class.^name} requires a value for attribute '{$.attribute.name.substr(2)}'" } }; multi trait_mod:<is>(Attribute:D $a, :$required!) { $a.set_build(method (|) { die X::Attribute::Required.new(:attribute($a), :class(self)) }) }; class Foo { has $.foo is required; }; Foo.new() 00:49
camelia rakudo 61580f: OUTPUT«Foo requires a value for attribute 'foo'␤ in method <anon> at /tmp/nrsmeFCfpn:1␤ in block at src/gen/CORE.setting:871␤ in method BUILDALL at src/gen/CORE.setting:826␤ in method bless at src/gen/CORE.setting:815␤ in method new at src/gen/CORE.setting:800␤ …»
TimToady nr: $_ = 'cooper'; $_ [R~]= 'mini'; .say 00:50
camelia niecza v24-95-ga6d4c5f: OUTPUT«minicooper␤»
..rakudo 61580f: OUTPUT«===SORRY!=== Error while compiling /tmp/siLWaxuTN_␤Preceding context expects a term, but found infix = instead␤at /tmp/siLWaxuTN_:1␤------> $_ = 'cooper'; $_ [R~]=⏏ 'mini'; .say␤»
TimToady rakudobug
japhb__ TimToady: Were you serious about wanting a trait name other than 'is required', or would that be good enough for now?
TimToady I would make a trait that long if I wanted to discourage its use :)
japhb__ I think the only serious suggestion people had that was shorter was 'is needed' 00:51
Of course, our collective lack of imagination doesn't mean there isn't a better word out ther. 00:52
*there
r: class X::Attribute::Needed is Exception { has $.attribute; has $.class; method message() { "{$.class.^name} needs a value for attribute '{$.attribute.name.substr(2)}'" } }; multi trait_mod:<is>(Attribute:D $a, :$needed!) { $a.set_build(method (|) { die X::Attribute::Needed.new(:attribute($a), :class(self)) }) }; class Foo { has $.foo is needed; }; Foo.new() 00:53
camelia rakudo 61580f: OUTPUT«Foo needs a value for attribute 'foo'␤ in method <anon> at /tmp/do0tff_7YN:1␤ in block at src/gen/CORE.setting:871␤ in method BUILDALL at src/gen/CORE.setting:826␤ in method bless at src/gen/CORE.setting:815␤ in method new at src/gen/CORE.setting:800␤ in…»
japhb__ Shortens the error message, even ...
Although perhaps X::Attribute::Value::Needed, hmmm 00:54
'is needed' feels so ... wimpy? 00:55
But it is accurate.
diakopter is want
is DOIT
japhb__ my $chocolate is want; 00:56
is used
is undefaulted # Uh ... no. 00:57
diakopter is lack;
japhb__ It's too bad we can't just say 'has $.foo!' 00:59
Because really, that has consistency with params quite nicely.
BenGoldberg r: my $foo!; 01:00
camelia rakudo 61580f: OUTPUT«===SORRY!=== Error while compiling /tmp/7fOC0Wq13_␤Negation metaoperator not followed by valid infix␤at /tmp/7fOC0Wq13_:1␤------> my $foo!⏏;␤ expecting any of:␤ scoped declarator␤ constraint␤ postfix␤…»
japhb__ TimToady: Actually, why didn't you want 'has $.foo!' ? 01:01
Too confusing in the 'has $!foo!' case?
Because that seems not a terribly strong issue, since new() wouldn't expose that attribute.
*default new() 01:02
diakopter or instead of has... 01:03
needs
that's kinda cutesy 01:04
requires? 01:05
demands?
lacks?
hasnt?
that's it.
hasn't Int $!foo; 01:06
japhb__
.oO( Oh, it hasn't, hasn't it? )
 01:07
geekosaur "hasn't" sounds like a candidate for perl6tercal 01:12
TimToady nr: .oO(rename 'Perl 6', 'OUTERCAL') 01:19
camelia niecza v24-95-ga6d4c5f: OUTPUT«===SORRY!===␤␤Undeclared routine:␤ 'rename' used at line 1␤␤Unhandled exception: Check failed␤␤ at /home/p6eval/niecza/boot/lib/CORE.setting line 1502 (die @ 5) ␤ at /home/p6eval/niecza/src/STD.pm6 line 1147 (P6.comp_unit @ 36) ␤ a…»
..rakudo 61580f: OUTPUT«rename is disallowed in restricted setting␤ in sub restricted at src/RESTRICTED.setting:2␤ in sub rename at src/RESTRICTED.setting:12␤ in block at /tmp/JkOENPpZWu:1␤␤»
diakopter INTRACAL
TimToady EXTERCAL 01:20
INTERSTANFORD
diakopter EXTRACAL
japhb__ CALSTANFORD
TimToady nr: $_ = 'fortnight'; s[^] = 'micro'; .say; 01:21
japhb__ Comes with a free axe head
camelia niecza v24-95-ga6d4c5f: OUTPUT«fortnight␤»
..rakudo 61580f: OUTPUT«microfortnight␤»
TimToady nieczabug
nr: $_ = 'fortnight'; s/^/micro/; .say; 01:22
camelia niecza v24-95-ga6d4c5f: OUTPUT«fortnight␤»
..rakudo 61580f: OUTPUT«microfortnight␤»
TimToady interesting
diakopter n: $_ = 'fortnight'; s[^] .= 'micro'(); .say;
camelia niecza v24-95-ga6d4c5f: OUTPUT«Potential difficulties:␤ Useless use of quotes at /tmp/ngAksri_b6 line 1:␤------> $_ = 'fortnight'; s[^] .= 'micro'⏏(); .say;␤␤fortnight␤»
TimToady hah 01:23
diakopter n: $_ = 'fortnight'; $_ = s[^] = 'micro'; .say;
camelia niecza v24-95-ga6d4c5f: OUTPUT«False␤»
TimToady n: $_ = 'fortnight'; $_ ~~ s[^] = 'micro'; .say;
camelia niecza v24-95-ga6d4c5f: OUTPUT«fortnight␤»
diakopter n: $_ = 'fortnight'; $_ = s[^] = '$0micro'; .say;
camelia niecza v24-95-ga6d4c5f: OUTPUT«False␤»
diakopter n: $_ = 'fortnight'; s[^] = '$0micro'; .say;
camelia niecza v24-95-ga6d4c5f: OUTPUT«fortnight␤»
TimToady n: $_ = 'fortnight'; .=subst(/^/,'micro'); .say; 01:24
camelia niecza v24-95-ga6d4c5f: OUTPUT«fortnight␤»
TimToady r: $_ = 'fortnight'; .=subst(/^/,'micro'); .say;
camelia rakudo 61580f: OUTPUT«microfortnight␤»
diakopter n: $_ = 'fortnight'; s/^/micro/; .say;
camelia niecza v24-95-ga6d4c5f: OUTPUT«fortnight␤»
TimToady r: $_ = 'fortnight'; .=subst(/^..../,'micro'); .say;
diakopter n: $_ = 'fortnight'; $_ =~ s/^/micro/; .say;
camelia rakudo 61580f: OUTPUT«micronight␤»
niecza v24-95-ga6d4c5f: OUTPUT«===SORRY!===␤␤Unsupported use of =~ to do pattern matching; in Perl 6 please use ~~ at /tmp/A0VDhzM5Hz line 1:␤------> $_ = 'fortnight'; $_ =~⏏ s/^/micro/; .say;␤␤Parse failed␤␤»
diakopter *facepalm*
TimToady it just fails on 0 width
diakopter n: $_ = 'fortnight'; $_ +~ s/^/micro/; .say; 01:25
camelia niecza v24-95-ga6d4c5f: OUTPUT«Unhandled exception: Cannot parse number: fortnight␤ at /home/p6eval/niecza/lib/CORE.setting line 1536 (die @ 5) ␤ at /home/p6eval/niecza/lib/CORE.setting line 3745 (ANON @ 10) ␤ at /home/p6eval/niecza/lib/CORE.setting line 3747 (NumSyntax.str2num @…»
TimToady must be testing the width as success
diakopter n: $_ = 'fortnight'; $_ ~~ s/^/micro/; .say;
camelia niecza v24-95-ga6d4c5f: OUTPUT«fortnight␤»
diakopter n: $_ = 'fortnight'; $_ ~~ s/^.*/micro/; .say;
camelia niecza v24-95-ga6d4c5f: OUTPUT«micro␤»
diakopter n: $_ = 'fortnight'; $_ ~~ s/^(.*)/$0micro/; .say;
camelia niecza v24-95-ga6d4c5f: OUTPUT«fortnightmicro␤»
TimToady r: $_ = 'fortnight'; .=subst(/^<?before ....>/,'micro'); .say;
camelia rakudo 61580f: OUTPUT«microfortnight␤»
TimToady duh
diakopter n: $_ = 'fortnight'; $_ ~~ s/^(.*)/micro$0/; .say; 01:26
camelia niecza v24-95-ga6d4c5f: OUTPUT«microfortnight␤»
TimToady n: $_ = 'fortnight'; .=subst(/^<?before ....>/,'micro'); .say;
camelia niecza v24-95-ga6d4c5f: OUTPUT«fortnight␤»
diakopter n: $_ = 'fortnight'; $_ ~~ s/^./micro$0/; .say;
camelia niecza v24-95-ga6d4c5f: OUTPUT«Use of uninitialized value in string context␤ at /home/p6eval/niecza/lib/CORE.setting line 1389 (warn @ 5) ␤ at /home/p6eval/niecza/lib/CORE.setting line 269 (Mu.Str @ 15) ␤ at <unknown> line 0 (ExitRunloop @ 0) ␤ at /tmp/IC8Hi0hxZ9 line 0 (ANON @ …»
diakopter n: $_ = 'fortnight'; s/^./micro$0/; .say;
camelia niecza v24-95-ga6d4c5f: OUTPUT«Use of uninitialized value in string context␤ at /home/p6eval/niecza/lib/CORE.setting line 1389 (warn @ 5) ␤ at /home/p6eval/niecza/lib/CORE.setting line 269 (Mu.Str @ 15) ␤ at <unknown> line 0 (ExitRunloop @ 0) ␤ at /tmp/3yeiSuMQp8 line 0 (ANON @ …»
diakopter n: $_ = 'fortnight'; s/^.*/micro$0/; .say;
camelia niecza v24-95-ga6d4c5f: OUTPUT«Use of uninitialized value in string context␤ at /home/p6eval/niecza/lib/CORE.setting line 1389 (warn @ 5) ␤ at /home/p6eval/niecza/lib/CORE.setting line 269 (Mu.Str @ 15) ␤ at <unknown> line 0 (ExitRunloop @ 0) ␤ at /tmp/v_GYeqkiLz line 0 (ANON @ …»
diakopter n: $_ = 'fortnight'; s/^(.*)/micro$0/; .say; 01:27
camelia niecza v24-95-ga6d4c5f: OUTPUT«microfortnight␤»
diakopter n: $_ = 'fortnight'; s/^(.*)/micro$0/.say;
TimToady </spam>
camelia niecza v24-95-ga6d4c5f: OUTPUT«True␤»
diakopter hey, I'm learning. again. 01:30
irclog.perlgeek.de/perl6/2007-08-06#i_76100 01:31
TimToady: ^
my first fuzz attempt 01:32
erm.
