github.com/moarvm/moarvm | IRC logs at irclog.perlgeek.de/moarvm/today
Set by moderator on 1 September 2013.
timotimo expects to be blown away by the performance improvements over parakudo 00:06
tadzik nqp/moar seems slightly slower than nqp/parrot now 00:11
I wonder if I'm on optimized build on either :) 00:12
diakopter parrot's optimized by default; moar is not right now 00:14
tadzik trying moar with --omg-optimized 00:15
argh. I need a thinkpad key 'S'
diakopter timotimo: it's not going to be hugely faster 00:16
tadzik on optimized build it seems about 20% faster than parrot 00:18
timotimo that sounds good :)
tadzik on nqp tests at lest
which may or may not mean somethin g:)
diakopter how much faster to run all the nqp test than nqp-jvm? :) 00:19
tadzik haha
diakopter tadzik: how are you still awake? holiday tomorrow like us usaians? 00:20
tadzik diakopter: nah, I now work tue-thu, and have a 4-day weekend 00:27
for writing my bachelor's, in theory
in practice I mostly play piano and write games
timotimo oh, games! 00:28
tadzik wanna try my asteroids? :) 00:31
I decided to be a programmer because I wanted to write games (I was 7-ish). Now I realized that I probably wrote more toy compilers than toy games 00:32
diakopter wait, compilers aren't games? 00:33
timotimo imagine perl 6 as a free to play
with microtransactions
want to buy a metaop? :D 00:34
tadzik hah. Free to play, pay to win ;)
. o O ( open source in a nutshell? )
01:06 FROGGS_ joined 01:07 foo_bar_baz joined 01:25 foo_bar_baz joined
dalek arVM: a186fca | jimmy++ | src/core/interp.c:
implemented compunitcodes op
 01:53
arVM: d92067d | jimmy++ | src/ (5 files):
moved ContainerRegistry to instance
 02:07
arVM: f07af83 | jimmy++ | / (7 files):
implemented getcodecuid op
 03:42
TimToady src/core/interp.c: In function ‘MVM_interp_run’: 04:34
src/core/interp.c:1229:21: error: duplicate case value
src/core/interp.c:1207:21: error: previously used here
case MVM_OP_compunitcodes: ended up in there twice 04:48
04:56 cognominal joined
dalek arVM: 5933e8f | jimmy++ | src/core/interp.c:
added missing getcodecuid op
 05:01
05:54 jlaire joined
jnthn JimmyZ: In a186fca3fa, maybe_cu and cu must be rooted, as they could become invalid after the MVM_repr_alloc_init call. But the simpler fix is just to move the line that allocates to be the first thing the op does 05:54
05:56 FROGGS joined
JimmyZ what? jnthn, you're getting up too early 05:56
jnthn JimmyZ: Yeah...gotta commute in 10 mins or so
JimmyZ heh
jnthn Thought I'd do some code-review while I slurp my coffee and my hair dries... :) 05:57
FROGGS *g*
morning @al
@all
jnthn
.oO( weird @all )
FROGGS and he's here in the room :o)
JimmyZ FROGGS, you're getting up too early also? 05:58
:P
FROGGS just a bit :o)
that is basically the time I want to be at work
jnthn JimmyZ: Does 56-role.t pass with these updates? 05:59
JimmyZ jnthn: nope
Cannot locate an outer frame for the call
same as 44 test 06:00
jnthn ok 06:02
I won't start looking at that now, otherwise I'll forget to take the train to $dayjob :)
JimmyZ :P
dalek arVM: 5328f7a | jimmy++ | src/core/interp.c:
avoids a MVMROOT
 06:04
JimmyZ jnthn: ^^ right?
