|
github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm Set by AlexDaniel on 12 June 2018. |
|||
|
00:17
brrt left
01:10
MasterDuke left
01:12
MasterDuke joined
01:13
MasterDuke left,
MasterDuke joined
01:47
MasterDuke left
06:14
Kaypie joined
06:15
Kaiepi left
06:32
brrt joined
06:55
brrt left
07:32
patrickb joined
07:36
brrt joined
07:47
zakharyas joined
|
|||
| brrt | \o | 07:52 | |
| nwc10 | o/ | ||
| timotimo | o/ | 08:59 | |
| jnthn | o/ | 09:06 | |
| AlexDani` | \o/ | 09:16 | |
|
09:16
AlexDani` is now known as AlexDaniel,
AlexDaniel left,
AlexDaniel joined
09:40
zakharyas left
09:56
AlexDaniel left,
AlexDaniel joined
10:10
zakharyas joined
10:53
zakharyas left
11:20
brrt left
11:38
Kaypie left,
Kaypie joined
12:04
squashable6 left
12:07
squashable6 joined
12:53
pamplemousse joined
13:26
Kaypie left,
Kaypie joined
13:32
domidumont joined
13:33
domidumont1 joined
13:36
domidumont left
13:42
domidumont joined
13:44
domidumont1 left
14:00
Kaypie is now known as Kaiepi
14:24
sena_kun joined
14:53
pamplemousse left
15:10
domidumont left
|
|||
| japhb | vOv vOv \o\ |o| /o/ vOv vOv | 15:33 | |
|
15:36
ggoebel left
16:06
patrickb left
16:18
brrt joined
17:03
brrt left
|
|||
| Geth | MoarVM/p6o-setup: ddb1f1c24e | (Jonathan Worthington)++ | 5 files Implement object setup at initialize in P6opaque * Allows specification of setup_prototype, which indicates that we should set up the object with the specified value at creation time * Basic code-gen of this when lowering to fastcreate |
17:26 | |
|
17:39
brrt joined
|
|||
| timotimo | finally perl6 admits that prototype-based OO is the best way to go | 17:39 | |
| brrt | I think perl6 was always sort of prototype-based.../ | 17:40 | |
| ? | |||
| I don't actually, really, know | |||
| timotimo | well, we've got support for that in 6model | ||
| jnthn | Maybe prototype wasn't the best word for that :) | 17:41 | |
| timotimo | "template" :P | 17:42 | |
| looking for the right place to put the ".linenumber and .filename on an MVMStaticFrame need a getcodelocation + smart_*ify rather than getattr_o emitted" special case so i don't have to force people to write lineno(sf) or similar in the confprog | 17:45 | ||
| one of the design considerations of the language was "no need to make it too user friendly", though | 17:48 | ||
|
18:20
brrt left
|
|||
| Geth | MoarVM/p6o-setup: 31046b6285 | (Jonathan Worthington)++ | src/6model/reprs/P6opaque.c Don't emit auto-viv ops for setup attributes |
18:21 | |
| jnthn | Before: JIT was successful and compilation took 802us | 18:24 | |
| Bytecode size: 7232 byte | |||
| After: JIT was successful and compilation took 640us | |||
| Bytecode size: 6656 byte | |||
| timotimo | *nice* | ||
| m: say 6656 / 7232 | 18:25 | ||
| camelia | 0.920354 | ||
| jnthn | Hm, a few new spectest failures from that latest one. Interesting. But...enough for today. | 18:35 | |
|
18:41
Woodi left
|
|||
| Kaiepi | are compound literals allowed or not? i was wondering since they're c99 | 19:26 | |
| like void *foo; bar(&(char)(foo)) | 19:27 | ||
| *&(char){foo} | |||
| ugexe | isnt the codebase c89? | 19:50 | |
|
19:53
brrt joined
|
|||
| timotimo | yeah, we're supporting MSVC | 20:00 | |
| it's got most of c89 | |||
|
20:05
brrt left
20:07
brrt joined
|
|||
| dogbert17 | jnthn: it's quite possible that some spectest failures are due to the RAKUDO_FIRST_FLAG problem. I didn't see that fix in your branch. | 20:13 | |
