»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or rakudo:, or /msg camelia p6: ... | irclog: irc.perl6.org or colabti.org/irclogger/irclogger_logs/perl6 | UTF-8 is our friend!
Set by moritz on 22 December 2015.
00:00 ggherdov left
AlexDaniel commitatble6: 2015.07.2 my @a = 'a'..'c'; say lazy @a; 00:00
committable6 AlexDaniel, ¦«2015.07.2»: a b c
AlexDaniel BenGoldberg: the change happened somewhere between 2015.07 and 2015.09, but can't tell when exactly
would've been nice to know because the answer is sometimes in the commit message 00:01
00:02 jackc2 joined
BenGoldberg Also, the Nil vs Any thing is a bit wierd. 00:02
00:02 ggherdov joined
gfldex_ AlexDaniel: Nil may be a dd artifact 00:03
AlexDaniel BenGoldberg: ↑
00:05 bjz joined
BenGoldberg wonders what would happen if whatever is creating those Nil values were to instead return Empty. 00:05
m: say +eager 'a'..'c';
camelia rakudo-moar aaec51: OUTPUT«3␤»
BenGoldberg m: say +lazy 'a'..'c';
camelia rakudo-moar aaec51: OUTPUT«Cannot .elems a lazy list␤ in block <unit> at <tmp> line 1␤␤Actually thrown at:␤ in block <unit> at <tmp> line 1␤␤»
BenGoldberg m: say +eager lazy 'a'..'c'; 00:06
camelia rakudo-moar aaec51: OUTPUT«3␤»
BenGoldberg m: .say for lazy 'a'..'c';
camelia rakudo-moar aaec51: OUTPUT«a␤b␤c␤»
00:06 cdg left 00:10 pierre_ joined 00:15 mcmillhj joined 00:21 bjz left 00:24 mcmillhj left 00:26 bartolin_ joined, Bucciarati_ joined, snowman_ left 00:27 rindolf left 00:28 bartolin left, Bucciarati left 00:36 mcmillhj joined, pierre_ left 00:44 pierre_ joined 00:46 pierre_ left, pierre_ joined 01:00 pierre_ left 01:04 eliasr left 01:05 aborazmeh joined, aborazmeh left, aborazmeh joined 01:08 pierre_ joined 01:12 pierre_ left 01:13 pierre_ joined 01:18 mcmillhj left 01:22 snowman_ joined 01:26 nadim left 01:41 FROGGS_ joined 01:45 FROGGS left, Jesmaybe joined 01:46 ilbot3 left 01:47 ilbot3 joined 02:00 pierre_ left 02:02 stigo joined 02:07 pierre_ joined 02:12 pierre_ left 02:24 Actualeyes joined 02:28 cpage joined, john51 joined 02:30 john51 left 02:32 aborazmeh left 02:33 john51 joined 02:35 john51 left 02:38 john51 joined, pierre_ joined 02:40 john51 left 02:43 john51 joined
seatek oh my god grammars now - one thing keeps leading to another. my head's going to explode. i need to set up a webcam so someone could call an ambulance. like it would matter. oh, my poor monitors! 02:48
these are looking pretty fun and interesting though
ugh. so much to rewrite now, though, if i use them. and then if i do, i'll be addicted, unable to use anything else ever again. 02:49
so good it must be evil 02:50
02:51 noganex_ joined, Jesmaybe left, pierre_ left
seatek i feel like i've had this dream before 02:52
02:54 pierre_ joined 02:59 itaipu joined 03:02 andrzejku_ joined
seatek this ».ast stuff feels like voodoo, like something tacked-on as an extra nifty feature. but this is how the language itself is composed? 03:04
have feeling learning grammars is going to make me learn about lot more in p6 than i can suspect right now 03:05
03:05 mcmillhj joined 03:10 _28_ria left 03:11 cognominal joined 03:12 _28_ria joined 03:13 dj_goku joined 03:16 Jesmaybe joined 03:21 itaipu left 03:22 itaipu joined 03:27 itaipu left 03:28 itaipu joined 03:34 mcmillhj left
jackc2 \HELP 03:38
03:39 jackc2 left 03:41 khw left 03:59 cibs left 04:01 cognominal left, cibs joined 04:02 andrzejku_ left, andrzejku_ joined
BenGoldberg m: dd gather take 42; 04:08
camelia rakudo-moar aaec51: OUTPUT«(42,).Seq␤»
04:12 BenGoldberg left 04:13 pierre_ left 04:16 xyf joined
xyf hello. 04:16
04:17 pierre_ joined 04:23 xyf left 04:25 canopus left 04:29 vendethiel left 04:30 vendethiel joined 04:33 Cabanossi joined 04:34 canopus joined 04:35 brillenfux joined 04:36 itaipu left 04:45 pierre_ left 04:52 wamba joined 05:13 Jesmaybe left 05:16 vendethiel- joined, vendethiel left 05:17 FROGGS_ left 05:29 skids left 05:46 wamba left, ufobat joined 05:52 RabidGravy joined 06:00 zacts left 06:03 zacts joined 06:05 abraxxa joined 06:19 Actualeyes left, neuraload joined 06:22 wamba joined 06:24 firstdayonthejob joined
nine seatek: yes, grammars are just awesome :) 06:27
06:29 RabidGravy left 06:33 firstdayonthejob left 06:38 cpage_ joined 06:41 broquaint joined 06:42 Actualeyes joined, literal_ joined, cpage left, broquain1 left, cpage_ is now known as cpage 06:43 wamba left 06:54 dataangel left 06:57 snowman_ left
dalek line-Perl5: b6bc0fc | niner++ | p5helper.c:
Fix warning due to unset return value in p5_init_callbacks
 06:58
07:02 wamba joined 07:03 bjz joined 07:08 Actualeyes left 07:11 CIAvash joined 07:13 domidumont joined 07:14 darutoko joined 07:16 bjz left 07:17 domidumont left 07:18 domidumont joined 07:24 Actualeyes joined 07:31 bjz joined
lizmat clickbaits p6weekly.wordpress.com/2016/10/24/...ning-star/ 07:34
07:35 girafe left 07:42 hchienjo joined 07:44 dakkar joined 07:45 aries_liuxueyang joined 07:47 hchienjo left 07:50 wamba left 07:55 nadim joined, aries_liuxueyang left 07:57 aries_liuxueyang joined 08:01 jonas3 joined 08:04 seatek left 08:08 pierre__ joined 08:12 dj_goku_ joined, pierre__ left 08:14 FROGGS joined 08:15 dj_goku left 08:16 aries_liuxueyang left 08:19 FROGGS left 08:24 aries_liuxueyang joined 08:27 pierre_ joined 08:28 aries_liuxueyang left
f3ew_ .win 15 08:38
08:41 Bucciarati_ is now known as Bucciarati
moritz lizmat++ 08:46
08:51 wamba joined 08:56 wamba left, bjz_ joined 08:58 bjz left, andrzejku_ left 08:59 andrzejku_ joined 09:05 wamba joined 09:07 andrzejku_ left, andrzejku_ joined 09:08 literal_ is now known as literal 09:09 grondilu joined
grondilu m: say log(8) %% log(2) 09:09
camelia rakudo-moar 84b4c8: OUTPUT«True␤»
grondilu weird, I get false in local 09:10
$ perl6 --version 09:11
This is Rakudo version 2016.10-37-g127b3be built on MoarVM version 2016.10-15-g715e39a
implementing Perl 6.c
m: print $*VM<version>
camelia rakudo-moar 84b4c8: OUTPUT«Type VM does not support associative indexing.␤ in block <unit> at <tmp> line 1␤␤Actually thrown at:␤ in block <unit> at <tmp> line 1␤␤»
grondilu m: print $*VERSION
camelia rakudo-moar 84b4c8: OUTPUT«Dynamic variable $*VERSION not found␤ in block <unit> at <tmp> line 1␤␤Actually thrown at:␤ in block <unit> at <tmp> line 1␤␤»
grondilu m: print $?VERSION
camelia rakudo-moar 84b4c8: OUTPUT«5===SORRY!5=== Error while compiling <tmp>␤Variable '$?VERSION' is not declared. Did you mean 'Version'?␤at <tmp>:1␤------> 3print 7⏏5$?VERSION␤»
grondilu I never ever can remember how to get that
m: say $*PERL 09:12
camelia rakudo-moar 84b4c8: OUTPUT«Perl 6 (6.c)␤»
grondilu m: say $*PERL.^methods
camelia rakudo-moar 84b4c8: OUTPUT«(BUILD VMnames DISTROnames KERNELnames Str gist compiler name auth version signature desc)␤»
grondilu m: say $*PERL.version
camelia rakudo-moar 84b4c8: OUTPUT«v6.c␤»
grondilu m: say $*PERL.compiler
camelia rakudo-moar 84b4c8: OUTPUT«rakudo (2016.10.53.g.84.b.4.c.83)␤»
09:15 rindolf joined
grondilu m: say log(8) % log(2) 09:22
camelia rakudo-moar 84b4c8: OUTPUT«0␤»
grondilu I get 0.693147180559945 in local
m: say log(8) % log(2) - log(2) 09:23
camelia rakudo-moar 84b4c8: OUTPUT«-0.693147180559945␤»
grondilu I get -1.11022302462516e-16
p6: say log(8) %% log(2) 09:24
camelia rakudo-moar 84b4c8: OUTPUT«True␤»
grondilu star: say log(8) %% log(2)
camelia star-m 2016.04: OUTPUT«True␤»
grondilu jvm: say log(8) %% log(2)
j: say log(8) %% log(2)
camelia rakudo-jvm 8d357a: OUTPUT«True␤»
grondilu tries on his Pi 09:25
grondilu gets the correct result there
09:26 aries_liuxueyang joined 09:27 TEttinger left
grondilu m: say map .log %% 2.log ?? '.' !! '!', 1..10; 09:28
camelia rakudo-moar 84b4c8: OUTPUT«No such method 'log' for invocant of type 'Any'␤ in block <unit> at <tmp> line 1␤␤»
grondilu m: say map *.log %% 2.log ?? '.' !! '!', 1..10;
camelia rakudo-moar 84b4c8: OUTPUT«Cannot resolve caller map(Str, Range); none of these signatures match:␤ (&code, + is raw)␤ in block <unit> at <tmp> line 1␤␤»
grondilu m: say join map { .log %% 2.log ?? '.' !! '!' }, 1..10; 09:29
camelia rakudo-moar 84b4c8: OUTPUT«␤»
grondilu m: say join '', map { .log %% 2.log ?? '.' !! '!' }, 1..10;
camelia rakudo-moar 84b4c8: OUTPUT«..!.!!!.!!␤»
grondilu am I missing something here?