irclog.perlgeek.de/perl6/2006-08-11#i_-188291 01:34
^ me joining #perl6 first time I think
01:35 fridim_ joined
diakopter no, it was the day before 01:36
no
01:37 jnap left
diakopter this appears to have been my first utterance irclog.perlgeek.de/perl6/2006-08-08#i_-193391 01:38
TimToady wow, it's the red-letter edition... 01:39
diakopter LOLOLOLOL irclog.perlgeek.de/perl6/2006-08-08#i_-193245
(and TreyHarris' reply)
and thus the pseudo-mock-trolling began.. 01:40
7 years flies by.. or something. 01:41
TimToady how fun flies when you're having time 01:42
01:42 FROGGS_ joined
diakopter I want a very wide graph showing the arcs of every nick's presence and vocalizations on the channel over the years 01:44
TimToady maybe you should implement a computer language that would be good at that :P 01:45
01:46 FROGGS left
diakopter nothing's super-awesome at mysql db processing except mysql 01:46
er, maybe he moved it to pg
wow, I actually have vague memories of backlogging this... at the time. 01:47
TimToady hmm, JVM compile blew up, trying a reconfig
diakopter ot of memory?
*ut 01:48
*out
japhb__ r: say True !?^ False 01:49
camelia rakudo 61580f: OUTPUT«===SORRY!=== Error while compiling /tmp/MZ5LBLRXVw␤Cannot negate ?^ because it is not iffy enough␤at /tmp/MZ5LBLRXVw:1␤------> say True !?^⏏ False␤»
japhb__ Say what?
diakopter yeah that error is pretty iffy 01:50
because it is not THIRTY_THIRD_DEGREE_GNOSTIC_FREEMASON_INNER_CIRCLE_INITIATION_CODEWORD enough 01:51
Perl 6, where even the errors coin you 01:52
TimToady gnostic-in, gnostic out
01:55 grondilu joined
grondilu r: constant fact = 1, [\*] 1 .. *; say fact[4] 01:57
camelia rakudo 61580f: OUTPUT«24␤»
japhb__ I can see where iffy is used in Grammar.nqp and in precedence.pm. But why is ?^ not iffy?
grondilu TimToady: constant fact = 1, [\*] 1 .. *; say fact[4] # is probably simpler for rosettacode.org/wiki/Factorial#via_Sequence
avuserow r: $_ = "abacab"; s:g[(.)] = ord($/[0]); .say; 01:58
camelia rakudo 61580f: OUTPUT«979897999798␤»
avuserow how could I write the above with, say, $a?
diakopter japhb__: if I had to guess [again], I'd conjecture iffiness has something to do with boolifiability
japhb__ My assumption is that 'iffy' roughly means 'produces a boolean result', in which case, just about any ? op should be iffy, neh? 01:59
r: $a = "abacab"; $a .=subst(/(.)/, ord($0)); $a.say; 02:01
camelia rakudo 61580f: OUTPUT«===SORRY!=== Error while compiling /tmp/oqdeeAGkGd␤Variable '$a' is not declared␤at /tmp/oqdeeAGkGd:1␤------> $a⏏ = "abacab"; $a .=subst(/(.)/, ord($0));␤ expecting any of:␤ postfix␤»
avuserow r: my $a = "abacab"; $a ~~ s:g/(.)/{ord $0}/;
camelia ( no output )
avuserow r: my $a = "abacab"; $a ~~ s:g/(.)/{ord $0}/; $a.say
camelia rakudo 61580f: OUTPUT«979897999798␤»
japhb__ r: my $a = "abacab"; $a .=subst(/(.)/, {ord($0)}, :g); $a.say;
camelia rakudo 61580f: OUTPUT«Nominal type check failed for parameter '$s'; expected Cool but got Any instead␤ in sub ord at src/gen/CORE.setting:3438␤ in block at /tmp/DeNoQrtafk:1␤␤»
japhb__ r: my $a = "abacab"; $a .=subst(/(.)/, *.ord, :g); $a.say; 02:02
camelia rakudo 61580f: OUTPUT«979897999798␤»
avuserow r: my $a = "abacab"; say $a.comb(/./).map: ord
camelia rakudo 61580f: OUTPUT«===SORRY!===␤CHECK FAILED:␤Calling 'ord' will never work with no arguments (lines 1, 1)␤ Expected: :(Cool $s)␤»
avuserow r: my $a = "abacab"; say $a.comb(/./).map:. ord
camelia rakudo 61580f: OUTPUT«===SORRY!=== Error while compiling /tmp/wHY5q93qeu␤Two terms in a row␤at /tmp/wHY5q93qeu:1␤------> my $a = "abacab"; say $a.comb(/./).map⏏:. ord␤ expecting any of:␤ method arguments␤ postfix␤ infix st…»
avuserow r: my $a = "abacab"; say $a.comb(/./).map: .ord
camelia rakudo 61580f: OUTPUT«No such method 'count' for invocant of type 'Int'␤ in method reify at src/gen/CORE.setting:7104␤ in method reify at src/gen/CORE.setting:7021␤ in method reify at src/gen/CORE.setting:7021␤ in method reify at src/gen/CORE.setting:7021␤ in method reify at sr…»
japhb__ avuserow: *.ord
avuserow okay, </spam> for me. :)
ohh
thanks japhb__++ :)
japhb__ np 02:03
So ... I'm still confused by iffy. diakopter and I seem to be thinking along the same lines, but Rakudo disagrees. Hmmm, what about std ...
avuserow comb is awesome :)
japhb__ std: say True !?^ False 02:04
camelia std 4cde04e: OUTPUT«===SORRY!===␤Cannot negate ?^ because additive operators are not iffy enough at /tmp/3gFX66L7_k line 1:␤------> say True !?^⏏ False␤Parse failed␤FAILED 00:00 42m␤»
japhb__ OK, at least Rakudo and STD agree. But I'm still confused.
avuserow: that it is. :-)
lue [backlog] I was thinking of a has-like keyword for required attributes too. (something like needs $.this; or wants $.attr;) I still have a sneaking suspicion that "this is a required attribute!" enters the realm of non-default constructors.
japhb__ lue: needs and wants sound like role/interface requirements 02:06
avuserow though I still don't see how (or if it is possible) to do s[foo] = "bar" on something other than $_
japhb__ lue: it just seems really common to want to know that (some subset of) your attributes are guaranteed initialized.
(By the caller, not by a default value) 02:07
lue yeah. If only english had a verb meaning "in possession of something on which additional information is needed by the possessor."
japhb__ r: my $a = "foo"; $a ~~ s[foo] = "bar"; say $a;
camelia rakudo 61580f: OUTPUT«bar␤»
japhb__ avuserow: like that ^^
avuserow huh. thought I tried that.
lue japhb__: the "it's a common desire" is why I'm middle-of-the-road on its necessity. On the one hand, people need it often. OTOH, is it really that much harder to write your own constructor :) ?
TimToady huh, parse went back up from 85 seconds to 91 02:08
japhb__ lue: Not hard, no. Just boilerplate. That's why we have a default new(), after all, to git rid of pointless boilerplate.
diakopter TimToady: we got 7% more features?
japhb__ diakopter: \o/
colomon \o/ 02:09
lue but I do like the consistency of C<has $.attr!> for this feature.
diakopter ..just play along....
japhb__ lue: Yeah, I'm still thinking that's the best idea yet.
(And it wasn't mine, someone else came up with it long ago.)
lue (and I think has $!attr! is the same as marking $a as read-write after using ::=. That is to say, pointless and wrong.)
japhb__ yeah. 02:10
lue After all, you can't set private attrs by default in the first place, so any concern with $!attr! is a non-issue, from where I stand.
japhb__ Hmmm, 'has $.foo!' implies you can Nil it either ...
lue: right
er, *can't Nil it 02:11
I kinda like that.
TimToady grondilu++ 02:13
lue I take it grondilu invented $.attr! ? 02:14
dalek d: a0bcfb1 | larry++ | STD.pm6:
?^ should be marked iffy
 02:16
lue If we go with has $.foo! , we should definitely allow has $.foo? too. 02:17
japhb__ \o/ # ?^ officially iffy
02:27 grondilu left 02:32 chano left 02:35 odoacre joined 02:39 odoacre_ left 02:41 benabik joined 02:45 odoacre left 02:53 xinming left 02:58 odoacre joined 03:00 xinming joined 03:06 raiph left 03:12 xinming left 03:14 xinming joined 03:18 xinming left 03:21 raiph joined 03:22 skids joined 03:24 xinming joined
TimToady lue: I was ++ing grondilu for [\*] 1..* to get factorial 03:27
BenGoldberg r: say [\*] 1..25; 03:33
camelia rakudo 61580f: OUTPUT«1 2 6 24 120 720 5040 40320 362880 3628800 39916800 479001600 6227020800 87178291200 1307674368000 20922789888000 355687428096000 6402373705728000 121645100408832000 2432902008176640000 51090942171709440000 1124000727777607680000 25852016738884976640000 6204484…»
BenGoldberg :)
So I've got a silly question: Why are the Set infix operators defined using names such as "\x2208", instead of "\c[element of]" 03:35
geekosaur because \c[] is only available with icu? 03:36
which is an optional dependency?
BenGoldberg Oh, ok.
