23:05 arnsholt joined 01:19 arnsholt joined 01:44 lue joined 01:46 jnap joined 02:34 FROGGS joined 04:21 colomon joined 04:26 jnap joined 05:27 jnap joined 06:28 jnap joined 07:26 crab2313 joined 07:29 jnap joined 07:52 FROGGS joined
FROGGS morning 08:12
nwc10 jnthn: setting! 08:17
Stage mast : 543772.499
Stage mbc : 3.022
FROGGS /o\
nwc10 paste.scsys.co.uk/297069 where we are. 08:18
FROGGS The following step can take a long time, please be patient. 08:19
moritz m: say (1144911 + 201539 + 543772) / 3600
camelia rakudo-moar d6a93b: OUTPUT«525.061667␤»
moritz m: say (1144911 + 201539 + 543772) / 3600 / 24 08:20
camelia rakudo-moar d6a93b: OUTPUT«21.877569␤»
08:25 tgt joined 08:29 jnap joined 08:39 V_S_C joined, odc joined
V_S_C When compiling module I get 08:47
When pre-compiling a module, its dependencies must be pre-compiled first.
FROGGS what does that module depend on? 08:48
V_S_C Please pre-compile ...
But I laready pre-compiled the dependency first
*already
I get success with simple modules like Acme::Meow 08:49
FROGGS V_S_C: you are on linux?
V_S_C No, MoarVM+Winx64
FROGGS k
so, can you please paste the use statements of you module, and the output of "dir" of the directory the deps are in? 08:50
V_S_C the modules are from modules.perl.org
FROGGS that does not matter
it is just that one of the deps must be outdated 08:51
V_S_C no, coz I restarted from scratch
FROGGS but error messages usually don't lie 08:52
V_S_C I'll put my writing of Masak++'s ufo
I rewrote it for MoarVM+Winx64 08:53
though I hardcoded some path for my system
so I can get more practice with perl6, than just rebuilding it 08:54
it breaks on modules with dependencies though 08:55
FROGGS: I'll comeback in awhile & give the url in here so others with Winx64 can have hands on 08:56
As it broke on dependencies, I tried proceding manually but the same error got reproduced 08:57
08:57 V_S_C left
hoelzro morning #moarvm 08:59
made my first PR last night! \o/
nwc10 wonders how jnthn is travelling to FOSDEM 09:00
timotimo \o/ 09:05
09:29 tgt joined 09:30 jnap joined
hoelzro is there any *good* docs on libuv? 09:36
s/is/are/ # me no English can good 09:37
FROGGS there is just that handbook, which is more some kind of introduction 09:40
hoelzro yeah, that's what I was afraid of 09:42
hoelzro puts on his wetsuit, gets ready to source dive 09:44
FROGGS I'll get you a hot chocolate when you are back :o)
hoelzro =) 09:49
diakopter hoelzro: I find the source to be quite easy 09:51
of course, I found the p5 source easy
so maybe my skewed is perspective 09:52
hoelzro heh
the source isn't that bad
I just made a stupid mistake because I was following the book examples too closely without thinking about it
should one use malloc/free in Moar code? or does Moar have a custom allocator routine (I'm allocating a uv_process_t) 09:54
diakopter malloc/free is fine
hoelzro very good 09:55
diakopter but you should associate it with something that knows when to free it
hoelzro yeah, I have a uv_close(req, free) in the exit cb 10:02
diakopter: do you know if failing to close a pipe should be considered fatal? 10:03
diakopter I wouldn't think so 10:04
but I don't know all the concerns
hoelzro alright 10:09
there's a JVM test that does nqp::close($pipe); nqp::close($pipe), so I'm leaning towards "doesn't care"
FROGGS hoelzro: almost 10:13
the first attempt should return something true-ish, and the others false 10:14
hoelzro oh, ok
FROGGS (in case close does return anything at all)
hoelzro that's doable
FROGGS r: say nqp::close($*STDERR) 10:15
camelia rakudo-jvm 1d6acc: OUTPUT«(timeout)»
..rakudo-parrot 1d6acc, rakudo-moar 1d6acc: OUTPUT«===SORRY!===␤Error while compiling op close: No registered operation handler for 'close'␤»
FROGGS r: say nqp::closefh($*STDERR)
camelia rakudo-moar 1d6acc: OUTPUT«close filehandle requires an object with REPR MVMOSHandle␤ in block at /tmp/tmpfile:1␤␤»
..rakudo-jvm 1d6acc: OUTPUT«closefh requires an object with the IOHandle REPR␤ in block at /tmp/tmpfile:1␤␤»
..rakudo-parrot 1d6acc: OUTPUT«Dynamic variable $*STDERR not found␤ in method <anon> at gen/parrot/CORE.setting:12232␤ in any at gen/parrot/Metamodel.nqp:2693␤ in any find_method_fallback at gen/parrot/Metamodel.nqp:2681␤ in any find_method at gen/parrot/Metamodel.nqp:948…»
FROGGS r: say nqp::closefh($*ERR)
camelia rakudo-moar 1d6acc: OUTPUT«close filehandle requires an object with REPR MVMOSHandle␤ in block at /tmp/tmpfile:1␤␤» 10:16
..rakudo-jvm 1d6acc: OUTPUT«closefh requires an object with the IOHandle REPR␤ in block at /tmp/tmpfile:1␤␤»
..rakudo-parrot 1d6acc: OUTPUT«True␤»
FROGGS p: say nqp::closefh($*ERR); say nqp::closefh($*ERR)
camelia rakudo-parrot 1d6acc: OUTPUT«True␤True␤»
FROGGS gah!