m: say join '', map { .log %% 2.log ?? '.' !! '!' }, 1..20;
camelia rakudo-moar 84b4c8: OUTPUT«..!.!!!.!!!!!!!.!!!!␤»
grondilu m: say join '', map { (.log %% 2.log) ?? '.' !! '!' }, 1..20; 09:31
camelia rakudo-moar 84b4c8: OUTPUT«..!.!!!.!!!!!!!.!!!!␤»
grondilu m: say join '', map { (1 +< .msb == $_) ?? '.' !! '!' }, 1..20; 09:32
camelia rakudo-moar 84b4c8: OUTPUT«..!.!!!.!!!!!!!.!!!!␤»
grondilu oh man, I'm so dumb
m: say join '', map { (.log %% 2.log) ?? '.' !! '!' }, 2 <<**<< 1..20; 09:33
camelia rakudo-moar 84b4c8: OUTPUT«.!.!!!.!!!!!!!.!!!!␤»
grondilu m: say join '', map { (1 +< .msb == $_) ?? '.' !! '!' }, 2 <<**<< 1..20;
camelia rakudo-moar 84b4c8: OUTPUT«.!.!!!.!!!!!!!.!!!!␤»
grondilu jeez
m: say 2**3 == 1 +< 3 09:34
camelia rakudo-moar 84b4c8: OUTPUT«True␤»
grondilu m: say 2**6 == 1 +< 6
camelia rakudo-moar 84b4c8: OUTPUT«True␤»
grondilu m: say (2**7).msb
camelia rakudo-moar 84b4c8: OUTPUT«7␤»
grondilu m: say (1 +< .msb == $_) given 2**2 09:35
camelia rakudo-moar 84b4c8: OUTPUT«True␤»
grondilu m: say (1 +< .msb == $_) given 2**3
camelia rakudo-moar 84b4c8: OUTPUT«True␤»
grondilu m: say (1 +< .msb == $_) given 2**4
camelia rakudo-moar 84b4c8: OUTPUT«True␤»
grondilu m: say (1 +< .msb == $_) for 2 <<**<< 1..5 09:36
camelia rakudo-moar 84b4c8: OUTPUT«True␤False␤True␤False␤»
grondilu m: say (1 +< .msb == $_) for 2 <<**<< 2..4
camelia rakudo-moar 84b4c8: OUTPUT«True␤»
grondilu ??
m: say (1 +< .msb == $_) for 2 <<**<< [2..4] 09:37
camelia rakudo-moar 84b4c8: OUTPUT«True␤True␤True␤»
grondilu ????
m: say join '', map { (1 +< .msb == $_) ?? '.' !! '!' }, 2 <<**<< [1..20];
camelia rakudo-moar 84b4c8: OUTPUT«....................␤»
grondilu m: say 2 <<**<< 1..20
camelia rakudo-moar 84b4c8: OUTPUT«2..20␤»
grondilu wat?
* 09:38
m: say join '', map { .log %% 2.log ?? '.' !! '!' }, 2 <<**<< [1..20]; 09:39
camelia rakudo-moar 84b4c8: OUTPUT«....................␤»
09:39 aries_liuxueyang left
grondilu anyway in local I get ..!.!!!.!!!!!!!.!!!! 09:40
09:46 aries_liuxueyang joined 09:48 maybekoo2 joined 09:56 pierre_ left 10:01 pierre_ joined 10:05 pierre_ left 10:07 maybekoo2 left 10:12 pierre_ joined
grondilu m: my num64 $x = log(8); say $x %% log(2) 10:20
camelia rakudo-moar 84b4c8: OUTPUT«True␤»
grondilu m: my num64 $x = log(8); say $x %% my num64 $ = log(2)
camelia rakudo-moar 84b4c8: OUTPUT«True␤»
grondilu gives up
10:22 itaipu joined 10:30 perlawhirl joined
viki grondilu: True is the correct result. ln(8)/ln(2) is 3, so there's no remainder 10:34
10:35 pierre_ left 10:38 labster left
viki I don't know how you get ..!.!!!.!!!!!!!.!!!! locally, considering all of those would be powers of 2 and ln(n)/ln(2) would be log₂(n), giving you the original power and no remainder. Are you sure you're escaping shell characters right, if you're running it in the shell? 10:39
10:40 aries_liuxueyang left
viki m: say 2 <<**<< (1..20) 10:40
camelia rakudo-moar 84b4c8: OUTPUT«(2 4 8 16 32 64 128 256 512 1024 2048 4096 8192 16384 32768 65536 131072 262144 524288 1048576)␤»
10:41 aries_liuxueyang joined, pierre_ joined
moritz viki: well, ln is a floating point operation, with all the caveats attached 10:44
m: say map { .log / 2.log }, 2 X** 1..20
camelia rakudo-moar 84b4c8: OUTPUT«5===SORRY!5=== Error while compiling <tmp>␤Missing comma after block argument to map␤at <tmp>:1␤------> 3say map { .log / 2.log }, 2 X** 1..207⏏5<EOL>␤»
moritz m: say map({ .log / 2.log }, 2 X** 1..20) 10:45
camelia rakudo-moar 84b4c8: OUTPUT«Cannot resolve caller map(Seq); none of these signatures match:␤ (&code, + is raw)␤ in block <unit> at <tmp> line 1␤␤»
moritz too stupid today
m: say map(*.log / 2.log, 2 X** 1..20)
camelia rakudo-moar 84b4c8: OUTPUT«Cannot resolve caller map(Seq); none of these signatures match:␤ (&code, + is raw)␤ in block <unit> at <tmp> line 1␤␤»
viki moritz: then log($n, $b) would be entirely broken. The caveats cancel out
moritz viki: having numerical errors != entirely broken 10:46
10:46 pierre_ left
moritz m: say log(2**$_, 2) for 1..20 10:46
camelia rakudo-moar 84b4c8: OUTPUT«1␤2␤3␤4␤5␤6␤7␤8␤9␤10␤11␤12␤13␤14␤15␤16␤17␤18␤19␤20␤»
moritz m: say log(2**$_) / log(2) for 1..20
camelia rakudo-moar 84b4c8: OUTPUT«1␤2␤3␤4␤5␤6␤7␤8␤9␤10␤11␤12␤13␤14␤15␤16␤17␤18␤19␤20␤»
moritz m: say log(2**$_) / log(2) for 21..50
camelia rakudo-moar 84b4c8: OUTPUT«21␤22␤23␤24␤25␤26␤27␤28␤29␤30␤31␤32␤33␤34␤35␤36␤37␤38␤39␤40␤41␤42␤43␤44␤45␤46␤47␤48␤49␤50␤»
10:47 Actualeyes left
moritz m: say log(2**$_) / log(2) for 51..80 10:48
camelia rakudo-moar 84b4c8: OUTPUT«51␤52␤53␤54␤55␤56␤57␤58␤59␤60␤61␤62␤63␤64␤65␤66␤67␤68␤69␤70␤71␤72␤73␤74␤75␤76␤77␤78␤79␤80␤»
moritz m: say log(2**$_) / log(2) for 100..119
camelia rakudo-moar 84b4c8: OUTPUT«100␤101␤102␤103␤104␤105␤106␤107␤108␤109␤110␤111␤112␤113␤114␤115␤116␤117␤118␤119␤»
viki m: say log(666**$_) / log(666) for 100..119 10:49
camelia rakudo-moar 84b4c8: OUTPUT«100␤101␤102␤103␤104␤105␤106␤107␤108␤109␤Inf␤Inf␤Inf␤Inf␤Inf␤Inf␤Inf␤Inf␤Inf␤Inf␤»
10:49 b^_^d left
moritz m: say log(2**$_) / log(2) for 150, 200, 250, 300 10:49
camelia rakudo-moar 84b4c8: OUTPUT«150␤200␤250␤300␤»
moritz m: say log(2**$_) / log(2) == $_ for 150, 200, 250, 300
camelia rakudo-moar 84b4c8: OUTPUT«True␤True␤False␤True␤»
moritz hah
stringification rounds off small errors
10:49 d^_^b joined
viki I see 10:50
m: say log(τ**$_) / log(τ) for 100..119
camelia rakudo-moar 84b4c8: OUTPUT«100␤101␤102␤103␤104␤105␤106␤107␤108␤109␤110␤111␤112␤113␤114␤115␤116␤117␤118␤119␤»
viki m: say log(τ**$_) / log(τ) == τ for 100..119
camelia rakudo-moar 84b4c8: OUTPUT«False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤»
viki m: say log(τ**$_) / log(τ) %% τ for 100..119
camelia rakudo-moar 84b4c8: OUTPUT«False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤»
viki stands corrected
m: say (log(τ**$_) / log(τ)) %% τ for 100..119 10:51
camelia rakudo-moar 84b4c8: OUTPUT«False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤»
10:51 bjz_ left, bjz joined
viki m: say (log(τ**$_) / log(τ)).round(10) %% τ.round(10) for 100..119 10:52
camelia rakudo-moar 84b4c8: OUTPUT«True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤»
viki m: say (log(τ**$_) / log(τ)).round(1e10) %% τ.round(1e-10) for 100..119
camelia rakudo-moar 84b4c8: OUTPUT«True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤»
viki bah
viki &
10:52 Actualeyes joined 10:57 pierre_ joined
grondilu the weirdest thing to me is that log(8) / log(2) %% 1 is true on my machine 11:10
and yet log(8) %% log(2) isn't 11:11
moritz does %% coerce its arguments? 11:12
m: say &infix:<%%>.candidates 11:13
camelia rakudo-moar 84b4c8: OUTPUT«(sub infix:<%%> () { #`(Sub|53188128) ... } sub infix:<%%> ($) { #`(Sub|53188280) ... } sub infix:<%%> (\a, \b) { #`(Sub|53188432) ... })␤»
moritz m: say 1.001e0 %% 1
camelia rakudo-moar 84b4c8: OUTPUT«False␤»
11:16 pierre_ left
grondilu suspects nqp is to blame 11:17
11:18 pierre_ joined
grondilu m: use nqp; say nqp::p6box_n(nqp::mod_n(nqp::unbox_n(log(8)), nqp::unbox_n(log(2)))) 11:19