TimToady icu is not really a dependency we want to keep 03:39
we've got to have our own unicode definitions anyway for portability to various platforms, so we might as well use that everywhere 03:41
plus ours will be optmized for NFG 03:42
std: say True !?^ False 03:43
camelia std a0bcfb1: OUTPUT«ok 00:00 42m␤»
TimToady afk &
03:45 eternaleye left 03:49 gattscha1do joined 03:52 gattschardo left 03:54 BenGoldberg left 03:58 prevost joined, eternaleye joined 04:00 preflex left 04:02 preflex joined, ChanServ sets mode: +v preflex, btyler joined 04:07 SamuraiJack__ joined 04:17 prevost left 04:23 fridim_ left 04:29 btyler left 04:52 japhb_ left
lue BenGoldberg: the \x form may be more cryptic, but it takes less characters to type. 05:02
:)
05:03 japhb_ joined 05:39 xinming_ joined 05:42 xinming left 05:47 xenoterracide left 05:59 FROGGS[mobile] joined 06:09 darutoko joined 06:19 japhb_ left
[Coke] yawns 06:27
06:30 japhb_ joined
dalek rl6-roast-data: 090c37b | coke++ | / (5 files):
today (automated commit)
 06:32
06:33 denis_boyun_ joined
[Coke] niecza back up to 844 fails 06:34
06:41 japhb_ left 06:45 raiph left, dmol joined 06:47 rindolf joined, GlitchMr left 06:48 GlitchMr joined 06:49 denis_boyun__ joined 06:50 FROGGS_ left, kaleem joined 06:51 denis_boyun_ left 06:53 japhb_ joined 06:55 rindolf left 07:08 iSlug joined 07:11 sciurius joined 07:12 rindolf joined 07:14 zakharyas joined 07:36 camelia joined
moritz r: say 42 07:36
07:36 ChanServ sets mode: +v camelia
camelia rakudo 61580f: OUTPUT«42␤» 07:36
diakopter r: "i'm slow".say 07:37
camelia rakudo 61580f: OUTPUT«i'm slow␤»
diakopter n: "i'm fast".say
moritz diakopter: the first execution with camelia is *always* slow
camelia niecza v24-98-g473bd20: OUTPUT«i'm fast␤»
moritz diakopter: I have no idea why
diakopter me neither :)
07:39 rindolf left
dalek albot/host07: cb218d7 | moritz++ | evalbot.pl:
nqp-moarvm
 07:41
07:41 camelia left
FROGGS[mobile] uhh 07:42
diakopter hee hee
FROGGS[mobile] moritz++
dalek albot/host07: e9922ec | moritz++ | evalbot.pl:
missing comma
diakopter moritz++ indeed
07:42 camelia joined
diakopter moritz: did you see my note about pugs path 07:42
tadzik: ping 07:43
moritz diakopter: yes. There's also a ~p6eval/.cabal/bin/pugs
07:43 ChanServ sets mode: +v camelia
diakopter oh 07:43
moritz diakopter: which I kinda prefer, because it means we can make /root/ permissions more restrictive again
nqp-moarvm: say(42)
camelia nqp-moarvm: OUTPUT«42␤»
moritz rnp: print 42
camelia rakudo 61580f, niecza v24-98-g473bd20: OUTPUT«42» 07:44
..pugs: OUTPUT«perl: warning: Setting locale failed.␤perl: warning: Please check that your locale settings:␤ LANGUAGE = "en_US:en",␤ LC_ALL = "en_US.ISO-8859-1",␤ LANG = "en_US.UTF-8"␤ are supported and installed on your system.␤perl: warning: Falling back to the standa…»
diakopter nqp-m: my $a := 1000000; while $a-- { }
camelia ( no output )
diakopter nqp-m: my $t := nqp::now; my $a := 1000000; while $a-- { }; say(nqp::now - $t);
camelia nqp-moarvm: OUTPUT«No registered operation handler for 'now'␤compile_op»
diakopter nqp-m: my $t := nqp::time_i; my $a := 1000000; while $a-- { }; say(nqp::time_i - $t);
camelia nqp-moarvm: OUTPUT«0␤»
diakopter whaa
nqp-m: my $t := nqp::time_i; my $a := 10000000; while $a-- { }; say(nqp::time_i - $t); 07:45
camelia nqp-moarvm: OUTPUT«5␤»
JimmyZ nqp-m: my $t := nqp::time_i; my $a := 1000000; while $a-- { }; say(nqp::time_n - $t);
camelia nqp-moarvm: OUTPUT«1.059592␤»
moritz nqp-m: my $t := nqp::time_n; my $a := 1000000; while $a-- { }; say(nqp::time_n - $t);
camelia nqp-moarvm: OUTPUT«0.728743␤»
diakopter ok
moritz p: 42 07:46
camelia pugs: OUTPUT«perl: warning: Setting locale failed.␤perl: warning: Please check that your locale settings:␤ LANGUAGE = "en_US:en",␤ LC_ALL = "en_US.ISO-8859-1",␤ LANG = "en_US.UTF-8"␤ are supported and installed on your system.␤perl: warning: Falling back to the standa…»
07:46 camelia left, camelia joined
JimmyZ MoarVM's not built with --optimize 07:46
I think
07:47 ChanServ sets mode: +v camelia
diakopter actually that machine's cpu isn't amazingly fast 07:47
but yes, it should be built with optimize :)
(if it's not)
yay; we can decrement integers... 07:48
07:48 denis_boyun__ left
diakopter nqp-j: my $t := nqp::time_n; my $a := 1000000; while $a-- { }; say(nqp::time_n - $t); 07:48
nqp-jvm: my $t := nqp::time_n; my $a := 1000000; while $a-- { }; say(nqp::time_n - $t); 07:49
camelia nqp-jvm: OUTPUT«0.09299993515014648␤»
diakopter heh.
moritz p: say 42
camelia pugs: OUTPUT«perl: warning: Setting locale failed.␤perl: warning: Please check that your locale settings:␤ LANGUAGE = "en_US:en",␤ LC_ALL = "en_US.ISO-8859-1",␤ LANG = "en_US.UTF-8"␤ are supported and installed on your system.␤perl: warning: Falling back to the standa…»
moritz grml
diakopter moritz: I remember fixing this on feather3 or wherever
07:49 fhelmberger joined
diakopter nqp-m: my $t := nqp::time_n; my int $a := 1000000; while $a-- { }; say(nqp::time_n - $t); 07:51
camelia nqp-moarvm: OUTPUT«0.739564␤»
diakopter nqp-jvm: my $t := nqp::time_n; my int $a := 1000000; while $a-- { }; say(nqp::time_n - $t);
camelia nqp-jvm: OUTPUT«0.0709998607635498␤»
diakopter meh
nqp: my $t := nqp::time_n; my int $a := 1000000; while $a-- { }; say(nqp::time_n - $t);
FROGGS[mobile] can we have a nqp-* command?
camelia nqp: OUTPUT«0.53498911857605␤»
moritz I can cope with the Debian locale system, but Ubuntu's freaks me out 07:52
diakopter they're that different?
moritz yes
JimmyZ the best one should be perl Configure.pl && make CGOTO=1
diakopter JimmyZ: well we need to tell it clang too
moritz FROGGS[mobile]: if somebody patches it in, yes
diakopter looks at the rebuild scdript on host07 07:53
JimmyZ I meant perl Configure.pl --optimize
07:53 ribasushi left
moritz p: say 42 07:53
camelia pugs: OUTPUT«42␤»
moritz \o/
diakopter moritz++ 07:54
p: say /3/
moritz ftr: added line "en_US.ISO-8859-1 ISO-8859-1" to /var/lib/locales/supported.d/local and then dpkg-reconfigure locales
camelia pugs: OUTPUT«<Regex>␤»
diakopter p: say 3~~/3/
camelia pugs: OUTPUT«Error eval perl5: "if (!$INC{'Pugs/Runtime/Match/HsBridge.pm'}) {␤ unshift @INC, '/home/p6eval/.cabal/share/Pugs-6.2.13.20120717/blib6/pugs/perl5/lib';␤ eval q[require 'Pugs/Runtime/Match/HsBridge.pm'] or die $@;␤}␤'Pugs::Runtime::Match::HsBridge'␤"␤**…»
diakopter poo
moritz: that one might be solvable 07:55
moritz: I don't see a rebuild-moarvm.pl do you want me to make one?
moritz rakudo-jvm: say 42
diakopter: yes please
camelia rakudo-jvm: OUTPUT«(signal XCPU)» 07:56
diakopter whee
*giggle*
07:56 ribasushi joined
moritz yes, needs evalserver 07:56
07:58 xinming_ left 08:00 xinming joined
diakopter r: ; 08:06
camelia ( no output )
diakopter JimmyZ: does clang take a very long time to do CGOTO=1 on interp.c? 08:07
apparently so... 08:08
JimmyZ I thought clang supports CGOTO well enought... 08:10
diakopter nqp-m: my $t := nqp::time_n; my int $a := 1000000; while $a-- { }; say(nqp::time_n - $t); 08:15
camelia nqp-moarvm: OUTPUT«0.452568␤»
diakopter better, anyway
still 5x slower than jvm :)
jnthn: ^
jnthn Well, of course. 08:16
JimmyZ diakopter: with CGOTO and --optimize?
moritz nqp-jvm: my $t := nqp::time_n; my int $a := 1000000; while $a-- { }; say(nqp::time_n - $t);
camelia nqp-jvm: OUTPUT«0.06500005722045898␤»
diakopter JimmyZ: yes, and clang
jnthn That's exactly the sort of thing a JIT makes a hot job of.
diakopter I'll try with gcc
08:17 Timbus|Away joined
diakopter nqp-m: my $t := nqp::time_n; my int $a := 1000000; while $a-- { }; say(nqp::time_n - $t); 08:18
08:18 Timbus left
camelia nqp-moarvm: OUTPUT«0.232732␤» 08:18
08:18 Timbus|Away is now known as Timbus
diakopter ^ with gcc 08:18
JimmyZ gcc is better than clang?
diakopter apparently at that benchmark ;)
micromicromicromicromicrobenchmark 08:19
jnthn nqp: my $t := nqp::time_n; my int $a := 1000000; while $a-- { }; say(nqp::time_n - $t); 08:20
camelia nqp: OUTPUT«0.495544195175171␤»
diakopter well, that's something
jnthn yeah, but it's all op dispatch overhead here. 08:21
diakopter yeah, but that's still something
jnthn aye; apparently Moar dispatches ops faster :) 08:22
diakopter wonders how the jvm would do if we disabled its JIT entirely.... ;) 08:23
nwc10 presumably that is acutally possible with an open source JVM. Does Sun's let you do it? 08:24
jnthn Probably "ok" given the bytecode was designed to be interpretable...
