|
github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm Set by AlexDaniel on 12 June 2018. |
|||
|
03:48
MasterDuke left
08:25
robertle joined
08:26
p6bannerbot sets mode: +v robertle
08:59
avar left,
avar joined,
avar left,
avar joined,
p6bannerbot sets mode: +v avar
09:00
p6bannerbot sets mode: +v avar
09:02
zakharyas joined
09:03
domidumont joined,
p6bannerbot sets mode: +v zakharyas
09:04
p6bannerbot sets mode: +v domidumont
09:13
ilmari_ left,
wictory[m] left,
AlexDaniel` left
09:18
GDPR left
09:19
GDPR joined,
p6bannerbot sets mode: +v GDPR
09:25
AlexDaniel` joined,
p6bannerbot sets mode: +v AlexDaniel`
09:51
ilmari_ joined,
wictory[m] joined,
p6bannerbot sets mode: +v ilmari_,
p6bannerbot sets mode: +v wictory[m]
09:59
domidumont left
11:17
Zoffix joined,
p6bannerbot sets mode: +v Zoffix
|
|||
| Zoffix | Perl 6 Advent Calendar 2018 Call for Authors: perl6.party/post/Perl-6-Advent-Cal...or-Authors | 11:21 | |
|
11:34
domidumont joined
11:35
p6bannerbot sets mode: +v domidumont
12:12
Zoffix left
12:13
zakharyas left
13:10
lizmat left
13:24
lizmat joined,
p6bannerbot sets mode: +v lizmat
13:43
zakharyas joined
13:44
p6bannerbot sets mode: +v zakharyas
13:57
dogbert17 joined
13:58
p6bannerbot sets mode: +v dogbert17
|
|||
| dogbert17 | ZOFFLOP: t/spec/MISC/bug-coverage-stress-6.d.t - Unhandled exception in code scheduled on thread 4 - continuationinvoke expects an MVMContinuation | 13:58 | |
| in block at SETTING::src/core/ThreadPoolScheduler.pm6 line 153 in block at SETTING::src/core/ThreadPoolScheduler.pm6 line 242 in block at SETTING::src/core/ThreadPoolScheduler.pm6 line 241 in method run-one at SETTING::src/core/ThreadPoolScheduler.pm6 line 236 in block at SETTING::src/core/ThreadPoolScheduler.pm6 line 278 | 14:00 | ||
|
14:07
domidumont left
14:27
SourceBaby joined
14:28
p6bannerbot sets mode: +v SourceBaby
14:50
AlexDaniel left
15:00
buggable left
15:01
buggable joined,
p6bannerbot sets mode: +v buggable
15:04
brrt joined
15:05
p6bannerbot sets mode: +v brrt
|
|||
| brrt | \o | 15:06 | |
| dogbert17 | o/ | 15:10 | |
| timotimo | oh, huh, we have functions starting in MVM_profile_ and MVM_profiler_ | 15:13 | |
|
15:20
fake_space_whale joined
15:21
p6bannerbot sets mode: +v fake_space_whale
|
|||
| brrt | ohai dogbert17, timotimo | 15:43 | |
| timotimo | hai there | ||
| brrt | I take it the fateful commits with the asyncsocket thing were reverted | ||
| it's a shame, I would've happily debugged that | |||
| timotimo | we can still unrevert it for some debugging joy :) | 15:44 | |
| dogbert17 | it'll probably be reverted ahain before long :-) | 15:53 | |
| *again | |||
| the flapper above is probably not JIT related but could perhaps be something for jnthn | 15:55 | ||
|
16:17
robertle left
16:18
AlexDaniel joined,
p6bannerbot sets mode: +v AlexDaniel
|
|||
| Geth | MoarVM: cd4f98aff4 | (Timo Paulssen)++ | 4 files split "cleared bytes" into managed/unmanaged since the unmanaged bytes amount is sometimes so gigantic, that the profiler accidentally underflows "cleared bytes" into the negative. |
16:31 | |
| timotimo | ^- that was actually an entire entry in my list of deliverables :) | ||
|
16:32
brrt left
19:01
fake_space_whale left
19:05
zakharyas left
19:16
patrickb joined,
p6bannerbot sets mode: +v patrickb
19:59
brrt joined
20:00
p6bannerbot sets mode: +v brrt
|
|||
| nine | I wonder what disadvantages would be there for doubling the interning arity limit? | 20:03 | |
| dogbert17 | looking a bit at the SEGV mentioned by XLiff and SqrtNegInf | 20:04 | |
| with no flags the example (gist.github.com/SqrtNegInf/2b0ccf9...c74ece006) SEGV's but with the JIT off I get | 20:05 | ||
| This type cannot unbox to a native integer: P6opaque, Mu | 20:06 | ||
| in sub crc32 at /home/dogbert/.rakudobrew/moar-master/install/share/perl6/site/sources/B11DB940E55C33C6937C1A1A778565879AE880A0 (String::CRC32) line 51 | |||
| as written in the gist | |||
| the SEGV is like this: #0 0x00007f8ab1ce83a6 in MVM_repr_at_pos_i (tc=0x1b27a50, obj=0x0, idx=-4294967273) at src/6model/reprconv.c:97 | 20:07 | ||
| 97 if (REPR(obj)->ID == MVM_REPR_ID_VMArray) { | |||
| jnthn | nine: Probably a drop-in-the-ocean amount of extra memory use (I think the array of interns is allocated as a fixed size) | 20:12 | |
|
20:13
zakharyas joined
20:14
p6bannerbot sets mode: +v zakharyas
|
|||
| timotimo | i.imgur.com/90X2yX6.gifv - moarperf fun | 20:28 | |
| tadzik | mmm | ||
| lizmat | .oO( wets the appetite ) |
20:36 | |
| timo: yeah, you do need to label your axes :-) | 20:39 | ||
| timotimo: hmmm.. on second thought, I'm not sure I'm following | 20:40 | ||
| timotimo | yeah, i do :) | ||
| i.imgur.com/rtKD4oL.png | 20:41 | ||
| and i just did | |||
| nine | Considering how much of spesh's magic depends on callsite interning, I'd argue for setting the limit not too low artificially | 20:44 | |
| lizmat | timotimo: I would s/Timing/Spent and put the "after prev" values into a new column called "Idle" | 21:05 | |
| and remove the word "spent" on each line | |||
|
21:17
zakharyas left
|
|||
| dogbert17 | looks as if the SEGV I blabbered about an hour ago is caused by github.com/MoarVM/MoarVM/commit/d8...70794a133b | 21:20 | |
| lizmat | timotimo: ^^^ | 21:26 | |
| timotimo | oh? crap? | 21:29 | |
| dogbert17 | it's a large change, could it just be a tiny mistake ... | 21:33 | |
| timotimo | could you put a default: MVM_panic(1, "oh no"); into that switch/case? | 21:54 | |
| though it wouldn't get past that "if" if it didn't take one of the branches | 21:55 | ||
|
22:38
brrt left
|
|||
| dogbert17 | timotimo: the MVM_panic wasn't called | 22:38 | |
|
23:05
ggoebel left
23:52
MasterDuke joined,
p6bannerbot sets mode: +v MasterDuke
23:53
MasterDuke left,
MasterDuke joined,
herbert.freenode.net sets mode: +v MasterDuke,
p6bannerbot sets mode: +v MasterDuke
|
|||