camelia rakudo-moar 84b4c8: OUTPUT«0␤»
grondilu I get log(2) locally
m: use nqp; say nqp::p6box_n(nqp::mod_n(nqp::unbox_n(nqp::log(8)), nqp::unbox_n(nqp::log(2))))
camelia rakudo-moar 84b4c8: OUTPUT«===SORRY!===␤No registered operation handler for 'log'␤»
grondilu m: use nqp; say nqp::p6box_n(nqp::mod_n(nqp::unbox_n(nqp::log(8e0)), nqp::unbox_n(nqp::log(2e0)))) 11:20
camelia rakudo-moar 84b4c8: OUTPUT«===SORRY!===␤No registered operation handler for 'log'␤»
grondilu m: use nqp; say nqp::p6box_n(nqp::mod_n(nqp::unbox_n(nqp::log_n(8)), nqp::unbox_n(nqp::log_n(2))))
camelia rakudo-moar 84b4c8: OUTPUT«This type cannot unbox to a native number: P6opaque, Int␤ in block <unit> at <tmp> line 1␤␤»
grondilu m: use nqp; say nqp::p6box_n(nqp::mod_n(nqp::unbox_n(nqp::log_n(8e0)), nqp::unbox_n(nqp::log_n(2e0))))
camelia rakudo-moar 84b4c8: OUTPUT«0␤»
grondilu hum
sorry that's ok
I still get log(2) locally 11:21
$ nqp --version
This is nqp version 2016.10-5-gb7335a9 built on MoarVM version 2016.10-15-g715e39a
is there a way to know which nqp version Perl 6 is using with "use nqp;"? 11:22
like nqp::version() or something? 11:23
timotimo well nqp::backendconfig()<version> is the moar version at least 11:24
grondilu m: use nqp; nqp::backendconfig()<version>
camelia ( no output )
grondilu m: use nqp; say nqp::backendconfig()<version>
camelia rakudo-moar 84b4c8: OUTPUT«2016.10-15-g715e39a␤»
timotimo but really, with "use nqp" you're not actually using nqp
like, you're not using the nqp compiler
you're just asking for nqp::foo ops from the perl6 compiler 11:25
tbrowder ref docs and modules: is this valid in p6: "use Foo 'bar'" where bar is a subroutine in module Foo?
grondilu yeah but I just want to check I'm using the same locally as in the bot here
turns out I do
timotimo then you'll want to know the moar version
tbrowder if so, how should it be exported?
timotimo tbrowder: needs to be '&bar' at the very least
tbrowder ok, didn't think of the sigil, thanks! 11:26
11:27 g4 joined, g4 left, g4 joined 11:44 gfldex_ is now known as gfldex 11:47 pierre_ left
tbrowder timtimo: hm, i cannot find any way to do "use Foo <&bar>;". i can't see that it's possible: proof otherwise would be beneficial! 11:55
11:57 zakharyas joined
tbrowder and the reason it's not may be that a subroutine exported by default in a module is not actually used unless it is explicitly called in a using program--which is a 'good thing' IMHO. 11:58
timotimo it would be kind of weird if your program would use all the subs you're not calling for you after exporting 12:00
"hey, you forgot to use this rm-rf sub, i'll just invoke it for you real quick"
grondilu lol 12:01
tbrowder agreed, but i'm trying to look at the case of the module distributor: what is the best practice for a module with miscellaneous types of subroutines? export all by default? or try to group them with tags? tag each one individually? what if i export sub foo and the user wants his own sub foo, but he wants all the other subs in my module? 12:09
12:11 CIAvash left
tbrowder or at least he wants to use a subset of all... 12:11
viki There's a ticket for use Foo <&bar>; IIRC that only works with require right now 12:16
tbrowder viki: ok, some progress! do you know how &bar is exported in that ticket? 12:18
viki What do you mean "how"?
It's not implemented yet.
And proposed syntax is exactly that; you give it a list of symbols to export
And IIRC ATM that'd clash with EXPORT sub args 12:19
tbrowder well, inside the module itself. is it "sub bar() is export {" or "sub bar() is export(:sometag) {" or ??? 12:20
viki tbrowder: the whole idea is for the user of the module to control what's imported into their namespace.
Right now that's controlled by the module's author and they're free to crap into user's namespace as they wish, which isn't great. 12:21
tbrowder: and I guess all symbols marked OK for export can be asked by user individually 12:22
tbrowder: there's something in the speculation about it. And I'm sure the ticket references that portion
tbrowder ok, so that has been the concern i've tried to express, and I'll add that to the docs. got a ticket number? 12:23
viki nope
12:23 bjz left 12:24 bjz joined
tbrowder "asked" by user meaning "called" by user? 12:24
viki I'd think the mechanism can be implemented by module author right now. Just take symbols via EXPORT and return a Map with them
tbrowder: asked by use Foo <&bar>; 12:25
12:25 ggoebel left
viki tbrowder: basically the way Perl 5 does it. 12:26
tbrowder hm, i can't see that's currently (as least as described in the docs) any different than marking each sub with "export...", but i'll try to test it. 12:27
thanks
viki tbrowder: marking each sub with export would import it on module's use 12:29
12:29 Khisanth joined, pierre_ joined
viki tbrowder: and you end up with problems like this: fpaste.scsys.co.uk/537887 12:31
12:31 rurban joined
viki tbrowder: I did not ask for `say` to be imported. And sure someone might not do this for `say` but it's easy to accidentally mess up some rare built in no one heard of, but your code actually uses. 12:31
And now you have a bug in your code and you don't know why. And it's hard to track it down to one of the dozen of used modules that modified a core built in you were using 12:32
tbrowder well i'm glad the problem is recognized, i have been nervous about use of export in my ecosystem modules. what do you recommend for current practice, something like "export(:ALL)" so at least the user has to knowingly import all? 12:33
dalek c: 5b70244 | (Francis Grizzly Smit)++ | doc/Type/Str.pod6:
fixed sprintf doc missing $a in order of arguments example
 12:34
viki m: SLICE_HUH 42, [], %, %
camelia rakudo-moar 84b4c8: OUTPUT«(signal SEGV)»
dalek c: 763aa8b | (Francis Grizzly Smit)++ | doc/Type/Str.pod6:
fixed sprintf doc missing $a in order of arguments example
 12:35
viki tbrowder: to me that looks like asking the user to do extra typing to get the same behaviour as without it
Don't think I have any recommendations :)
12:35 pierre_ left, brillenfux left
tbrowder okay, thanks, hope a smart person or persons fix this soonish! 12:36
but i could tag EVERY sub with a unique tag to force the user to use them explicitly... 12:37
what would you as a user of my module think? i think i could even tag them with the subroutine name itself! 12:39
viki Depends on the module. And as I've said, you can implement the use Foo <&bar &ber>; with EXPORT sub: github.com/zoffixznet/perl6-SPEC-F...pm6#L7-L12 12:40
well that module almost does it; it doesn't require the &: github.com/zoffixznet/perl6-SPEC-Func#synopsis
My pet peeve in Perl 5 land is module that provide a single function but require you to explicitly import it.... That's where "best practice" trumps common sense. 12:41
viki &
tbrowder but as i read the docs "*@wanted" is the same as marking a sub "export(:wanted)" 12:42
i agree with the single-sub complaint...