diakopter nwc10: we're using Sun's on that box 08:25
jnthn I think tehe's sa flag
diakopter p6eval@host07:~/MoarVM$ java -version
java version "1.7.0_40"
Java(TM) SE Runtime Environment (build 1.7.0_40-b43)
Java HotSpot(TM) 64-Bit Server VM (build 24.0-b56, mixed mode)
jnthn *there's, a
nwc10 bytecode *has* to be interpretable, else it would be stupid, because it would be slower than regular ahead-of-time compiling 08:26
diakopter yep, things run berry berry slowly
08:26 odoacre_ joined
diakopter but startup time is hella fasta! 08:28
jnthn diakopter: really? hm
08:30 odoacre left 08:34 kivutar left 08:35 iSlug left, FROGGS[mobile] left 08:55 iSlug joined 09:02 mathw_ is now known as mathw
mathw Good morning o/ 09:02
moritz good morning mathw 09:03
mathw hey moritz
jnthn o/ 09:04
09:11 iSlug left 09:14 daniel-s joined 09:16 dmol left 09:21 FROGGS[mobile] joined
mathw jnthn! 09:21
I wrote a little Perl 6 at the weekend 09:23
it was... nice
hyphens in variable names 09:24
so good. That's one of the things I like about Lisps as well
jnthn yeah, it looks kinda nice 09:25
09:25 gfldex left
jnthn I guess reading AMOP was the first time I'd read any significant amount of Lisp... :) 09:25
09:27 gfldex joined 09:28 daniel-s left
moritz uh, rakudo-parrot spectest looks quite dirty 09:28
lots of fallout in set/bag tests, but also S32-io/note.t and S02-types/declare.t 09:29
and one in S05-metasyntax/delimiters.t
jnthn The note one aside, I'm guessing fallout from the set/bag renaming stuff
moritz ah, my t/spec got stuck :/ 09:30
never mind then
09:31 daniel-s joined
moritz but my desktop froze during the spectest run for a while, so I guess there's still something wrong 09:34
09:39 daniel-s left
tadzik diakopter: pong 09:45
09:46 gfldex left
moritz t/spec/S02-types/array.t uses excessive amounts of memory after test 81 10:02
I set a ulimit to 2GB, and it runs into that limit
ah, it's dies_ok { Mu.[0] }, 'but Mu has no .[]'; 10:03
.to lizmat it seems that Mu.[0] causes infinite recursion; maybe related to the postcircumfix rework?
yoleaux moritz: I'll pass your message to lizmat.
10:06 rjbs- left, rjbs joined 10:10 daxim joined
moritz r: Mu.[0] 10:11
camelia rakudo 61580f: OUTPUT«Cannot call 'postcircumfix:<[ ]>'; none of these signatures match:␤:(\SELF, int $pos)␤:(\SELF, int $pos, Mu \$BIND)␤:(\SELF, int $pos, :delete(:$delete)!, *%other)␤:(\SELF, int $pos, :exists(:$exists)!, *%other)␤:(\SELF, int $pos, :kv(:$kv)!, *%other)…»
moritz hopes the resource limits work on the new box 10:12
huh.
moritz gets spam with "dispose of that old mop", and thinks "what's wrong with our MOP?" 10:14
jnthn oh, I'm getting loads of mop spam too 10:15
Apparently MOPs can spread bacteria too 10:16
.oO( Who on earth is going to buy a new mop 'cus they get spam about it???! )
10:16 kivutar joined
moritz who on earth is going to buy anything because they get spam about it? 10:17
tadzik enough people so it makes sense to send spam 10:18
moritz and that's really the heart of the problem 10:19
jnthn I know, it's just the choice of mops as a thing to spam about... :) 10:20
mathw yeah maybe the market for genital enlargement products has peaked 10:24
moritz
.oO( peak oil was yesterday. Today it's peak ... )
 10:26
10:27 rindolf joined 10:30 denis_boyun joined 10:31 dmol joined 10:33 denis_boyun left 10:34 denisboyun joined, gfldex joined
JimmyZ saw jnthn++'s comments to AMOP in google book 10:47
10:51 camelia left 10:52 camelia joined, ChanServ sets mode: +v camelia 10:55 kivutar left
moritz ok, camelia now runs inside a tmux session on host07.perl6.com 10:56
you can get to it as user p6eval with tmux attach -t 2 10:57
nwc10 how do I ask camelia which backends she has?
moritz camelia: help
camelia moritz: Usage: <(rakudo-jvm|nqp-jvm|star|pugs|nqp|std|niecza|rakudo|b|nqp-moarvm|nom|r|nqp-m|nqp-mvm|pnr|p6|nr|n|npr|perl6|prn|rn|r-j|r-jvm|p|rnp|nrp|rpn)(?^::\s) $perl6_program>
10:57 daxim left
moritz nwc10: it also lists all aliases, which is why it's a bit messy :-) 10:57
jnthn r-j: say('hi from jvm') 10:58
camelia rakudo-jvm: OUTPUT«(signal XCPU)» 10:59
jnthn aww
moritz I didn't say they all *work* :-)
10:59 kivutar joined
nwc10 moritz: feature request - could they all work please? :-) 10:59
jnthn :P
moritz jnthn: just decrease startup time of rakudo-jvm enough to be usable with the existing CPU limits, and then that wouldn't happen anymore :-)
nwc10 nice answer 11:00
moritz though in the short term we'll likely use an evalserver
does the evalserver mix with --setting=RESTRICTED?
jnthn I think so
11:01 kivutar left
colomon woah 11:02
stage parse went from 108 two nights ago to 135 last night. :(
11:03 daxim joined
jnthn huh 11:03
We merged stuff yesterday that got rid of almost a million Cursor allocations and it was generally reported as a win. What changed since then? 11:04
colomon hmm…. oh, it's possible I had a $work test running to slow things down? 11:05
jnthn Dunno
colomon oh, that's looking like a real possibility.
jnthn I know if I build multiple things at once then they get slower, though :)
colomon give me a moment. 11:06
jnthn k
moritz some of those modern CPUs have dynamic overclocking, which only works if just a single core is active (due to thermal restraints)
colomon My $work tests generally run on four cores at once 11:07
(nothing fancy, I just have 4 instances of the test going.) 11:08
jnthn I suspect that the real bottleneck ends up being not cores, but memory access
11:11 denisboyun left
colomon will have another number for you in a moment 11:11
nwc10 jnthn: is this a bug in the JVM, or just running out of memory? 11:17
java.lang.ArrayStoreException: <unknown> in finish_code_object (src/Perl6/World.nqp:1070)
(etc)
jnthn nwc10: That's a "huh wtf" 11:19
nwc10 OK, quite possibly a JVM bug 11:20
jnthn docs.oracle.com/javase/7/docs/api/j...ption.html # docs for it
nwc10 that's the first run in a while that didn't coredump the JVM and issue an "if you want to report a bug" report
jnthn JVM bugs aside, it's a little hard to imagine a scenario where that arises on one machine but not another... 11:21
moritz ok, I have now twice reverse-engineered how the JVM evalserver works 11:30
(in the past)
I swear that if I do it a third time (for camelia), I'll actually document it 11:31
11:31 gattscha1do left
colomon parse time on this trial 97 seconds, so that's a 10% shaved off since two days ago. jnthn++ 11:35
timotimo so, are we going to reduce the initial size of ListIter's internal counter thingie down from 100_000? 11:49
i'm still in favor of giving qrpa a very big factor for downsizing; something like "reduce to 1/2 the size if less than 1/16 of the slots are in use" 11:50
cool, my rakudo-parrot parse time is down to 106 seconds from about 120s 11:51
timotimo should pull in the newest changes as well 11:52
11:52 iSlug joined
moritz ListIter stores a list of slots? 11:53
either that sounds confused, or I am confused
timotimo it has a $rpa that stores things
if $eager is true, it gets pre-sized to 100_000
donaldh found out yesterday evening, that even before the first line in the evalserver is run, 15 of those 100_000 element big arrays get allocated (and not yet freed) 11:55
lizmat good *, #perl6! 11:58
yoleaux 10:03Z <moritz> lizmat: it seems that Mu.[0] causes infinite recursion; maybe related to the postcircumfix rework?
lizmat moritz: All tests successful.
Files=1, Tests=99, 8 wallclock secs ( 0.03 usr 0.01 sys + 7.17 cusr 1.21 csys = 8.42 CPU)
this is on parakudo, are you using jakudo?
moritz lizmat: parakudo 11:59
lizmat weird, lemme check
jnthn r: Mu.[0]
camelia rakudo 61580f: OUTPUT«Cannot call 'postcircumfix:<[ ]>'; none of these signatures match:␤:(\SELF, int $pos)␤:(\SELF, int $pos, Mu \$BIND)␤:(\SELF, int $pos, :delete(:$delete)!, *%other)␤:(\SELF, int $pos, :exists(:$exists)!, *%other)␤:(\SELF, int $pos, :kv(:$kv)!, *%other)…»
moritz Files=1, Tests=99 that's not a full spectest run, no?
lizmat nope, just that file 12:00
moritz oh 12:01
hm, weird
I also did a 'git clean -xdf' in all the repos involved
lizmat it *does* take about 3 seconds to fail 12:02
moritz lizmat: locally start a shell with a ulimit -v 1048576
lizmat and all CPU, so maybe you're just seeing it more
moritz lizmat: and then try ./perl6 -e 'Mu.[0]'
lizmat it seems to grow about 250 MB for that 12:03
moritz so there's something wrong with that
lizmat $ ulimit -v 1048576
$ perl6 -e 'Mu.[0]'
Cannot call 'postcircumfix:<[ ]>'; none of these signatures match:
and a long list of signatures 12:04
moritz doesn't understand this at all
lizmat ICU related maybe ? 12:05
moritz dunno, would surprise me
given that it's not related to string handling at all, afaict
12:05 iSlug left
lizmat maybe ulimit doesn't work on OS X 12:05
moritz wipes clean everything, and rebuilds
lizmat $ ulimit -v 1048576
$ ulimit
unlimited
timotimo that's correct, you'd need to ulimit -a to see the actual limit 12:06
moritz or ulimit -v 12:07
lizmat max memory size (kbytes, -m) 1048576 # so it did work
errands& 12:10
12:14 skids left 12:15 rindolf left 12:17 iSlug joined 12:24 bluescreen10 left 12:38 grondilu joined 12:42 mtk left 12:43 colomon left
masak good afternoon, #perl6 12:44
12:44 rindolf joined
moritz \o masak 12:44
12:49 ribasushi left 12:54 ribasushi joined 12:55 ajr joined 12:56 ajr is now known as Guest69469, Guest69469 is now known as ajr_ 13:07 mtk joined 13:10 PacoAir joined
moritz after removing everything and rebuilding from scratch, I still get that error from S02-types/array.t :/ 13:17
and Any.[0] takes 0.4s (including startup time), Mu.[0] takes 2.2s 13:18
I don't think that it takes 2s to list all the candidates in that error message
13:18 denis_boyun joined 13:19 kay joined
moritz /usr/bin/time ./perl6 -e 'try Mu.[0]' 13:19
1.02user 0.34system 0:01.38elapsed 99%CPU (0avgtext+0avgdata 1499688maxresident)k
0inputs+0outputs (0major+379426minor)pagefaults 0swaps
/usr/bin/time ./perl6 -e 'try Any.[0]'
0.33user 0.08system 0:00.42elapsed 97%CPU (0avgtext+0avgdata 300788maxresident)k
0inputs+0outputs (0major+79697minor)pagefaults 0swaps
13:19 kay is now known as Guest51551
masak .oO( slow as the bellow of a cow ) 13:19
timotimo wow, so much more memory used 13:20
moritz aye
timotimo r: 1499688 / 300788
camelia ( no output )
timotimo r: say 1499688 / 300788
camelia rakudo 61580f: OUTPUT«4.985864␤»
moritz so I'm pretty sure there's something seriously wrong here, but I don't know what it could be, because everything I could think of would fatally die on all setups, not just mine
13:21 kaleem left 13:23 bbkr_ joined
masak moritz: maybe submit a rakudobug? 13:23
the problem sounds palpable enough to be an observable. 13:24
bbkr_ hi, how can I do quotemeta() in P6?
moritz bbkr_: it's the default for regex interpolation
r: my $str = '$str'; say '$str' ~~ /$str/
camelia rakudo 61580f: OUTPUT«｢$str｣␤␤»
moritz which is why there's no special method or sub for it 13:25
13:26 kbaker joined
bbkr_ I need it for qx{}. to pass password param from MAIN safely to system command 13:29
qx:s{something --password=$pass } does not quote meta characters by default 13:30
moritz right. But for that you don't need quotemeta, but shellquote (don't think that exists either, but it's not the same) 13:31
geekosaur shudders at passing passwords on command line
bbkr_ thanks 13:32
GlitchMr run "command", "something", "-password=$pass";
But I don't think it's good idea to pass passwords on command line.