jnthn JimmyZ: yes 06:05
FROGGS humm, we fail more tests than yesterday O.o 06:06
this is new: t/nqp/44-try-catch.t .................. Could not locate compile-time value for symbol return
ahh no, we have more test files 06:07
t/nqp/56-role.t ....................... Cannot locate an outer frame for the call
at nqp-src/NQP.nqp:251 (nqp.moarvm:ȷ{��*:137) <--- that $!name is weird
jnthn commute & 06:10
JimmyZ \\
dalek arVM: d28c899 | (Tobias Leich)++ | nqp-cc/src/QASTOperationsMAST.nqp:
import more constants from jvm-backend
 06:26
JimmyZ FROGGS: looks like 77 test is different from nqp 06:36
FROGGS true
JimmyZ I'd like to copy 77 test to 19-file-ops.t in NQP
FROGGS yes, but it needs at least fudging for vm/parrot and/or jvm 06:37
JimmyZ oh
FROGGS or wait, there is a new file-io test in nqp since a few days, maybe you just take this one 06:38
(committed by donaldh)
JimmyZ I avoid to commit NQP 06:41
FROGGS JimmyZ: you dont have to, but you can steal the new test file from nqp 06:43
JimmyZ we don't need to steal the new test... nqp-cc will go away 06:45
FROGGS sure 06:46
07:29 odc joined
dalek arVM: 7f0dad7 | jimmy++ | src/core/interp.c:
removed a bunch of MVMROOT
 08:54
09:00 donaldh joined
dalek arVM: b4aaa4a | jimmy++ | src/core/interp.c:
removed another pieces of MVMROOT
 09:07
09:56 foo_bar_baz joined
dalek arVM: f186138 | (Tobias Leich)++ | src/io/fileops.c:
bring nqp::readlinefh in line with jvm/parrot

That is, read and return trailing newlines.		 10:23
JimmyZ FROGGS: didn't you break t/nqp/77-file-io.t? 11:24
jnthn JimmyZ: 7f0dad7 was a net clarity win, but b4aaa4a makes things worse... 11:25
If we can re-order things so we naturally only access things once, go for it, but re-fetching stuff we need to refer to multiple times, then masking the smell with comments, is a bad idea. 11:26
FROGGS JimmyZ: that is because the tests are not in line with nqp@parrot and nqp@jvm 11:31
JimmyZ FROGGS: so it should contain whatever it reads CRLF? 11:33
FROGGS JimmyZ: yes, that is what nqp@parrot/jvm do, even if I'm not fond of it :o) 11:34
jnthn I fixed at least the slurp test to cope with the fact you might have \\r\\n 11:35
Dunno about the other tests though
FROGGS and btw, 19 would pass if CREDITS file were there 11:36
jnthn Just hack the test to look for a differnt file? Or throw a dummy CREDITS file in the repo? :) 11:38
FROGGS jnthn: I can copy it over...
JimmyZ jnthn: gist.github.com/zhuomingliang/6411968 11:39
jnthn: Is it OK? 11:40
since we fetch tc->instance->hll_syms only twice, 'always fetch it' is better than 'fetch it and then MVMROOT' 11:42
jnthn JimmyZ: I don't think it is...]
We fetch hll_name twice in here too 11:43
JimmyZ oh, yes, you prefer two MVMROOT?
dalek arVM: 8389856 | (Tobias Leich)++ | nqp-cc/t/nqp/ (2 files):
use CREDITS file from nqp for testing
 11:44
JimmyZ I don't like two MVMROOT though
jnthn I think it was clearer before... I'm fine if we get rid of MVMROOT for things we only fetch once just by re-ordering code, like the patch for the bigint ops
But fetching the same thing twice just to avoid using MVMROOT feels icky and will make refactoring harder. 11:45
JimmyZ jnthn: so keep it as before
is it OK?
jnthn Yeah. 7f0dad7 I like, but b4aaa4a I'd rather just revert.
JimmyZ well, I don't like b4aaa4a after pushing... 11:46
jnthn That's ok, version control exists to help us cope with not realizing we don't like things until later :) 11:47
JimmyZ just hate pushing and reverting it
*and then
jnthn yes, it's nice to realize stuff before pushing. But we're all human :)
dalek arVM: 9d614e0 | jimmy++ | src/core/interp.c:
Revert "removed another pieces of MVMROOT"

This reverts commit b4aaa4a913b83b96261d52f4509c2ddacf38ac54.		 11:49
FROGGS .oO( except jnthn *cough* )
dalek arVM: 549e22c | jimmy++ | src/ (2 files):
implemented rename op, and fixed a potential read line bug
 12:00
diakopter g'morn 12:01
FROGGS: yes, we already passed 3 or 4 of the 7 I copied 12:02
lizmat morn diakopter
JimmyZ diakopter: 03:00am there?
diakopter o/ :)
5 a.m.
JimmyZ you're working behave as honeybee? 12:03
diakopter somehow caught food poisoning 1-2 days ago; slept all day yesterday when I wasn't otherwise miserable
stinks to be sick on a paid holiday ("labor" day) 12:05
at some point yesterday I hallucinated that I was working on (and finished!) a Perl 6 implementation written in the modeled world (2d dot grid) of this game Sugar Crush Saga... 12:07
woolfy Let your dreams come true as soon as you get better (I wish you get better soon). 12:10
diakopter thanks. hee-hee :)
woolfy oO(Sugar Crush, that's a game, hmm, never saw it, just read about it) 12:11
diakopter it's very cute. reminds me of the movie Wreck It Ralph, which is one of the best movies *ever* (it was released this year)... I *highly* recommend watching it. very amusing and touching. 12:13
woolfy oO(wha? huh?) 12:14
www.imdb.com/title/tt1772341/ (has a rating of 7.8 which does not indicate it is one of the best movies ever)(is a movie from 2012, which is not this year) 12:15
diakopter, go back to bed! :-) 12:16
JimmyZ could anyone help me download www.youtube.com/watch?v=TIUUVEojULE ? 12:18
lizmat woolfy: go back to bed, you're not feeling well either! 12:19
JimmyZ I nerver can visit youtube in china
woolfy ok ok bye
moritz JimmyZ: moritz.faui2k3.org/tmp/jnthn-moarvm.webm 12:21
12:21 donaldh joined
JimmyZ moritz: thanks very much 12:22
jnthn
.oO( wow, really effective firewall there, China :P )
 12:47
diakopter: Get well soon.