ok, in current rakudo land: i distriube a module where each of 1..N subs is marked "is export(:s1)",..."is export(:sN)" 12:45
12:46 harrison_ joined
perlpilot What I want to know is how to explicitly *not* export something that may conflict locally. 12:46
12:47 harrison_ left
tbrowder well, that would be cool, but it seems that's not possible at the moment. maybe you could make that comment on the ticket i found RT #127305 12:48
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=127305
12:54 perlawhirl left
gfldex lizmat: i wrote you a blogpost so you don't have to go and hunt down something unrelated for the next weekly :-> gfldex.wordpress.com/2016/10/25/yo...ou-an-get/ 12:54
lizmat an get ? 12:55
:)
gfldex i didn't proveread that part it seams :)
lizmat nice blog :-)
keep them coming :-)
13:00 TheovdH joined 13:01 zakharyas left
lizmat gfldex: re caching "subset C where $ //= ::('M::C');" 13:01
isn't a bare sigil in that context a state var, so "subset C where $ = ::('M::C')" is maybe enough ? 13:02
gfldex lizmat: i believe so but it's hard to test. Also, I like //= better because it clearly shows my intend. Code golfs are neat, not necessary. :) 13:04
lizmat okidoki 13:05
tbrowder timotimo:, viki: and perlpilot: i added a comment to that ticket 127305, please critique it when you get a chance 13:09
13:10 cdg joined, cdg left, cdg joined
lizmat tbrowder: if you want to refer to a discussion, you can refer to it in a URL 13:10
perlpilot tbrowder: you said what I meant better than I did :) 13:11
lizmat tbrowder: you mean: irclog.perlgeek.de/perl6/2016-10-25#i_13461499 ? 13:12
TheovdH Hi, is there a way to get consecutive members (i.e. named and unnamed subpatterns) from a Match object? Or do I need to create an explicit action for each rule if I want such a list?
lizmat if the match object is in $/, you can use $0, $1 and $<named> or is that not what you mean?
perlpilot TheovdH: if you put them in (), then it's just $/[0], $/[1], $/[2], etc. 13:13
lizmat $/[0,1,2] # as a slice
$/<foo bar baz> # named captures as a slice 13:14
jnthn m: say ('bcd' ~~ /(.)$<x>=.(.)/).caps
camelia rakudo-moar 364e67: OUTPUT«(0 => ｢b｣ x => ｢c｣ 1 => ｢d｣)␤»
13:14 bjz left, bjz_ joined
perlpilot jnthn++ 13:14
tbrowder lizmat: thanks, i'll add that 13:15
lizmat tbrowder: you know how to create such a link ? 13:16
tbrowder um, no...
but i think i can look at the logs and figure it out 13:17
viki tbrowder: the time stamps in the logs are the links to those lines 13:20
moritz TheovdH: .caps might be what you want 13:21
TheovdH: and .chunks if you want non-captured text too 13:22
TheovdH limat: perlpilot: moritz: Thanks. I need to experiment some more to access elements deeper in the Match object. 13:24
moritz TheovdH: if you try to go more than a few levels deep, it might be easier to use action methods that reduce it to a more managable data structure 13:26
13:26 rurban left 13:27 pierre_ joined 13:35 pmurias joined
dalek c: 9eaf716 | coke++ | xt/aspell.t:
add notes on how to make tests pass
 13:35
13:38 aborazmeh joined 13:39 aborazmeh left, aborazmeh joined 13:43 aborazmeh left 13:46 khw joined
TheovdH lizmat: moritz: What's wrong with something like $/<path><in><parse><tree>[3] ? 13:47
lizmat nothing 13:48
conceptually :-)
moritz TheovdH: coupling
DrForr waves at TheovdH .
lizmat moritz: elaborate ?
moritz lizmat: I like to structure my code to be structured in weakly coupled units, where unit A doesn't need to know about the inner workings of unit B 13:49
arnsholt If you ever want to change the grammar, it'll suck
lizmat ah, ok, yes, agree 13:50
arnsholt Since you'll have to change all over the code, rather than localized
moritz lizmat: but if the parser for a table needs to know how each cell type is parsed internally, they are tightly coupled, changes become a PITA
shoving a well-defined data structure into $/.ast aka $/.made is a very natural mini-API that prevents such tight coupling 13:51
13:52 grondilu left
TheovdH moritz: building such an API still requires me to write an action that processes the members consecutively, which is the essence of my problem. 13:54
moritz: I am aware that I am struggling with Perl6 more than with the concept of parsing. 13:55
13:55 colomon joined, bjz_ left, grondilu joined
arnsholt TheovdH: Wait, you want to extract a list of *unrelated* captures? 13:56
moritz TheovdH: you don't need actions for all named regexes, fwiw 13:58
TheovdH arnsholt: consecutive. the members of rule in their original order. 13:59
arnsholt Sure, but not related by anything other than being siblings in the parse tree
TheovdH arnsholt: correct. 14:00
arnsholt That seems like a smell to me. Why do you need to do it that way?
jnthn TheovdH: Do you have a small example of the rule?
If it's something like [<foo> | <bar> | <baz>]* then you may find it neater to factor it as a protoregex 14:01
TheovdH jnthn: silly example: token chars() { <char> $<xyzzy>=':' <char> <char> $<xozzy>='!' }
jnthn Did you mean for xozzy to be xyzzy (e.g. they're overlapping in name)? 14:02
arnsholt If you need to sequence side-effects, that's a pretty clear case for action methods for the subrules invoked
TheovdH jnthn: no 14:03
jnthn OK, just checking
But I don't see why $/.caps in the action method won't give you what's needed 14:04
TheovdH This is probably just my deplorable grasp of Perl 6, but how would I just join the members?
14:04 Upasaka joined
jnthn What, you just want the whole string that this rule matched? 14:04
moritz that's ~$/
DrForr D*mn, if I weren't in a meeting I'd be able to make some suggestions...
arnsholt But *why*? What is it that makes it important to process siblings in order of mention in the rule? 14:05
TheovdH jnthn: Aaah. I think you are right. 14:07
14:07 skids joined 14:08 dmaestro joined
TheovdH jnthn: moritz: Yes. It was my struggle with Perl6 operators. 14:09
tbrowder lizmat and viki: thanks, i see the link in the logs
TheovdH ansholt: I am working on systematic conversion of some input. By default everything is maintained in order. 14:11
dmaestro Is there another channel for rakudo-specific user experience questions, or is this the right place? 14:14
lizmat dmaestro: this is the right place
dmaestro Thank! 14:15
I tried to upgrade to the latest rakudo star on my MacBook, and the downloadable .dmg is for Mac OS X 10.12 - any chance of providing an El Capitan build (10.11), since Sierra is pretty new and I may not upgrade real soon. 14:16
Here's what I get: 14:17
its-MacBook-Pro-DLS:~ douglasschrag$ perl6 -v dyld: lazy symbol binding failed: Symbol not found: _clock_gettime Referenced from: /Applications/Rakudo/lib/libmoar.dylib (which was built for Mac OS X 10.12) Expected in: /usr/lib/libSystem.B.dylib dyld: Symbol not found: _clock_gettime Referenced from: /Applications/Rakudo/lib/libmoar.dylib (which was built for Mac OS X 10.12) Expected in: /usr/lib/libSystem.B.dylib Trace/BPT
viki stmuk_: ping ^
tbrowder 6 ref RT #127305: the IRC thread-start link was added to the ticket
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=127305
14:22 pierre_ left, pierre_ joined
viki dmaestro: stmuk_ is the Rakudo Star release manager, maybe when they're around they can help out. 14:28
dmaestro Thanks, I'll keep looking.