13:32 ribasushi left
moritz GlitchMr: but that doesn't capture standard output 13:33
which qx does
GlitchMr oh, right
Escaping is possible, but probably error prone, and OS specific. 13:34
Replace ' with '\'', and put everything in single quotes.
timotimo quoting is simply ugly :(
GlitchMr Still, Perl 6 needs something like IPC::Run.
bbkr_ indeed, perl6 -e 'run "echo", "a b\"c"' works as expected. I do not need to capture output and password is not sensitive at all in my case so this solution solves my question. thanks again 13:35
moritz GlitchMr: agreed 13:36
GlitchMr Also, it would be fun if code like run("echo", "something") ==> run("some", "other", "command") would just work (the command doesn't have to be run). 13:38
13:38 rindolf left
bbkr_ does run set any status variable like $! in P5? 13:38
GlitchMr Run gives status as return code.
13:38 ribasushi joined
GlitchMr When used as boolean, it's true when return code is 0. When used as number, it returns return code. 13:39
13:39 skids joined
timotimo .o(still b0rked in loops, i believe) 13:40
cannot show it in the evalbot, though
bbkr_ GlitchMr: works, thanks
timotimo r: for ^1000 { 1 but False }
camelia ( no output )
timotimo r: for ^1000 { my $i = 1 but False; 1 }
camelia rakudo 61580f: OUTPUT«(signal ABRT)»
timotimo there we go
masak lovely. 13:42
timotimo p6bench is hit by that :(
GlitchMr I still have no idea why run() likes to fail.
Something is broken with "but Boolean", but I've no idea why. 13:43
masak GlitchMr: um... it's not &run that fails, as evidenced by the above?
GlitchMr Yeah, I know.
timotimo r: class ProcessResult is Int is Bool { has $.rval; method Bool { not $.rval }; method Int { $.rval } }; given ProcessResult.new(:rval(1)) { say +$_; say so $_ }
camelia rakudo 61580f: OUTPUT«===SORRY!===␤Duplicate box_target for native int␤»
GlitchMr I know the problem is with "but Boolean", but, I don't know why it fails.
timotimo whoops?
GlitchMr Something about double free(), but... 13:44
gist.github.com/GlitchMr/6826186 13:46
Not really useful
13:47 btyler joined
GlitchMr "Address 0x8 is not stack'd, malloc'd or (recently) free'd" 13:47
huh?
0x8 doesn't look like an address.
moritz it looks like a NULL-Pointer that's been incremented by 64 bit 13:48
GlitchMr hm, makes sense.
timotimo so trying to access a null pointer with an offset of 1 int 13:49
somewhere in the bigint ops
GlitchMr What does bigint have with "but"?
moritz GlitchMr: maybe it's not the "but", but the boolification protocol 13:50
GlitchMr That would make sense.
I don't know much about Perl 6 internals. 13:51
I wonder how JVM reacts to that.
Nothing, no crashes, no anything. 13:53
But that makes sense, it's JVM, not Parrot. Entire lowlevel codebase is different.
(I was hoping JVM version of NQP would throw an exception) 13:54
13:57 jnap joined 14:05 bruges_ left 14:06 bruges joined 14:08 iSlug left, benabik left 14:10 xenoterracide joined, SmokeMachine joined, kivutar joined 14:12 bruges left, bruges joined 14:13 bluescreen10 joined 14:19 fhelmberger left 14:20 SamuraiJack__ left 14:33 lizmat left, lizmat joined 14:34 bluescreen10 left 14:41 risou_awy_ is now known as risou 14:48 fhelmberger joined 14:49 benabik joined 14:52 hummeleB1 joined, fhelmberger left 14:56 ajr joined, ajr_ left 14:57 ajr is now known as Guest16641 14:58 Guest16641 is now known as ajr_ 15:10 raiph joined 15:16 rindolf joined 15:23 daxim left, dmol left
dalek rlito: ed1912f | (Flavio S. Glock)++ | misc/pretty_print.pl:
Perlito5 - pretty-printer stub
 15:29
moritz abstrusegoose++ # "Big data just means increasing the font size" 15:33
15:33 zakharyas left 15:34 kivutar left 15:39 daxim joined, daxim left 15:41 donaldh joined
donaldh moritz: is it a coincidence that this xkcd.com/ mouseover is the converse? 15:43
I should say xkcd.com/1273/ for future proofing 15:44
If nobody else gets to 100_000 eager ListIter before the end of UTC day, I'll take a look. 15:47
timotimo is currently in "use keyboard as little as possible to get out of RSI mode ASAP" mode 15:48
donaldh FWIW, it looks like MapIter is doing it righter.
TimToady wonders if this is a result of trying to say "well, if if a list of unknown finity has 100_000 elements, maybe we'd better think of it as infinite anyway" 15:51
*finitude, I guess 15:52
15:52 Guest51551 left
TimToady but even if we make such a guess and stop evaluating a mostly-eager list, that doesn't mean we should allocate 100_000 right off the bat 15:53
timotimo agreed, 100_000 seems really excessive
15:53 bbkr__ joined
TimToady unless we plan to always copy it to something shorter when we find the end 15:54
donaldh Well MapIter just maxxes out at 100_000 for the $rpa
So if it is infinite, we don't need infinite memory. 15:55
japhb__ "maxes out"?
donaldh $count = 100_000 if $count > 100_000
won't allocate an array bigger than 100_000 when eager. 15:56
timotimo so ... will only the first 100_000 elements ever be stored?
donaldh dunno about that.
15:56 bbkr left
moritz fwiw the 100_000 is how many elements are reified *at once* 15:57
not how many are reified at all
timotimo ah, ok
donaldh what moritz said
That seems like more sane logic for ListIter too right ? 15:58
TimToady how does ListIter differ?
donaldh When eager, it presizes its $rpa to 100_000 regardless 16:00
Not reify up to 100_000 *at once*
TimToady yeah, I see that 16:03
to me that seems unlikely to be the right balance point for an initial guess 16:04
the 'r' stands for "resizeable" doesn't it? 16:05
donaldh hadn't got around to asking what rpa stands for 16:07
geekosaur resizable parrot-m(something)-code array? 16:08
originally at least
TimToady but that's only for reify(*), so maybe it's the fault of the callers 16:09
(partly) 16:10
arnsholt geekosaur: ResizablePMCArray, where PMC is what Parrot calls an object 16:11
TimToady $!listiter.reify(*) appears to be called only from MapIter
geekosaur yeh, I was trying to expand PMC at the same time 16:12
arnsholt Parrot Magic Cookie initially, later retconned to PolyMorphic Container, IIRC
[Coke] let me know if you fix the pugs issue on camelia, as that is making the daily runs about 20x as painful as they need to be.
geekosaur since replacing one indigestible initialism with another is less than helpful
16:14 dmol joined
TimToady MapIter calls ListIter.reify(*) if count is *, so it's also allocating 100_000 in that case 16:17
1000 might be much saner 16:18
donaldh will experiment with it later. 16:19
TimToady 4 might be even saner than that--see Perl 5
donaldh decommute&
16:19 donaldh left
[Coke] nqp: nqp:say(1,2,3); 16:20
camelia nqp: OUTPUT«Confused at line 2, near "nqp:say(1,"␤current instr.: 'panic' pc 14748 (src/stage2/gen/NQPHLL.pir:5229) (src/stage2/gen/NQPHLL.nqp:279)␤»
TimToady of course, Perl 5 was partly assuming a powers-of-two allocator, so growing the array by 5/4 each time only copies when crossing 2**n boundaries
[Coke] nqp: nqp:deb(1,"hi");
camelia nqp: OUTPUT«Confused at line 2, near "nqp:deb(1,"␤current instr.: 'panic' pc 14748 (src/stage2/gen/NQPHLL.pir:5229) (src/stage2/gen/NQPHLL.nqp:279)␤»
[Coke] nqp: nqp::deb(1,"hi"); #whoops 16:21
camelia nqp: OUTPUT«Error while compiling block : Error while compiling op deb (source text: "nqp::deb(1,\"hi\")"): No registered operation handler for 'deb'␤current instr.: '' pc 55082 (src/stage2/QAST.pir:20123) (src/stage2/QAST.nqp:3316)␤»
[Coke] shame, there was a lot of documentation for that opcode in the old doc file.
japhb__ TimToady: if Perl 5 was assuming a powers-of-two allocator would be used anyway, why bother to only grow the array by 5/4? Why not always grow by 2? 16:22
moritz list vs. hash?
TimToady hashes always grow by 2 16:23
(in p5)
moritz it's much more costly to rehash a hash than to to grow a list
TimToady though p5 does tricks to make that fairly cheap too
it's obvious from the masked hash value which entries to copy and which to leave
since doubling the hash table merely means you're paying attention to one more already-existing bit 16:24
japhb__ No, my point is, what does it actually *gain* to use 5/4 as the array growth ratio? I can see that 5 is relatively prime to 2, but what magical property does that bring?
TimToady we don't actually ever "rehash"
it guarantees growth if you start at 4 :)
and is still quite parsimonious with memory on systems where you're rather copy than overallocate 16:25
japhb__ Okay ... but why not 2 as the growth ratio, if that's the allocation ratio?
TimToady it doesn't try to guess which malloc is in use
japhb__ Were the small-object pools not power-of-two?