timotimo no firewall can handle aggressively adaptive protocols (like humans) 12:58
13:05 Woodi joined
Woodi hallo :) 13:05
lizmat Woodi o/
dalek arVM: fd6faee | (Tobias Leich)++ | nqp-cc/t/nqp/ (4 files):
merge 77-file-io.t into 19-file-ops.t
 13:12
13:36 FROGGS joined 14:28 donaldh joined 14:48 FROGGS joined
diakopter woolfy: it's a good movie :P 16:15
woolfy diakopter: the description suggests so, I just never heard of it. I will add it to my want-list 16:16
Liz is a bit stunned: how did diakopter know that woolfy came out of bed... :-) 16:18
diakopter just woke up, again 16:23
I have a feeling I'll wake up again in another 4 hours 16:24
woolfy too 16:28
17:24 FROGGS joined
diakopter might be getting hungry, finally 18:22
FROGGS that's a good sign 18:23
timotimo i hope you'll recover soon! 18:28
18:44 jnthn joined
diakopter jnthn: the 24-module.t failure relates to the prior_invocation thingie 18:49
jnthn diakopter: Yeah. prior_invocation must die, though. 18:52
diakopter what was it bandaiding? 18:53
I can try to fix it if you tell me how
jnthn diakopter: It was before I talked with Pm at YAPC::NA about closures :) 18:59
I think it's one I need to work out; I'm not completely sure I can provide instructions that'll work, and also I want to fix this stuff on JVM too.
diakopter oh 19:00
masak .oO( priori invocati ) 19:21
FROGGS .oO( whispered masak with a flick of his wand ) 19:22
sorear
.oO( masak is porting Perligata? )
 19:23
19:50 BabsSeed joined
jnthn hm 20:29
Non-zero exit status: 5
On 49-regex-interpolation.t
We haz a segfaulter...
benabik 5 == segfault?
jnthn yeah
(Note, Windows :)) 20:30
benabik shudders.
;-)
jnthn Urgh. Memory corruption. 20:33
timotimo damn politicians everywhere 20:34
jnthn Anything with a thread ID of 13 in a single-threaded program is le hosed
timotimo identity crisis?
jnthn It's st is null too 20:35
*Its
0x0000000000b12b20
uh, sorry
Hm, yes, object pointer is into fromspace... 20:38
That's always wrong.
timotimo fromspace is the "take from here" space of the GC? 20:39
jnthn It's the semi-space we were allocating in before the last GC. 20:40
And that we evacuated all living things from during the last GC
sorear haha 13 20:49
jnthn unlucky, huh? :P 20:50
sorear nm was confusing 13 (SIGPIPE) with 11 (SIGSEGV)
dalek arVM: fabda86 | jnthn++ | src/gc/debug.h:
Add header with handy GC debugging macro.

Have hacked this into code a couple of times while debugging GC issues and figured it's best shared and easily available.		 20:56
jnthn static MVMObject *env_hash; 21:00
if (!env_hash) {
argh!!
timotimo oops :)
jnthn And not only is it static
It never gets permrooted.
Meaning it gets GC'd.
So some time later, env_hash points into heck knows what.
jnthn is learning that the "is it in fromspace" sanity test is one of the unforseen debugging advantages of having chosen a semi-space collector for the nursery... 21:02
dalek arVM: 94ade18 | jnthn++ | src/ (3 files):
Hang env_hash off instance, mark it.

Fixes a segfault that 49-regex-interpolation.t triggered, but really anything unlucky enough could have.		 21:10
jnthn Well, there's my patch for today 21:14
lizmat is considering zapping tests that do chop(@array) and chop(%hash) 21:34
tests originate from 2008, aka 5+ years ago, and seem to have no bearing on reality or the future
t/spec/S32-str/chop.t for the ones interested 21:35
oops ww 21:36
diakopter hi 22:56
23:04 cognominal joined 23:13 benabik joined 23:24 foo_bar_baz joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!