dalek c: 8970224 | (Tom Browder)++ | doc/Language/modules.pod6:
add some more info on module import vs. export
 14:30
14:31 rindolf left 14:36 rindolf joined 14:37 snowman_ joined 14:38 CIAvash joined 14:43 darutoko left 14:51 jonas3 left 15:03 cdg left
Ulti dmaestro: Sierra isn't much of a change FYI, if you update XCode to the latest version under 10.11 you are in for a world of hurt if you dont also upgrade to 10.12 15:04
dmaestro: in fact you have that error
15:04 g4 left
Ulti I couldnt even build Rakudo on 10.11 with the latest XCode because of that linking error 15:05
I gave up fixing it because there are so many other reported issues
15:07 pierre_ left
geekosaur you can download the xcode 7 clt from developer.apple.com downloads area and reinstall them over the xcode 8 tools 15:07
that's the official workaround until xcode 8.1 ships
15:08 domidumont left
Ulti lame 15:14
skids wonders how apple devs pronounce "clt" :-)
Ulti but yeah I have had nothing go wrong just moving to Sierra, I haven't even rebuilt my homebrew stuff
15:17 wamba left
dalek c: 2d88020 | (Tom Browder)++ | doc/Language/modules.pod6:
correct errors after trying actual code

Seems as if module import/export is very flexible, but a little tedious on the export side (which can be easily automated) by the user until rakudo does it for you).		 15:20
15:21 FROGGS[mobile] joined
viki skids: "silt" :) 15:21
15:28 rudolfochrist joined
dmaestro Ulti: Thanks for the info. It's just really disruptive to _require_ a MacOS X upgrade to install Rakudo Star. It's not that hard for me, but I do need to schedule IT assistance on my work MacBook; and I've used XCode and could do what's needed, but I can't see the worth - how many other users might be turned off by that? The best alternative for most users of El Capitan would be just to stick with *2016.07 15:32
But I _want_ the concurrency fixes Jonathon worked so hard on :-) 15:33
viki
.oO( there's always sauce... )
 15:34
As in monthly compiler releases: rakudo.org/downloads/rakudo/
15:34 bsk_ joined
mst dmaestro: we don't require an OS X upgrade 15:35
dmaestro: however, you have upgraded to a version of XCode that will not work without an OS X upgrade
dmaestro: if you downgrade XCode to a working version I'd expect it to build fine
15:36 bsk_ left
dmaestro Well, only to install from the Rakudo* .dmg !! No, I'm not using XCode to install. Anyway, if I have to work at the build level, I'll be doing it on Linux. 15:36
mst hm?
oh, sorry, hmmm
it may be possible to re-rev the dmg with an older version of XCode or whatever
but we're now into "too much mac specific stuff" for me to have a good opinion
so my apologies for the noise 15:37
dmaestro I want it on my work laptop only to facilitate learning, not really a need.
mst: no problem :-)
dalek c: 7eef1f4 | (Tom Browder)++ | doc/Language/modules.pod6:
clarify tag separator comments in the example
15:49 abraxxa left, CIAvash left 15:59 cognominal joined 16:16 wamba joined, rurban joined 16:29 nicq20 joined, canopus left
nicq20 Hello! o/ 16:29
viki \o 16:30
[Coke] dmaestro: if you're willing to install xcode, we can step you through a build from source, which doesn't solve the original problem, but would get you something to play with; also, if you have docker installed, you can run perl6-in-docker. 16:34
16:35 canopus joined 16:39 nicq20 left 16:40 dakkar left 16:49 cdg joined 16:52 wamba left
dmaestro Coke: Thanks for the offer, I'm interested in trying since I do have XCode installed already. 16:56
16:56 Actualeyes left
dmaestro That's if v7.3.1 isn't too out-of-date. 16:58
17:00 Actualeyes joined 17:08 rurban left 17:15 rurban joined 17:16 rurban left
[Coke] I'm on 8.0 on 10.11.6 17:19
17:43 rurban joined, domidumont joined 17:52 rurban left 17:56 cdg left
dalek c: 0ce48fe | (Tom Browder)++ | doc/Language/modules.pod6:
capitalize first word in a sentence
 18:01
dmaestro Coke: Looks like I need to update to get a better 'tar' - so I'll go with 8.0 18:12
18:17 lostinfog joined 18:20 dmaestro left 18:21 dmaestro joined, cibs left 18:22 RabidGravy joined, rudolfochrist left 18:23 cibs joined 18:27 FROGGS joined, FROGGS[mobile] left 18:33 andrzejku joined 18:39 wamba joined
viki mst: what's considered a distro for PSIXDISTS? ( re github.com/perl6/ecosystem/issues/255 ) 18:41
mst: like, what's blowing up? Is it making automatic releases from each git repo commit or what? 18:42
DrForr! 18:43
mst viki: you'd want to ask ranguard about that, I just got involved because he was about to stop trying after your original responses suggested nobody cared :P
viki :( 18:44
mst I think he may be on holiday at the moment though, so what I'd suggest is that you ask questions on the ticket 18:45
and then I'll kick him to have a look when he's got sufficient internet
18:46 labster joined
viki mst: but what's amusing is he proved my point. Here he's added the "mandatory" meta test but never actually ran it so the repo still contains a META error: github.com/drforr/perl6-Perl6-Parser/pull/1 18:46
18:46 setty1 joined 18:48 sufrostico left
mst viki: well, yes, but that point was nothing to do with what we were actually discussing 18:50
viki Considering the ticket is titled "Module list should check modules META.info is correct" I thought it did. 18:51
Or rather, considering the poposition was to make it a requirement that a module that's added to ecosystem has Test::META as prepres, I thought it id. 18:52
DrForr's module has Test::META as prereq, yet has broken META, ergo making Test::META a mandatory requirement for ecosystem membership doesn't solve any problems.
</grumpy>
[Coke] I love the word "poposition", and we have to make that mean something
mst viki: no, the Test::Meta-as-requirement was nothing to do with ranguard or I 18:54
viki: it was when you started going on about that as if ranguard had proposed it that he noped out
gfldex [Coke]: see dict.leo.org/ende/index_de.html#/se...&pos=0
18:55 girafe joined
gfldex so it's basically where you bum is 18:55
mst viki: you inventned a 'proposition to make Test::META a requirement' - that was the problem. nobody ever wanted that, it made no sense, and then you shouted at other people for it making no sense :P
viki: meanwhile, what we *actually* want is a modulelist that is *probably* uploadable
TheovdH arnsholt: as a first exercise I created a routine that takes a Match object and writes a parse tree. Example of a projection where order is relevant. 18:56
viki mst: well, that's what I see from this. RabidGravy said there's Test::META and ranguard says can ecosystem-team make it a requirement to META.list: i.imgur.com/3QcjvPD.png 18:57
18:57 nicq20 joined
nicq20 Hello o/ 18:57
viki \o
mst viki: oh, yes, sorry, RabidGravy suggested it
viki: I'm sorry that the ecosystem team proposed a solutioln the ecosystem team didn't like, but that still isn't ranguard's fault 18:58
viki I'll wait for ranguard to comment on the ticket. I suspect modules.perl6.org dist builder can generate a list of modules that "probably upload"
It already has a JSON validator. Needs a checker that source URL actually has source... not sure what else. 18:59
mst viki: point, basically, is that for us it was us: "here's a problem" you: "ok, here's the solution" us: "ok, could you require the solution then" you: "THAT ISN'T A SOLUTION" us: "wtf?"
viki :}
mst can you see how it reads like that from outside?
viki Yes.
mst anyway, now we've established that we're all idiots, hopefully we can get somewhere :) 19:00
19:02 toolforger1 joined
El_Che in his best Dr Phil's voice: so, mst, when did you discovered you were an idiot? Did it hurt? 19:02
mst I find it's the sort of thing one doesn't so much discover as remember, repeatedly 19:07
[Coke] OW! (oh, right)
breinbaas (IQ is an imaginary number) 19:08
RabidGravy so what data is it the PSIXDIST is trying to consume?
toolforger1 \o/ 19:09
Anybody got an idea how to best wrap one's mind around the rakudo-for-jvm buildchain? 19:10
psch toolforger1: what do you mean? the chain is quite a bit shorter than the one for rakudo-moar :)
toolforger1 Well, I am having trouble understanding the makefile 19:11
19:11 dylanwh_ is now known as dylanwh
dalek href="https://modules.perl6.org:">modules.perl6.org: 64878f9 | (Zoffix Znet)++ | lib/ModulesPerl6/DbBuilder/Dist/Source/GitLab.pm:
Mark GitLab dists as always fresh

We need a proper check for when new commits arrive, but for now, mark each dist fetch as fresh so postprocessor run properly.		 19:12
toolforger1 I find it difficult to combine the information from Configure.pl, makefile.in, and the various scripts that Configure calls
viki toolforger1: what are you trying to accomplish?
toolforger1 Oh... wait... I'm wrangling NQP, not Rakudo 19:13
I want to extract the regex part of NQP so it becomes useful to Java guys
viki ooooooh 19:14
psch toolforger1: oh, that was on the ml recently i think?
toolforger1 (maybe that's foolish)
:-)
Yeah, I talked about it there
Patrick was helpful but obviously code is better than words :-) 19:15
psch toolforger1: i'd start from the other end. as in, try to load a CompUnit containing only QRegex from Java
toolforger1: the EvalServer should give at least a rudimentary how to load CompUnits, although i suspect that QRegex doesn't have an entryQbid
toolforger1 Issue is that the Java class name is some hex thing - looks like an SHA hash or something
so the interface is going to be 100% unstable
psch $ find . -iname 'QRegex*jar' 19:16
./install/share/nqp/lib/QRegex.jar
toolforger1 Oh, it's in a jar already?