Ah!
TimToady I think it was powers of two up to some value 16:26
then it was ad hoc
japhb__ gotcha.
TimToady well, the BSD allocator I stole :)
SysV malloc was stingier
japhb__ The great thing about BSD is how very hard it is to steal from them. :-)
TimToady people on SysV with sufficient memory tended to pick Perl's malloc over the system malloc anyway 16:27
but the other reason p5 could be relatively stingy and biased towards copying was that it did a lot more to remember what happened last time 16:29
calling the same subroutine would tend to reuse the storage from last time
at least, calling the same subroutine at the same recursion level would 16:30
that was another reason to bias in favor of assuming small initial sizes 16:31
one would not really have to keep around the actual data structure like p5 did, just remember how big it got last time in similar situations 16:32
though keeping it around does save the cost of reallocating
might not be practical with the way frames are handled now though 16:33
though if we made frames by cloning a protoframe, the protoframe could probably be tweaked with that info 16:34
slava does perl6 have 'thunks' ? 16:37
TimToady not as first class items yet
we use thunks all the time though
$a || $b <- RHS is a thunk 16:38
has $.foo = stuff; # stuff is a thunk
dalek p: c659ef3 | coke++ | docs/nqp-opcode.txt:
Remove old nqp op list.

We now have a test (t/docs/*) that provides a list of codes from the vms. Other comments in here are design considerations that are now resolved, and docs for an obsolete opcode.
See docs/ops.markdown going forward.
p: 18f113a | coke++ | docs/ops.markdown:
document most exceptional ops
TimToady thunks are mentioned in S's 3, 4, 5, and 6 16:39
[Coke] 3 ops now that are in the docs, but don't exist in parrot: writefh, isfilelink, openasync 16:41
TimToady arguments to macros are thunks, more or less
except you can't call ASTs as code directly, afaik 16:42
r: macro runlater ($a) { macro runnow { $a }; runnow }; runlater(say hi) 16:49
camelia rakudo 61580f: OUTPUT«===SORRY!===␤too few positional arguments: 2 passed, 3 (or more) expected␤»
16:50 colomon joined
TimToady r: macro runlater ($a) { macro runnow() { $a }; runnow() }; runlater(say hi) 16:50
camelia rakudo 61580f: OUTPUT«===SORRY!===␤too few positional arguments: 2 passed, 3 (or more) expected␤»
TimToady hmm
16:51 stevan_ joined 16:55 ajr_ left
TimToady r: macro runlater ($a) { $a }; runlater(say 'hi') 16:55
camelia rakudo 61580f: OUTPUT«hi␤»
slava what about my $a := somecode();
not a thunk?
TimToady that's just calling somecode and binding the results 16:56
the question is, how do you get something called 'somecode' that isn't a block or routine 16:57
slava TimToady: I am not smart enough to answer that :P 16:58
moritz eval!
TimToady wonders if adding more exclamation marks will make it more sensical... 17:00
17:01 colomon left
TimToady how does the compiler turn an AST into a thunk, for || and such? 17:01
and should we spec that? 17:02
or provide a first-class API to thunks?
slava TimToady: I hope those questions are not intended for me ...
TimToady no, they're aimed more at everyone :)
diakopter nqp-m: my int $i := 200_000_000; while --$i { nqp::iscclass(pir::const::CCLASS_ALPHABETIC, "«", 0); } 17:03
camelia nqp-moarvm: OUTPUT«To compile on the MoarVM backend, QAST::VM must have an alternative 'moar' or 'moarop'␤as_mast»
slava ahh, good, then I will just stay quiet and read. :)
moritz TimToady: I guess I just missed some context
TimToady: and IMHO it would be helpful to have some API to turn an AST into a code object
TimToady hands moritz some more excalamation marks 17:04
*excl
moritz preferably with a mechanism to provide the context in which it is compiled
r: !!!
camelia rakudo 61580f: OUTPUT«Stub code executed␤ in block at /tmp/Q7QcyE5TCo:1␤␤»
TimToady well, a macro is the natural place to gather that info, if we can type the argument to Thunk as well as AST 17:05
we'll probably end up with mixed macros that can return any of AST, Thunk, or already-evaluted data
(for each parameter) 17:06
17:07 broquain1 is now known as broquaint
TimToady alternately, always supply ASTs, but make it obvious and easy how to compile and/or run bits of AST 17:14
diakopter nqp-m: sub A($k, $x1, $x2, $x3, $x4, $x5) { if ($k <= 0) { return $x4() + $x5() }; my $B := sub () { $k:=$k-1; A($k, $B, $x1, $x2, $x3, $x4) }; $B() }; sub K($n) { sub () { $n } }; say(A(18, K(1), K(-1), K(-1), K(1), K(0) ))
[Coke] can someone explain why pow_I needs 2 types? (what do each of them do?) looks like the first one is result type...
camelia nqp-moarvm: OUTPUT«-35602␤»
diakopter jnthn: ^ that result is wrong O_O
moritz [Coke]: pow_I with negative power boxes into a Num 17:15
[Coke]: and with a postive power returns Int 17:18
but since nqp doesn't know anything about Int and Num, we have provide those type objects so that nqp knows what to return
[Coke] moritz: danke. 17:19
timotimo diakopter: how come you can just --$i on nqp on moar? 17:21
oh. was that not working a limitation of rakudo only? 17:22
dalek p: e0bc2b7 | coke++ | docs/ops.markdown:
Fixup docs for some _I ops

  moritz++		 17:26
[Coke] .to pmurias - fixed the existing _I opcodes i the docs.
yoleaux [Coke]: I'll pass your message to pmurias.
17:30 FROGGS[mobile] left 17:34 rindolf left 17:36 FROGGS[mobile] joined 17:37 nine_ left 17:38 nine joined 17:39 jnap left
diakopter timotimo: someone (jnthn?) added it to all the NQPs a few months ago 17:42
17:42 nine left 17:43 nine joined 17:44 ajr joined, ajr is now known as Guest42820 17:45 Guest42820 is now known as ajr_
timotimo oh nice 17:54
diakopter nqp-m: sub A($k, $x1, $x2, $x3, $x4, $x5) { return $x4() + $x5() if $k <= 0; my $B := sub () { --$k; return A($k, $B, $x1, $x2, $x3, $x4) }; return $B() }; sub K($n) { sub () { $n } }; say(A(18, K(1), K(-1), K(-1), K(1), K(0) )); 18:01
camelia nqp-moarvm: OUTPUT«-35600␤» 18:02
diakopter jnthn: !!!!! another wrong (and different) result...
18:02 SmokeMachine left
diakopter nqp-m: sub A($k, $x1, $x2, $x3, $x4, $x5) { return $x4() + $x5() if $k <= 0; my $B := sub () { $k := $k - 1; return A($k, $B, $x1, $x2, $x3, $x4) }; return $B() }; sub K($n) { sub () { $n } }; say(A(18, K(1), K(-1), K(-1), K(1), K(0) )); 18:02
camelia nqp-moarvm: OUTPUT«-35601␤» 18:03
diakopter *sigh*
arnsholt That's the man or boy test, isn't it?
diakopter timotimo: well there you go, problem with the --
yep
-35,601 is the correct answer
jnthn: bug: $k-- doesn't do the same thing as $k := $k - 1 18:04
er, --$k
if I had to guess, it does decrement the local, but not the lexical or something 18:05
timotimo .o( my lexical to local optimisation hasn't hit master, right? ) 18:13
.o( of course it hasn't! )
ajr_ If you have a little spare time, and/or care about Unicode, you might wander over to The Register for a look at this: www.theregister.co.uk/2013/10/04/ve...b_unicode/ For all the levity, there's some meat there and in the comments. 18:17
diakopter ajr_: yeah, but Perl 6 has dynamic per-string and/or per-interpreter codepages, with mappings to hopefully fit entirely in 8-bits 18:23
ajr_ Maybe that's a marketing slogan, for some definitions of "market"? :-)* 18:24
diakopter ... fsdo "has" 18:27
bbl
timotimo hehe.
18:28 FROGGS[mobile] left
arnsholt ajr_: Whee! That was a fun rant =D 18:32
ajr_ Verity is usually good for that.
geekosaur tempted to say right but for the wrong reasons 18:34
utf16/ucs2 is kinda the worst of both worlds 18:35
since if you're doing significant CJK you will probably find yourself wishing for UTF32, and if not you probably prefer UTF8 :)
dalek kudo/nom: 01c2e2a | (Elizabeth Mattijsen)++ | src/core/ (6 files):
Initial implementation of .grab for Set/Bag/Mix
 18:36
geekosaur thinks about a string representation that can use utf32 or utf8 as appropriate 18:37
lue hello world o/
timotimo grab is for when you want to change a random element out of the bag?
lue geekosaur: utf16 would be so much better if it worked like utf8
timotimo er, "a bunch of random elements" rather 18:38
geekosaur indeed
lizmat timotimo: yes
timotimo mhm, interesting
lizmat well, not change,
remove from
timotimo there's a "BIND" involved... so i thought it's for changing stuff
but okay
18:38 kbaker left
timotimo ah, i see how it works 18:39
lizmat well, I just called it BIND because .pick and .grab are *very* similar
lue
.oO(We would also get characters past 0x10_FFFF)
18:40 darutoko left
timotimo did the fixes to make the classfile/jast writing faster hit yet? are they in some branch i could try out? 18:44
currently takes 22s for the jast stage and 5.8s for the classfile stage
slava is there any prediction on when moarvm will be distributed with rakudo star? 18:48
grondilu slava: hopefully next month, IIRC 18:50
PerlJam slava: I predict that the 2014-06 release of R* will have moarvm. :-)
PerlJam makes an upper bound, but not a least upper-bound
lue Where is MoarVM right now? Can NQP self-compile on it yet? 18:51
grondilu not yet
lizmat but pretty damn near, is my impression 18:52
grondilu nqp runs on moarvm and passed all nqp tests. Now the team is working on bootstraping
(that is compiling itself, not with the cross-compiler)
dalek kudo/nom: 0a2efdb | (Elizabeth Mattijsen)++ | src/core/Baggy.pm:
For some reason, one must specify $_ here? Otherwise:

No such method 'key' for invocant of type 'Nil'
   in method pick at src/gen/CORE.setting:15268
   in block at t/spec/S02-types/bag.rakudo.parrot:281		 18:54
lue can't wait to replace the norewegian blue with MoarVM for his rakudo-ing
lizmat too
18:55 bluescreen10 joined
lue s/nore/nor/ 18:56
grondilu there hasn't been any commit on MoarVM lately. What's up?