hoelzro heh
well, I'll worry about return values later
FROGGS sure :o)
hoelzro does Moar have some sort of string builder? I'm trying to impl readallfh for pipes, so I don't know the size in advance 10:17
should I just read chunks and concatenate the strings?
moritz probably read chunks into an array, call nqp::join at the end
FROGGS it could even be that close returns the filehandle it gets... and its truthness says weather it is opened or not
moritz even if it's not fast yet, join() will eventually be fixed to be O(N), not O(N²) 10:18
no idea what the current state is
FROGGS moritz: he is doing that in C
so I guess you look for concat
hoelzro join could also work 10:20
10:25 V_S_C joined
jnthn hoelzro: For readallfh you need to read chunks *of bytes*, and then decode the whole lot of 'em, I'd say. 10:25
V_S_C FROGGS: github.com/iProcess/mufow64 10:26
hoelzro jnthn: that makes sense 10:27
V_S_C as FastCGI shows on modules.perl.org without experimintal S11 support
so I tried it with github.com/dwarring/PDF-Grammar/ also
FROGGS V_S_C: and FastCGI fails to compile?
V_S_C It stopped keeping demanding constants.pm compilation 10:28
I even did that manually 10:29
FROGGS V_S_C: so, just to be clear, which fails to compile?
V_S_C I'll rerun it, gimme 2 mins 10:30
10:31 jnap joined
V_S_C Please pre-compile C:\Rakudo64\languages\perl6\lib/FastCGI/Constants.pm6 10:32
jnthn V_S_C: What are you doing to pre-compile it?
V_S_C perl6 --target=mbc --output=C:\RakuDo64\languages\perl6\lib\FastCGI\Protocol.mbc C:\RakuDo64\languages\perl6\lib\FastCGI\Protocol.pm6 10:33
jnthn: perl6 --target=mbc --output=C:\RakuDo64\languages\perl6\lib\FastCGI\Protocol.mbc C:\RakuDo64\languages\perl6\lib\FastCGI\Protocol.pm6 10:34
FROGGS that does not precompile C:\Rakudo64\languages\perl6\lib/FastCGI/Constants.pm6
jnthn V_S_C: You need the extension .moarvm, not .mbc
FROGGS and the output filename should be moarvm
yeah
V_S_C But I already precompiled Constants.pm6
Oh the output filename must be it 10:35
FROGGS yeah
V_S_C coz I tried pir & pbc after failing with mbc
jnthn It varies by backend
pir on Parrot, jar on JVM, moarvm on MoarVM 10:36
V_S_C I know but after mbc, moarvm never occured to me
I imagined moarvm to be for implementation backend.. 10:37
I knew there's no MIR, & I tried PIR, PBC expecting legacy extensions recycling 10:38
Success! 10:40
FROGGS \o/
V_S_C FROGGS: jnthn: github.com/iProcess/mufow64 10:45
Its dumbed down with hardcoded path, but now I'll work on adding meaning to it. :) 10:46
& I've already corrected it for using moarvm extension. :) 10:47
10:47 V_S_C left 11:09 tgt joined 11:29 tgt joined 11:32 jnap joined 11:58 tgt joined 12:22 tgt joined 12:32 jnap joined 13:15 tgt joined 13:33 jnap joined 13:40 tgt joined 13:49 woolfy1 left 14:04 [Coke] joined 14:14 tgt joined 14:34 jnap joined 15:07 tgt joined 15:08 jnap joined 15:17 tgt joined 16:52 FROGGS[mobile] joined 16:56 FROGGS[mobile]2 joined
timotimo has someone tried compiling and/or running MoarVM + rakudo on a raspberry pi? 17:21
diakopter nick got moarvm, don't know about rakudo 17:23
timotimo with only 100 megabytes of baseline memory usage for a perl6 program, it could be feasible 17:28
moritz moarvm compiles fine on the rpi 17:40
nqp-m doesn't
it gets an "index out of bounds" error somewhere in nqp compilation
timotimo aww :( 17:44
and the performance is probably also not very awesome?