psch yeah
might still be you have that same problem you mentioned when loading it though :)
toolforger1 How does the NQP runtime find the right class name? 19:17
psch i think that's bootclassloader shenanigans..? 19:18
toolforger1 Being a Java guy myself, I can answer that with a solid "no"
19:18 girafe left
toolforger1 Class loaders are just another namespace hierarchy, essentially 19:19
psch well, QRegex.jar is a basically a compilation unit for nqp
toolforger1 plus the ability to load stuff from whatever bytestream source you want, be it files, URLs, or zip file elements 19:20
psch so it's, in a way, nqp bytecode wrapped in java bytecode
toolforger1 so I'd need the bytecode interpreter as well, but that's easy to find out
psch yeah, that's how i hacked together a half-backed attempt for perl6-in-a-jar 19:21
toolforger1 These jars consist of a class plus something serialized, NQP bytecode I've been assuming
psch as in, wrap the whole rakudo interpreter into a single jar, make that interpret the precompiled .class that was output by rakudo before... 19:22
Ulti dmaestro: well I always just use rakudobrew personally.. building Rakudo doesnt take very long and that way you can play with the latest perf improvements everyone here delivers weekly
toolforger1 I wanted to try and cut down; a full Perl6 interpreter in a .jar just to get new-style regexes seemed a bit over the top 19:23
19:23 acrussell joined
psch yeah, i'd agree with that 19:23
i'm not sure how much you can get around the concept itself though, considering lots of the regex engine is written in nqp 19:24
as in, you might have to at least ship a (potentially stripped) nqp runtime
toolforger1 Yeah, you'd need as much of the NQP runtime as necessary to run regex
Looking at the NQP docs, there's quite a lot of bytescodes that are definitely unnecessary. I don't know how easy it would be to strip the unneded bytecodes out though 19:25
btw I am seeing a whole slew of QRegex jars
DrForr viki: I'm working on some talk proposals, but /msg me what's going wrong and I'll fix it tomorrow.
viki DrForr: just a trailing comma that isn't valid JSON. I sent a PR 19:26
psch from the rakudo build dir i have 6, 4 of which are build artifacts
or, well, bootstrap
DrForr Thanks, which module?
viki DrForr: github.com/drforr/perl6-Perl6-Parser/pull/2
DrForr Ah. Glad to know people are actually looking at that :) 19:27
viki FWIW, I found it by looking at modules.perl6.org/update.log :}
toolforger1 four in gen/jvm/stage[12], two in src/vm/jvm/stage0, and (NQPP[56])?QRegex.jar in the nqp directory itself
DrForr Done.
viki \o/ 19:28
psch toolforger1: well, QRegex is the base regex language, the prefix should be self-explanatory
DrForr I'll keep that in mind, but gotta get these proposals out the door.
psch toolforger1: so you'd need QRegex and NQPP6Regex i think
toolforger1 Yeah - P5 regexes are already covered via PCRE, it's the P6 ones that I find awesome 19:29
viki :D
toolforger1 Ah... the ones in stage2 are identical to the ones in gen/jvm/stage2 19:30
psch stage0 and stage1 are bootstrap steps
well, i suppose technically stage2 is too :) 19:31
toolforger1 hehe 19:32
Eek. Main class is in default package. 19:33
Best practice is to put it in a package that you own the domain name for, so this should go to org.perl, or org.rakudo, or org.perl6 19:34
Fortunately there are tools that can take a class file and its dependencies, and move everything to a different package 19:35
psch org.perl6.nqp
is where everything else should sit 19:36
toolforger1 Changing the package name would probably be easier :-)
psch well, minus org.objectweb.asm
and other deps too of course vOv
toolforger1 Yeah, the pulled-in libraries should stay where they are
psch i think this is the appropriate moment for "patches welcome", isn't it? :)
toolforger1 Well, I'd need to know where the code is that writes these classes :-) 19:37
psch grepping for ClassWriter instances should narrow it down significantly
it's probably not org.perl6.nqp.sixmodel.P6Opaque though :) 19:38
[Coke] I would recommend rakudo.org if we have to pick one.
toolforger1 perl6.org or rakudo.org is essentially a question of who's responsible for managing the namespace 19:40
psch [Coke]: well, the rakudo-jvm specific java files for the perl6 binary sit in org.perl6.rakudo
namespace that is
[Coke] ok. 19:41
psch s/binary/runtime/
[Coke] dusts off his java for work.
toolforger1 If NQP's usefulness is restricted to Rakudo, then it should be org.perl6.rakudo.nqp
psch it's not 19:42
there's a somewhat working python implementation on nqp
El_Che the parrot dream is alive, it seems
FROGGS *g*
samcv has anybody had issues with vim's perl6 syntax highlighting? i notice it being pretty slow, so for larger files i'm using the 'perl' syntax highlighting instead 19:43
dalek osystem: c98e552 | RabidGravy++ | META.list:
Add Monitor::Monit

See github.com/jonathanstowe/Monitor-Monit
toolforger1 Is there a chance of any other perl6 subproject to be named NQP?
samcv there may be nothing i can do about this, but was wonering your experiences
toolforger1 If no, org.perl6.nqp would be fine
FROGGS org.perl6.nqp is correct, aye
toolforger1 Okay, that goes into the patch 19:44
(assuming I manage to write one)
FROGGS toolforger1++
viki samcv: BTW, you gave me an email address to add to book mailing list... I'm no longer writing one. Contact moritz isntead.
samcv ok thanks for letting me know
dalek href="https://modules.perl6.org:">modules.perl6.org: 960c08c | (Zoffix Znet)++ | lib/ModulesPerl6/DbBuilder/Dist/PostProcessor/METAChecker.pm:
Add METAChecker postprocessor

Add extra checks to validate distributions in the Ecosystem.
Part of the PSIXDISTS indexing issue
  github.com/perl6/ecosystem/issues/255		 19:45
toolforger1 I am seeing ClassWriter references in nqp/runtime, nqp/sixmodel, and nqp/jav2bc - where should I look first?
psch sixmodel is the object model, so that writes nqp and rakudo level classes 19:46
jast2bc is the compilation step after we turn a QAST into a JAST 19:47
well, runtime is only interop, to C and Java respectively 19:48
so, probably jast2bc i think? :)
toolforger1 looking...
"public static JavaClass buildClass()" seems pretty spot on 19:49
goes down to JastCompiler.compileJast 19:50
arnsholt Yeah, jast2bc is what you want 19:51
sixmodel is to generate Java classes backing 6model classes at compile-time 19:52
Not sure what's in runtime off-hand
19:52 itaipu left
psch NativeCallOps and BootJavaInterop 19:52
arnsholt Right, so some bits of NativeCall and JVM interop in there too 19:53
toolforger1 I hope that these aren't necessary for Regex :-)
arnsholt Ah, right. The NativeCall bits are for callbacks
dmaestro Ulti: Thanks, I took the opportunity to try out the basic toolchain and (with [Coke]'s help) was able to build rakudo.
toolforger1 the NativeCall stuff I mean
arnsholt Nope =)
toolforger1 Yeah, I'd leave out callbacks initially; maybe add callbacks into Java later 19:54
dmaestro I learned that upgrading XCode does not automatically install the command-line tools ;-)
19:54 domidumont left
toolforger1 JVMInterop is probably for converting between Perl and Java strings 19:55
arnsholt Not quite. It's for calling Java code from Perl 6 code
19:55 cdg joined
arnsholt Like NativeCall is for C 19:55
psch j: use java::lang::System:from<JavaRuntime>; System.get_out.println("hi") 19:56
camelia rakudo-jvm 8d357a: OUTPUT«hi␤»
toolforger1 Hmm... seems like it is useful for callbacks
19:56 girafe joined
psch ^^^ that's a lot BootJavaInterop and some RakudoJavaInterop 19:56
19:56 canopus left
toolforger1 Most of the work is probably packaging everything so it interoperated with the Java ecosystem 19:57
I.e. upload to Maven Central, use dependencies instead of the 3rdparty directory
psch RakudoJavaInterop mostly matters to not have to write "method/println/(Ljava/lang/String;)V"
oh, and probably some marshalling from rakudo Str to java.lang.String
toolforger1 Yeah, this needs to be somewhere, maybe in a common subpackage 19:58
19:58 canopus joined
toolforger1 That's easy to figure out 19:58
It's the entry points that are hard to find
The call graph inside Java is easy with the right Eclipse tooling 19:59
19:59 itaipu joined
toolforger1 finding the dependencies from NQP bytecode would be hard again 19:59
19:59 ufobat left
toolforger1 Heh. I most definitely want a Maven or Gradle build instead of the 3rdparty stuff. Eclipse will then automagically retrieve the sources from Maven Central with the right tooling. 20:00
(not just Eclipse, every Java-capable IDE can do this) 20:01
arnsholt Yeah, making the JVM build use more Java standard tools is on my list of things I'd like to do
psch that would be nice 20:02
toolforger1 I can definitely lend a hand with this
psch although i'm pretty used to doing it the way i've been doing it... :)
arnsholt But just integrating stuff into the extant Makefile was *so* much easier when the JVM backend landed =)
toolforger1 ;-P
You know how programmers acquire bad karma :-P
DrForr Proposal 1 for OSCON Austin going out...
toolforger1 (not that I'd have done it very differently if I weren't intimately familiar with the toolchains in question) 20:03
arnsholt Yeah, none of us are
Thus the Makefile =)
And it's complicated by the fact that only *part* of the build is in Java-land 20:04
toolforger1 I can contribute that knowledge
arnsholt Excellent!