'norwegian blue' ? 18:57
grondilu doesn't get the reference
timotimo lizmat: seems like it would be parsed as a hash rather than a code block if you don't use $_
r: { .key => .value }.WHAT.say
camelia rakudo 61580f: OUTPUT«No such method 'key' for invocant of type 'Nil'␤ in block at /tmp/e67ZvmRSKz:1␤␤»
PerlJam grondilu: I thought it was a parrot.
timotimo r: $_ = (a=>1); { .key => .value }.WHAT.say
camelia rakudo 61580f: OUTPUT«(Hash)␤»
timotimo r: $_ = (a=>1); { $_.key => $_.value }.WHAT.say 18:58
camelia rakudo 61580f: OUTPUT«(Block)␤»
btyler monty python joke: en.wikipedia.org/wiki/Dead_Parrot_sketch
lizmat they stun easily
timotimo: ah, indeed, timotimo++
lue grondilu: see btyler's message a couple lines up. 19:01
btyler (although, really, reading the wiki article about it is a sure way to kill the joke...it's on youtube and worth the few minutes :) 19:03
lue thinks he's using the phrase "norwegian blue" similar to how people use "the scottish play". For some reason he has the silly notion saying "Parrot" is likely to cause trouble. 19:05
19:05 lesshaste joined
lesshaste hi 19:05
what's the main aim of perl 6 ?
grondilu lue: btyler: yup got it 19:06
lesshaste: the same as perl 5, only in a more modern way. As I understand it, anyway. 19:07
lesshaste grondilu, in what way modern ? :) 19:08
I think you just moved my question to the word modern :)
grondilu rather "postmodern", as once explained by TimToady
PerlJam lesshaste: have you seen perl6.org yet?
lesshaste yes
it is short on the "why?"
slava lesshaste: because perl6 is better 19:09
PerlJam lesshaste: Perl5 had some warts; Perl6 was a way to fix them in a backwards incompatible way.
grondilu wall.org/~larry/pm.html
19:10 SamuraiJack__ joined
PerlJam lesshaste: for instance, Perl 6 has pervasive laziness. You can't get that in Perl 5 without some deep changes to the language. 19:10
lesshaste: same thing for the re-imagining of regular expressions as "grammars"
lesshaste oh dear
that makes the author look insane
PerlJam lesshaste: same thing for the object model.
(though p5-mop is an excellent start at giving p5 a real, built-in object model) 19:11
timotimo lesshaste: where have i seen your nickname before? o_O
lesshaste moresped? 19:12
19:13 lesshaste left
timotimo o_O 19:13
geekosaur they were just asking about timing in #haskell earlier today 19:16
19:17 FROGGS joined
geekosaur also that last exchange... "less haste more speed" which I think I first encountered in LOtR (the innkeeper at Bree, I think) 19:18
dalek kudo/nom: 4fd3721 | (Elizabeth Mattijsen)++ | src/core/Setty.pm:
We need Set.grab to return a List, rather than a Parcel
 19:32
19:32 kaare_ joined, SamuraiJack__ left
tipdbmp How can I pretty print something (array, hash, etc.), what's Perl 6's Data::Dumper or even better Data::Printer? 19:36
PerlJam tipdbmp: $thingy.perl 19:37
r: my @a = <foo bar baz>; say @a.perl;
camelia rakudo 0a2efd: OUTPUT«Array.new("foo", "bar", "baz")␤»
tipdbmp Yes, I knew that but I wouldn't call that pretty.
PerlJam r: my @a = <foo bar baz>; say @a.gist;
camelia rakudo 0a2efd: OUTPUT«foo bar baz␤»
PerlJam better?
PerlJam doesn't know what sort of "pretty" you're looking for.
or ... I wouldn't call Data::Dumper pretty either 19:38
19:39 risou is now known as risou_awy
FROGGS o/ 19:39
tipdbmp PerlJam, maybe in indent mode 2 or 3, but what I mean by pretty is to have \n and ' ' indentation, .perl is a single line? 19:40
lizmat TimToady: just wondering, since we have an .Array coercer, shouldn't .list also be .List ?
PerlJam tipdbmp: I think you'll have to write a module to get that version of "pretty" 19:41
19:42 raiph left
tipdbmp Hm, okay =). 19:43
19:45 jnap joined 19:46 japhb_ left 19:47 kaare_ left
benabik Hm. Does { .foo } not imply a $_ in the block? Should it? 19:57
19:57 not_gerd joined 19:58 japhb_ joined
timotimo well, how do you do { .key => .value } if you actually want the outer $_'s .key and the outer $_'s value? 19:58
TimToady S04:1637
synopsebot Link: perlcabal.org/syn/S04.html#line_1637
TimToady you use a temporary in that case
benabik So judging from 0a2efdb, it's a Rakudobug? 19:59
timotimo huh?
TimToady I believe it's been reported, but I could be wrong...
not_gerd tipdbmp: see gist.github.com/gerdr/6831810 for some code that indents .perl output
dalek ast: adcc1c7 | (Elizabeth Mattijsen)++ | S02-types/ (4 files):
Add Set/Bag tests for .grab and .total
 20:01
ast: 57a3771 | (Elizabeth Mattijsen)++ | S02-types/ (4 files):
Fudging for niecza / pugs for .total and .grab tests
TimToady lizmat: you really don't want me to make design decisions while I have a migraine... 20:02
even if it's the relatively painless sort
lizmat hopes the migraine will soon go away, not just because of design decisions 20:03
:-)
timotimo not_gerd: your code doesn't handle smileys in quoted text well, it seems to me 20:04
not_gerd timotimo: what's your input data 20:05
20:08 raiph joined
timotimo sorry, i was just running the code in my head ;) 20:09
tipdbmp not_gerd, thanks, it's something. 20:15
20:17 ssutch left
TimToady lizmat: but I'd say offhand that .list only requires that it return something that can act like a list, not necessarily a List 20:19
woolfy We got a devroom at Fosdem 2014, Saturday, February 1, in Brussels: fosdem.org/2014/news/2013-10-02-ac...-devrooms/ (see also wendyga.wordpress.com/2013/09/23/fo...-speakers/ )
diakopter woolfy: hi! 20:20
woolfy I hope some people here want to come to Brussels and speak about Perl 6 in our Perl-devroom. I really hope Perl 6 will be represented.
We got a devroom!!!
diakopter: hi!
diakopter okay, so no particular group got more than 1 day
TimToady
.oO(why would we want to de-vroom?)
 20:21
woolfy diakopter: indeed. Well, they have a limited number of rooms. They might add some more later on. But I am happy as it is that we got a devroom. 20:22
lue
.oO(I thought we were talking about d eVroom)
woolfy TimToady: because we have too much Vroom
Vroom Vroom Vrooooooooom.....
diakopter woolfy: are you still accepting talk proposals for the Perl broom?
woolfy And since Perl is the solution to everything, Perl will also assist in this difficult matter.
20:23 hummeleB1 left
woolfy diakopter: ehm, I just asked for proposals... :-) 20:23
"(10:19:55 PM) woolfy: I hope some people here want to come to Brussels and speak about Perl 6 in our Perl-devroom. I really hope Perl 6 will be represented."
diakopter oh, it says it was published 9/23
lizmat TimToady: ok, whereas .Array would be a true coercer
diakopter oh, I issed hat essage 20:24
woolfy Oh, my blog... well, yes, that blog was both about the main room (for which the deadline is passed) and our devroom (for which the deadline lays far into the future).
TimToady plus lists are sort of an honorary built-in native type, though not so much as in Perl 5...
and lists are more about data flow, not data structure, in Perl 20:26
lizmat on that note: is there a reason why {} and [] slices return a Parcel, rather than a list ?
TimToady because they're generally more like tuplies of known arity
20:26 stevan_ left
TimToady *tuples 20:27
lizmat so shouldn't .pick / .roll / .grab not return Parcels as well ?
TimToady now you're asking for design again :) 20:28
there should be a smiley for "half my head feels like it's missing"
.)
lue
.oO( \c[RIGHT PARENTHESIS BOTTOM HALF] )
 20:29
lizmat TimToady: sorry :-( I know I'm a pain in any normal kind of situation already :-) 20:32
TimToady on camelia aliases, I suspect at some point we should retire pugs to not have a p alias, and instead have p, j, and m aliases for parrot, jvm, and moar, with r racing all three against each other 20:36
moritz eventually, yes 20:38
TimToady maybe with capitals for nqp
or a modifier that just applies across the board 20:39
just ideas from my right brane, which is happy it's not the side that makes decisions .) 20:40
dalek kudo/nom: c962e51 | (Elizabeth Mattijsen)++ | src/core/ (3 files):
Make .total constant attribute for Set/Bag/Mix
TimToady does it even need to be an attribute? 20:42
if the total method just closes over the role parameter?
well, or over something calculated by the role, if not a parameter 20:43
lizmat is stupid, a state var will do the trick nicely and be cheaper if not used
TimToady I suppose there's not much harm in storing one number with the rest of the data
partly why I was thinking the immutable form was a different role than the mutable 20:44
20:45 BenGoldberg joined 20:46 konundra joined
lizmat submethod total { state $total = [+] self.values } # would also do the trick, and be without burden to the object if not used 20:48
hmmm… perhaps not 20:56
moritz nah, state isn't scoped per instance 20:57
but rather per closure clone 20:58
lizmat indeed :-(
20:58 skids left 21:03 konundra left
lizmat ok leaving that idea for future optimizers :-) 21:04
21:05 denis_boyun left 21:08 donaldh joined
dalek kudo/nom: 9f6f443 | (Elizabeth Mattijsen)++ | src/core/ (2 files):
Burdening all Set/Bag objects with .total calculation, doesn't make sense
 21:10
21:12 ssutch joined 21:18 not_gerd left
donaldh jnthn: does eliminating priorInvocation on JVM require jnthn-fu ? 21:25
21:25 Mouq joined
Mouq #perl6++ 21:25
diakopter donaldh: O 21:27
donaldh: I'd guess if you studied the relevant commits in moarvm, you could figure it out
donaldh diakopter: yes, I'm thinking the same. 21:28
diakopter they're pretty small/few iirc 21:29
donaldh Oh, it was just at the weekend.