moritz the rpi is rather CPU starved 17:46
[Coke] I went to school at RPI, so this is all very confusing. :) 17:48
timotimo :) 17:49
i'm considering getting an raspi for some unrelated thing, and putting some perl6 on it just seemed natural :P
moritz then you can debug nqp-m on it :-) 17:56
and I'll get a working nqp-m at least :-)
timotimo er ... maybe? i guess?
the trouble now is that i only have one device that can display HDMI and that's my TV
which obnoxiously does overscan
so if i have a terminal on it, i'll miss the first and last 5 lines and the first and last 10 characters 17:57
moritz considers buying a GPS receiver too, and then use that with the rpi 18:00
ssh exists :-)
18:04 FROGGS joined 18:07 benabik joined 18:10 tgt joined
timotimo has to be set up first, though :\ 18:15
well, i guess that's doable
what are you going to use the gps receiver for?
can the raspi work off of a battery?
18:16 arnsholt joined
PerlJam timotimo: www.raspberrypi.org/faqs#powerBatteries :) 18:18
timotimo ah, fair enough 18:20
18:52 arnsholt joined 19:08 arnsholt joined 19:24 arnsholt joined
timotimo motivation will dictate wether i'll spend my evening tuits on $dayjob to make the monday deadline or moarvm to get either heap allocation statistics or however much of little big ints i can build 19:45
19:49 tgt joined 19:51 arnsholt joined 19:59 arnsholt joined 20:11 tgt joined 20:32 arnsholt joined 20:45 arnsholt joined 20:50 arnsholt joined
diakopter does nqp not build master from latest moar? 21:08
I'm getting a bytecode validation error on stage0\nqp.moarvm
21:13 arnsholt joined
diakopter FROGGS: ? 21:15
moritz tries
FROGGS diakopter: I can try in a few minutes 21:16
diakopter actually I'm wrong, the bytecode validation error isn't *on* that file
moritz nqp-m built fine here 21:17
diakopter gist.github.com/diakopter/eebc0aaf991340a65616
nwc10 jnthn: slow thing finished 21:21
$ echo $? 21:22
0
moritz diakopter: that's not nqp master, is it? 21:23
diakopter nwc10: which slow thing was this? build rakudo with gc run every allocation,ish?
nwc10 yes. it's taken 24 days
[Coke] holy crap.
diakopter moritz: it's nqp master
moritz: nope, you're right 21:24
I was in the wrong dir 21:25
*FACESPIKE*
moritz: you put TimToday to rakudo credits 21:26
moritz I could tell because latest nqp/master shouldn't complain about --git-protocol
diakopter also his github name is often larry, I thought
moritz github.com/TimToady/
FROGGS larry is the display name 21:27
[Coke] github.com/larry/
FROGGS like it shows my real name, which is not my github nick :o)
diakopter moritz: you put TimToday to rakudo CREDITS
moritz diakopter: yes, because that's his nickname now, both on IRC and github 21:28
diakopter no it's not
TimToday: you around?
moritz it's not?
did you even look at the link?
FROGGS Today?
diakopter did you? :P 21:29
moritz did you look around here in the channel?
yes, I did
diakopter I was just pointing out a typo
in the commit
moritz oh.
FROGGS *narf*
github.com/rakudo/rakudo/commit/1d...321750c4ad
maybe TimToday is the best friend of Captain Future :P 21:30
diakopter but my other point was that dalek (at one time) used rakudo's CREDITS to correlate git display names to irc names, not github usernames 21:32
(but I don't know whether it still does that)
21:45 arnsholt joined 22:48 FROGGS[mobile] joined 22:54 arnsholt joined