This is how open source progresses =)
toolforger1 :-)
You add a dependence on the build tool
There are two that I am familiar with, Maven and Gradle 20:05
arnsholt Anyways, I suspect what we'll want is an incantation that builds the Java bits, and then that invocation can be called from the Makefile to build that part of the project
toolforger1 Maven is probably unsuitable, it is extremely strict in how to set up the build, and essentially requires you set up your own Maven repository on a well-known server
arnsholt Yeah, that might be a bit of an obstacle 20:06
The strictness can probably be dealt with though, since we mostly only care about the output JARs
toolforger1 Gradle, you can package a stub that will automatically download the Gradle tool, all declared dependencies, put them into a local cache so they won't get downloaded again (cache is SHA1-secured), and runs the buildscript
Maven's main strictness is that you have a lifecycle with a fixes set of steps 20:07
DrForr From gradle to crave?
toolforger1 *fixed
arnsholt (Another thing I've been wanting to look into is making the compiler JARs fat JARs, so that they can be invoked as just "java -jar nqp.jar @args" rather than the classpath hackery we have today)
toolforger1 so if you want to insert an extra step, you need to contort everything with Maven
psch arnsholt: fwiw, i've ripped bootclasspath out of rakudo
toolforger1 Fatjars are easy with both Gradle and Maven
psch arnsholt: not sure how that'd go with nqp though 20:08
toolforger1 well, sort of - fatjars offer unique challenges
20:08 ufobat joined
toolforger1 i.e. you need a good fatjar tool, then you invoke it fr 20:09
the application integrator simply invokes that tool as a plugin
If you upload the jars to Maven Central, then both Maven and Gradle can download and use the thing
arnsholt Yeah, that's the bits I haven't figured out =) 20:10
toolforger1 so you can leave it to the application integrator whether they want a fatjar, a bunch of class files, a bunch of jars, jars-packages-into-war, or whatever packaging option there is
arnsholt I just know that the current state of affairs annoys me =)
toolforger1 I.e. the good news is that you do not need to worry about fatjars in Java land, you just need to upload to Maven Central 20:11
Maven Central will archive all old versions that are not marked as SNAPSHOT, so callers can use whatever old version they need for build stability
arnsholt toolforger1: Oh, if you want to look at the build process without actually building all of the things, make -n does a dry run 20:12
toolforger1 or if they hunt a bug, they can always reproduce an old build, without having to archive the old jars themselves
arnsholt That should give you an idea of what goes on
toolforger1 Been there, done that, went into MEGO mode :-D
arnsholt Good, good
Anyways, bedtime here for me
toolforger1 I think the best approach would be to use a "gradle build" call inside the makefile 20:13
arnsholt If you prod me during business-ish hours CET tomorrow I'll probably respond without too much delay
In case of more questions
toolforger1 CET time here myself
arnsholt (But psch has more up-to-date info than me. It's been a while since I worked on the JVM code) 20:14
toolforger1 1:30 commute time, so I tend to be on late-ish and not too quick doing stuff (unfortunately)
will be back tomorrow - g'night arnsholt
arnsholt No rush, this is open source after all
psch ehh, my brain's learning objective c currently
arnsholt Just so that you know when you have the best probability of catching me =) 20:15
psch so i'd have to look for most in-depth things
toolforger1 @arnsholt ok, that's what I meant to communicate, too :-)
Good luck with ObjC, psch... you have my sympathy
psch well, at least it's a job i guess :) 20:16
toolforger1 Heh. I am currently doing Appway.
20:17 TEttinger joined
toolforger1 "Full-fleged BPMN tool", but not really enterprise-ready - does not integrate with any sane SCM because sources are saved as SHA1-identified XML 20:17
I.e. XML files that have SHA1 hashes as filename
psch why does corp like xml so much..?
toolforger1 It's standard. 20:18
psch right. it also does everything i suppose
so it always works!
toolforger1 An extensible one - if you sh*tty customer sends you XML with nonstandard fields, you can easily identify and ignore them
psch anyway, gonna get bedwards too, CET waits for no one o/
toolforger1 \o/
20:19 wamba left
toolforger1 I'd still prefer YAML, but ah well I'm not going to be asked 20:19
off to bed for me, too
g'night all!
20:20 toolforger1 left 20:22 andrzejku__ joined 20:24 Brock is now known as awwaiid
dalek line-Python: 981960c | (Brock Wilcox)++ | / (3 files):
Add an option to EVAL to differentiate eval vs file mode. Python!!!
 20:25
line-Python: 5bb1d89 | (Brock Wilcox)++ | / (5 files):
Merge remote-tracking branch 'origin/master' into fix-EVAL

Explicitly call python2 in installation
Some environments have python point to python3. In these environments the installation fails.
20:25 dalek left, dalek joined, ChanServ sets mode: +v dalek
line-Python: 82c41f0 | niner++ | / (2 files):
Merge branch 'explicit-path' of github.com/parabolize/Inline-Python
20:32 ufobat left 20:33 itaipu left 20:35 nicq20 left
masak how likely is `python2` to be available on a system? 20:35
it's on mine, but I had never heard of it before :)
pmurias psch: what do you mean by nqp bytecode? 20:37
samcv i have python2 on my system, and python is python 3
@masak
depends on the distro though
geekosaur www.python.org/dev/peps/pep-0394/
so I expect it will be fairly reliable
samcv usually doing python2 is more catchall than 'python' 20:38
i'm pretty sure even distros where 'python' uses python2 still have a symlink called python2 pointing to it
20:39 bjz joined, rurban joined
grondilu has 'python2' on his debian/stretch 20:39
dalek line-Python: cf7956e | (Brock Wilcox)++ | README.md:
Update README with EVAL example and fixed travis link
 20:42
line-Python: 0660c00 | niner++ | README.md:
Merge pull request #13 from awwaiid/master

Update README with EVAL example and fixed travis link
masak today's autopun spotting: twitter.com/nixcraft/status/662040529362784256 20:54
timotimo "trick or syncookie" <3
20:57 Zoffix joined
Zoffix 20:57
vendethiel- timotimo: but cookies are for the dark side :o
timotimo ohai ven
masak vendethiel-! \o/ 20:58
vendethiel- o/
20:59 inra joined 21:01 andrzejk1 joined, inra left 21:04 skids left
grondilu oh that's interesting. Remember how I got False in local with log(8) %% log(2) ? I've tried compiling MoarVM with clang instead of the default gcc, turns out with that version I get true. 21:09
so maybe gcc is to blame
I'm using gcc (Debian 5.4.1-3) 5.4.1 20161019
Zoffix grondilu, is floating point noise supposed to be deterministic? 21:10
21:10 RabidGravy left
vendethiel- Zoffix: yeah, hardware specific 21:11
maybe UB
Zoffix Ah 21:12
21:13 kurahaupo__ joined
timotimo something might have shuffled the order of floating point calculations around 21:14
grondilu: i'd be interested to know if the result changes when you execute the same code enough that it gets jitted
though i'm not sure if we have jit-implementations of much of the floating point arithmetic
addition and subtraction for sure, but maybe not logarithms 21:15
21:15 bjz_ joined, bjz left
Zoffix m: log(8) %% log(2) or die for ^1000_0000 21:15
camelia ( no output )
colomon why would you expect log(8) %% log(2) to work?
Zoffix ?
What does "to work" mean in this context? 21:16
21:16 rindolf left
colomon it’s an int/rat operator and you’re passing it floats? 21:16
vendethiel- grondilu: I'm interested in knowing if it changes depending on gcc/clang's optimization level :P
Zoffix Yeah, that's the issue. We get different results for same input 21:17
On different builds of the same versions of rakudo/moar
colomon stops asking questions and pulls up the source 21:18
Zoffix :D
perlpilot colomon: maybe the expected it result is to complain :) 21:19
s/it//
21:19 mcmillhj joined, firstdayonthejob joined
Zoffix m: say gather { take $_ unless log(2**$_) %% log(2) for ^100 } 21:20
camelia rakudo-moar 4829f1: OUTPUT«(33 35 37 39 51 55 59 66 70 74 78 81 85 89 95)␤»
colomon hmm, there’s an explicit infix:<%> for Num
grondilu I'm using debian/stretch on a VirtualBox VM, so it should be easily reproducable.
colomon … and you’ve got to dive into NQP to figure out what it does 21:21
off the top of my head, I’m thinking infix:<%> makes sense for Nums, but infix:<%%> (which is defined in terms of infix:<%>) does not 21:22
Zoffix I get same results as camelia on two different jessie boxes, on Bodhi box based off ubuntu trusty, and a Windows 10 box 21:23
colomon because it amounts to checking a floating point operation for exact equality.