Mouq nqp-m: nqp::say('hi') # :D 21:31
camelia nqp-moarvm: OUTPUT«hi␤»
dalek rlito: a204083 | (Flavio S. Glock)++ | misc/pretty_print.pl:
Perlito5 - pretty-printer AST
 21:32
rlito: 913bc28 | (Flavio S. Glock)++ | misc/pretty_print.pl:
Perlito5 - pretty-printer precedence
rlito: e1b541e | (Flavio S. Glock)++ | misc/pretty_print.pl:
Perlito5 - pretty-printer precedence, special cases
donaldh Okay, so closure changes in nqp then codegen changes required to make use of it in backend. 21:36
21:40 bruges left 21:43 bruges joined 21:46 benabik left 21:52 Ben_Goldberg joined 21:53 BenGoldberg left
dalek kudo/nom: be96b97 | (Elizabeth Mattijsen)++ | src/core/Mixy.pm:
Make sure that Mixy allows for negative values when building
 21:58
kudo/nom: 560b416 | (Elizabeth Mattijsen)++ | src/core/Baggy.pm:
Make sure we coerce multiplication factors to Int for Baggy
 22:02
kudo/nom: 4b41c01 | (Elizabeth Mattijsen)++ | src/core/ (4 files):
Make sure Mixy / Baggy coerce correctly to each other
 22:12
22:21 skids joined 22:28 btyler left
TimToady we should probably add an optional key type to those roles at some point 22:31
lizmat hmmm… why not now ? 22:33
TimToady now is a point...oh wait, now it's in the past :)
lizmat well, maybe not today, trying to finish up Mix/MixHash tests before sleep 22:34
diakopter reads MixMaster 22:35
lizmat r: my $m=MixHash.new; $m<d>--; say $m # should we allow decrementing on non-existent elements in a MixHash ? 22:36
camelia rakudo 560b41: OUTPUT«MixHash.new(d(-1))␤»
lizmat I guess it makes sense since we allow negative values 22:37
22:37 REPLeffect left
lizmat only 0 makes elements disappear 22:37
TimToady make sense to me
(design brane is starting to wake up a little)
diakopter nqp: sub A($k, $x1, $x2, $x3, $x4, $x5) { return $x4() + $x5() if $k <= 0; my $B := sub () { $k := $k - 1; return A($k, $B, $x1, $x2, $x3, $x4) }; return $B() }; sub K($n) { sub () { $n } }; say(A(18, K(1), K(-1), K(-1), K(1), K(0) ));
camelia nqp: OUTPUT«maximum recursion depth exceeded␤current instr.: '' pc 262 ((file unknown):60546898) (/tmp/5Sshcl1doW:1)␤»
diakopter mwahaha 22:38
nqp-jvm: sub A($k, $x1, $x2, $x3, $x4, $x5) { return $x4() + $x5() if $k <= 0; my $B := sub () { $k := $k - 1; return A($k, $B, $x1, $x2, $x3, $x4) }; return $B() }; sub K($n) { sub () { $n } }; say(A(18, K(1), K(-1), K(-1), K(1), K(0) ));
22:38 REPLeffect joined
camelia nqp-jvm: OUTPUT«Unhandled exception: java.lang.StackOverflowError␤ in ␤ in ␤ in A␤ in ␤ in A␤ in ␤ in A␤ in ␤ in A␤ in ␤ in A␤ in ␤ in A␤ in ␤ in A␤ in ␤ in A␤ in ␤ in A␤ in ␤ in A␤ in ␤ in A␤ in ␤ in A␤ i…» 22:38
diakopter mwahaha
there's the benefit of moarvm
well, one of them
(yes, I know you can turn off the recursion limit on parrot)
(but that's not how its distributed) ;) 22:39
*it's
donaldh and increase the stack on JVM
diakopter true.
dalek p: bb33397 | (Donald Hunter)++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/CompilationUnit.java:
Avoid allocating zero length arrays in CodeRef.
 22:41
p: 5ef6bcc | (Donald Hunter)++ | src/vm/jvm/runtime/org/perl6/nqp/sixmodel/reprs/VMArray.java:
Presize arrays when deserializing.
p: 5cab6ef | (Donald Hunter)++ | src/vm/jvm/runtime/org/perl6/nqp/sixmodel/reprs/VMArrayInstance.java:
Allow min size of VMArrayInstance to be 4 elems.
diakopter nqp-jvm: sub A($k, $x1, $x2, $x3, $x4, $x5) { return $x4() + $x5() if $k <= 0; my $B := sub () { $k := $k - 1; return A($k, $B, $x1, $x2, $x3, $x4) }; return $B() }; sub K($n) { sub () { $n } }; say(A(12, K(1), K(-1), K(-1), K(1), K(0) )); 22:42
camelia nqp-jvm: OUTPUT«Unhandled exception: java.lang.StackOverflowError␤ in ␤ in ␤ in A␤ in ␤ in A␤ in ␤ in A␤ in ␤ in A␤ in ␤ in A␤ in ␤ in A␤ in ␤ in A␤ in ␤ in A␤ in ␤ in A␤ in ␤ in A␤ in ␤ in A␤ in ␤ in A␤ i…»
diakopter nqp-jvm: sub A($k, $x1, $x2, $x3, $x4, $x5) { return $x4() + $x5() if $k <= 0; my $B := sub () { $k := $k - 1; return A($k, $B, $x1, $x2, $x3, $x4) }; return $B() }; sub K($n) { sub () { $n } }; say(A(8, K(1), K(-1), K(-1), K(1), K(0) )); 22:43
camelia nqp-jvm: OUTPUT«0␤»
diakopter wrong; the correct answer is -10
nqp-m: sub A($k, $x1, $x2, $x3, $x4, $x5) { return $x4() + $x5() if $k <= 0; my $B := sub () { $k := $k - 1; return A($k, $B, $x1, $x2, $x3, $x4) }; return $B() }; sub K($n) { sub () { $n } }; say(A(8, K(1), K(-1), K(-1), K(1), K(0) ));
camelia nqp-moarvm: OUTPUT«-10␤»
diakopter whoa that was fast
nqp-m: sub A($k, $x1, $x2, $x3, $x4, $x5) { return $x4() + $x5() if $k <= 0; my $B := sub () { $k := $k - 1; return A($k, $B, $x1, $x2, $x3, $x4) }; return $B() }; sub K($n) { sub () { $n } }; say(A(8, K(1), K(-1), K(-1), K(1), K(0) ));
camelia nqp-moarvm: OUTPUT«-10␤»
diakopter nqp-m: sub A($k, $x1, $x2, $x3, $x4, $x5) { return $x4() + $x5() if $k <= 0; my $B := sub () { $k := $k - 1; return A($k, $B, $x1, $x2, $x3, $x4) }; return $B() }; sub K($n) { sub () { $n } }; say(A(8, K(1), K(-1), K(-1), K(1), K(0) ));
camelia nqp-moarvm: OUTPUT«-10␤»
diakopter bleh
well ok 22:44
nqp: sub A($k, $x1, $x2, $x3, $x4, $x5) { return $x4() + $x5() if $k <= 0; my $B := sub () { $k := $k - 1; return A($k, $B, $x1, $x2, $x3, $x4) }; return $B() }; sub K($n) { sub () { $n } }; say(A(8, K(1), K(-1), K(-1), K(1), K(0) ));
camelia nqp: OUTPUT«0␤»
diakopter ha; nqp on rakudo gets it wrong too
er.
nqp on parrot.
donaldh :)
diakopter
.oO( you know someone is performance obsessed when the regression suite of the vm included the man-or-boy benchmark from the earliest days... )
 22:45
jnthn diakopter: Oh, I was mostly happy to see it from the point of view of testing closure semantics. 22:46
donaldh: You can probably make quite some headway from the moar patches
diakopter jnthn: see nqp on parrot and jvm get the answer wrong
dalek ast: a33884d | (Elizabeth Mattijsen)++ | S02-types/ (4 files):
Added tests for Mix/MixHash and fixed some Bag/BagHash tests
donaldh jnthn: yeah, needs more time to grok the moar patches than I have this weekend.
jnthn donaldh: OK, no hurry :) 22:47
lizmat that's 300+ more tests
diakopter hopes that leads donaldh to start contributing to moar..
donaldh is happy in jvm land for now.
diakopter lizmat: whoa
lizmat mostly cat-licensing 22:48
diakopter say "We do get here, right?";
jnthn lizmat: Nice! Glad we're adding test coverage :)
ingy sorear: pings 22:51
donaldh can I push directly from the same git branch to master more than once?
ingy sorear: could you drop by freenode #yaml 22:52
donaldh i.e. using git push origin mybranch:master ?
dalek p: 464b5d4 | (Donald Hunter)++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/StaticCodeInfo.java:
Avoid allocating HashMaps when lexical vars are null.
 22:54
donaldh realised it was a kinda silly question 22:55
goodnight o/ 22:57
22:57 donaldh left
dalek kudo/nom: 93c6271 | (Elizabeth Mattijsen)++ | t/spectest.data:
Make sure we run the Mix/MixHash tests as well
 22:58
lizmat and on that thought, I wish everybody a good * 22:59
and get some shuteye&
jnthn 'night, lizmat++
23:01 raiph left 23:02 bbkr__ left 23:06 bbkr__ joined 23:11 raiph joined
dalek rlito: 24620a5 | (Flavio S. Glock)++ | / (6 files):
Perlito5 - perl5 - new pretty-printer backend
 23:11
ingy anyone here use Mono? 23:31
23:34 ajr_ left
TimToady n: say "Hi, Ingy, from mono." 23:34
camelia niecza v24-98-g473bd20: OUTPUT«Hi, Ingy, from mono.␤»
ingy TimToady++ 23:35
I am trying to figure out how to run csharp 1liners from commandline 23:36
it has a repl
TimToady csharp would be more jnthn++'s bailiwick 23:37
ingy well I think it's more mono than c#
thus my pining for sorear 23:38
jnthn Hm, wasn't aware of a repl, but I'm almost always using the MS toolchain rather than Mono.
TimToady sorear++ has been kinda heads down at his job lately 23:39
ingy echo 'Console.WriteLine ("hello");' | csharp
# FTW!
jnthn That works? :)
ingy :D
jnthn Nice :)
ingy it was just a guess but it works 23:40
jnthn afaik, csc always wants the boring boilerplate... :)
ingy I made this: gist.github.com/ingydotnet/6833981 23:41
script for multi-loading yaml
ingy has climbed back into the helm of the yaml ship this week
and pushing forward with YAML2 23:42
one thought is to write a reference y2 implementation in p6
diakopter jnthn: any ideas why that man-or-boy one-liner fails in nqp parrot & jvm? 23:43
ingy there are a few reasons why I want to do that
jnthn diakopter: no
diakopter: No idea if it's closure related or not 23:44
diakopter: If it is, JVM is having changes there soon anyway
diakopter: So can revisit it then.
23:47 raiph left
jnthn time for some rest...'night o/ 23:47
[Coke] nqp-moar: say(3); # where is this coming from? 23:56
nqp-moarvm: say(3); # where is this coming from?
camelia nqp-moarvm: OUTPUT«3␤»
[Coke] it's not in nqp yet... 23:57

Powered by Cro and the Raku® Programming Language.

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

Thank you!