Zoffix with Bodhi running in Virtual Box
colomon m: for 1..10 -> $n { say log(2**$n) } 21:25
camelia rakudo-moar 4829f1: OUTPUT«0.693147180559945␤1.38629436111989␤2.07944154167984␤2.77258872223978␤3.46573590279973␤4.15888308335967␤4.85203026391962␤5.54517744447956␤6.23832462503951␤6.93147180559945␤»
colomon m: for 1..10 -> $n { say log(2**$n) / log(2) }
camelia rakudo-moar 4829f1: OUTPUT«1␤2␤3␤4␤5␤6␤7␤8␤9␤10␤»
colomon m: for 1..10 -> $n { say log(2**$n) / log(2) - $n }
camelia rakudo-moar 4829f1: OUTPUT«0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤»
colomon hmm 21:26
masak m: for 1..10 -> $n { say log(2**$n) / log(2) == $n }
camelia rakudo-moar 4829f1: OUTPUT«True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤»
masak hmmmmm :)
masak .oO( coincidence? I think not! )
colomon I’m inclined to think so 21:27
masak colomon: I'm torn on the infix:<%%> for Nums thing 21:28
colomon: on the one hand, I would never even think to do that
colomon: on the other hand, maybe it would make sense for some Nums; for example Nums that are sums of powers of 1/2 21:29
colomon m: for 1..10 -> $n { say log(3**$n) / log(3) }
camelia rakudo-moar 4829f1: OUTPUT«1␤2␤3␤4␤5␤6␤7␤8␤9␤10␤»
masak I think what you're seeing for those really small numbers is simply double precision being more than enough to end up on the integer precisely 21:30
perlpilot thinks there should be a different operator for %% on Num
masak m: say log(3 ** 100) / log(3)
camelia rakudo-moar 4829f1: OUTPUT«100␤»
masak perlpilot: why? it's doing the same numeric operation
colomon masak: isn’t log(2) an irrational number?
masak perlpilot: a lot of operations on Num are inexact by nature 21:31
perlpilot masak: I don't have a good rationale. It just feels righter to me.
masak colomon: yes, but that likely all gets lost in the fuzz and froth of doubles
21:31 bjz_ left
masak perlpilot: I think I only approve of an explosion of operators when the operators are actually spiritually different. Perl typically ends up doing a rather good job of using *the same* operator for things that are conceptually the same thing. 21:32
perlpilot masak: but, perhaps I'm still reeling from being surprised that it worked in the first place, so it still feels foreign to me.
21:33 Zoffix left
masak heh 21:33
colomon masak: but it’s really weird for the fuzz and froth of doubles to just cancel out and leave you nice integer answers.
masak well, I do agree that it working in the first place was probably to some extent lucky happenstance
colomon: yes, I can't fully explain that. but I don't mind much, either
m: say .1e0 + .2e0 == .3e0 21:34
camelia rakudo-moar 4829f1: OUTPUT«False␤»
masak there, better now? :P
perlpilot colomon: Maybe it's only really weird for those of us who were brought up learning the foibles of floating point ... maybe it's "natural" if you ignore the fact that computers are involved :) 21:35
masak objection. I was brought up learning the foibles of floating point. 21:36
I'm aware of the dangers of loss of precision. I would expect someone using infix:<%%> on Num to be similarly aware.
that's why I said it might still make sense if you limit yourself to Nums which are sums of powers of 1/2. 21:37
21:38 colomon_ joined
masak m: say 0.5e0 %% 0.125e0 # I would expect this to always be True, no matter how crappy the Num precision 21:38
camelia rakudo-moar 4829f1: OUTPUT«True␤»
21:39 colomon left, colomon_ is now known as colomon
colomon but it seems ill-advised to allow an operation that is only well-defined for a tiny percentage of the possible arguments to it. 21:44
21:45 pmurias left
grondilu FYI it would never have occurred to me to use %% with floating points, but TimToady used it on RosettaCode. 21:47
(not that I want to point fingers or anything :) )
colomon is guessing he wanted to know if any integer was a power of two 21:50
m: for 1..200 -> $n { say log(2**$n) / log(2) == $n } 21:52
camelia rakudo-moar 4829f1: OUTPUT«True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤False␤True␤False␤True␤True␤True␤True␤True␤Tr…»
kurahaupo__ m: say 0.1 %% 0.01
camelia rakudo-moar 4829f1: OUTPUT«True␤»
colomon yeah, bad plan
21:53 inra joined 21:54 espadrine joined 21:56 acrussell left, colomon left, colomon joined 22:05 rurban left
grondilu colomon: it was for the Fractran example in RC, but possibly some places else as well. 22:10
I think testing for power-of-two-ness should rather be 2**.msb == $_
22:14 inra left
grondilu or maybe .base(2).comb.one == 1 22:14
22:15 TheovdH left
grondilu m: say join "", map { .base(2).comb.one == 1 ?? '.' !! '!' }, 1 .. 10; 22:15
camelia rakudo-moar 4829f1: OUTPUT«..!.!!!.!!␤»
grondilu ouch 22:16
22:16 colomon left, colomon joined
grondilu m: say join "", map { .base(2).comb.one == 1 ?? '.' !! '!' }, 2 <<**<< [1 .. 10]; 22:17
camelia rakudo-moar 4829f1: OUTPUT«..........␤»
22:18 setty1 left
colomon m: sub is-power-of-two(Int $n) { 2 ** $n.msb == $n }; for ^200 -> $n { say is-power-of-two(2 ** $n); } 22:19
camelia rakudo-moar 4829f1: OUTPUT«True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True…»
22:20 bjz joined
colomon m: sub is-power-of-two(Int $n) { 2 ** $n.msb == $n }; (^200).map(-> $n { say is-power-of-two(2 ** $n) }).categorize-list.say 22:22
camelia rakudo-moar 4829f1: OUTPUT«No such method 'categorize-list' for invocant of type 'Seq'␤ in block <unit> at <tmp> line 1␤␤»
colomon m: sub is-power-of-two(Int $n) { 2 ** $n.msb == $n }; (^200).map(-> $n { say is-power-of-two(2 ** $n) }).List.categorize-list.say
camelia rakudo-moar 4829f1: OUTPUT«No such method 'categorize-list' for invocant of type 'List'␤ in block <unit> at <tmp> line 1␤␤»
colomon needs to finish making noms rather than sort out what stupid thing he is doing wrong there 22:23
22:23 dmaestro left 22:24 espadrine is now known as espadrine_, espadrine_ is now known as espadrine`, espadrine` is now known as espadrine 22:28 andrzejku__ left, andrzejku left, andrzejk1 left 22:31 BenGoldberg joined
grondilu m: sub is-power-of-two(Int $n) { 2**$n.msb == $n }; (^200).map({ is-power-of-two(2**$_)}).bag 22:33
camelia rakudo-moar 4829f1: OUTPUT«No such method 'bag' for invocant of type 'Seq'␤ in block <unit> at <tmp> line 1␤␤»
grondilu m: sub is-power-of-two(Int $n) { 2**$n.msb == $n }; (^200).map({ is-power-of-two(2**$_)}).Bag
camelia ( no output )
grondilu m: sub is-power-of-two(Int $n) { 2**$n.msb == $n }; (^200).map({ is-power-of-two(2**$_)}).Bag.say
camelia rakudo-moar 4829f1: OUTPUT«bag(True(200))␤»
22:34 bjz left 22:37 grondilu left 22:40 espadrine left 22:42 bjz joined 22:45 cibs left
dalek href="https://modules.perl6.org:">modules.perl6.org: 8f32095 | (Zoffix Znet)++ | lib/ModulesPerl6/DbBuilder/Dist/PostProcessor/METAChecker.pm:
Make METAChecker check source URL only for fresh dists
 22:46
22:46 cibs joined 22:47 bjz left
viki m: for 77..79 -> $n { say log(2**$n) / log(2) == $n } 22:48
camelia rakudo-moar 4829f1: OUTPUT«True␤False␤True␤»
viki masak: ^ yeah, it's a coincidence
s: &infix:<%%>, \(2e0, 2e0) 22:49
SourceBaby viki, Sauce is at github.com/rakudo/rakudo/blob/4829...ic.pm#L220
viki m: dd 2e0.Real
camelia rakudo-moar 4829f1: OUTPUT«2e0␤»
viki m: dd 2e0.narrow
camelia rakudo-moar 4829f1: OUTPUT«2␤»
22:51 skids joined, itaipu joined 22:52 mcmillhj left, firstdayonthejob left
viki .categorize-list is only available on Baggies and Maps 23:00
colomon viki++
23:03 sufrostico joined
viki m: sub is-power-of-two(Int $n) { (2 ** ($n.msb//0)) == $n }; (% .categorize-list: {is-power-of-two $_}, ^2000){True}.say 23:06
camelia rakudo-moar 4829f1: OUTPUT«[1 2 4 8 16 32 64 128 256 512 1024]␤»
viki m: sub is-power-of-two(Int $n) { (2 ** ($n.msb//0)) == $n }; ^2000 .categorize(*.&is-power-of-two){True}.say 23:07
camelia rakudo-moar 4829f1: OUTPUT«5===SORRY!5=== Error while compiling <tmp>␤Unexpected block in infix position (missing statement control word before the expression?)␤at <tmp>:1␤------> 3}; ^2000 .categorize(*.&is-power-of-two)7⏏5{True}.say␤ expecting any of:␤ …»
viki your mother
m: sub is-power-of-two(Int $n) { (2 ** ($n.msb//0)) == $n }; (^2000).categorize(*.&is-power-of-two){True}.say
camelia rakudo-moar 4829f1: OUTPUT«[1 2 4 8 16 32 64 128 256 512 1024]␤»
23:09 itaipu left 23:10 lostinfog left
viki m: ^2000 .grep({$_ == 2 ** (.msb//0)}).say 23:12
camelia rakudo-moar 4829f1: OUTPUT«(1 2 4 8 16 32 64 128 256 512 1024)␤»
23:24 vendethiel- left, kurahaupo__ left 23:25 cognominal left 23:52 geraud left

Powered by Cro and the Raku® Programming Language.

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

Thank you!