|
20:26
sivoais left
20:27
sivoais joined
20:38
ggoebel joined
|
|||
| jnthn | dogbert17: I didn't see that particular one fail, though :) | 20:58 | |
|
21:16
Kaiepi left
21:17
brrt left
21:26
brrt joined
|
|||
| dogbert17 | this one looks at bit suspicious though, doesn't happen all the time | 21:45 | |
| dogbert@dogbert-VirtualBox ~/repos/rakudo $ ./perl6 -Ilib t/spec/S15-nfg/many-threads.t | |||
| moar: src/6model/sc.c:401: MVM_SC_WB_OBJ: Assertion `!(obj->header.flags & MVM_CF_FORWARDER_VALID)' failed. | 21:46 | ||
| jnthn | dogbert17: On branches, or not? | ||
| dogbert17 | on the branches | 21:47 | |
| jnthn | Ah, phew :) | ||
|
21:53
Kaiepi joined
21:55
Kaiepi left,
Kaiepi joined
21:58
Kaiepi left
22:29
brrt left
22:45
Kaiepi joined
|
|||
| Geth | MoarVM/p6o-setup: 2eda4c1ef6 | (Jonathan Worthington)++ | src/6model/reprs/P6opaque.c Explode prototype clone into instructions The escape analyzer will be able to deal a lot better with these. |
22:54 | |
| timotimo | good morning jnthn :) | ||
| jnthn | :) | ||
| Didn't sleep yet ;) | |||
| Figured I'd get that bit of relatively easy code churned out tonight so I can look at the trickier things tomorrow | |||
| timotimo | i just figured out that if i put a stringification thing into the confprog, that'd naively entail allocation of mvm objects :| | 22:55 | |
| jnthn | d'oh, yes | 22:59 | |
| What do you need to stringify? | |||
| timotimo | integers :) | ||
| but just for log output | |||
| so log can just grow a variant that understands integers, and one that can understand nums | 23:00 | ||
|
23:00
Kaiepi left
|
|||
| timotimo | anyway, confprog can now decide based on a static frame's code location as getcodelocation defines it | 23:01 | |
| i.e. look at the first annotation and take the file name and line number of that | |||
| jnthn | ooh, nice | 23:02 | |
| timotimo | that's probably the only thing of interest for profiling | ||
| apart from "exclude anything from SETTING::" | |||
| Geth | MoarVM/p6o-setup: bc65492ec0 | (Jonathan Worthington)++ | 7 files Add a no-write-barrier form of sp_bind_o Since we can know in various cases we generate that the write barrier will never possibly be hit. |
23:11 | |
| MoarVM/p6o-setup: ed4a1b882c | (Jonathan Worthington)++ | src/6model/reprs/P6opaque.c Use sp_bind_o_nowb in object setup code-gen |
|||
| jnthn | Another easy one :) | ||
| Geth | MoarVM/configurable-subsystems: 8f26551470 | (Timo Paulssen)++ | 4 files confprog: implement lineno and filename for MVMStaticFrame |
23:15 | |
|
23:15
Kaiepi joined
23:16
AlexDaniel left
23:29
Kaiepi left
23:31
Kaiepi joined
|
|||
| timotimo | confprog is speculated to perhaps get support for C strings, fwiw. so stringification of ints and nums isn't off the table | 23:44 | |
| jnthn | Ah, this is telling: with a small nursery and GC debug on I replicate loads of those assertion failures dogbert17 mentioned earlier. But if I disable JIT they go away. | ||
| Certainly I'm too tired to work out what exactly is going on there | |||
| But it's a good clue for tomorrow | 23:45 | ||
| Curiously, I don't think I changed anything JIT-y in my branch, so I've probably uncovered an existing bug | |||
| timotimo | quite probably | 23:46 | |
| will be interesting to see | |||
| i'm pretty glad today was a cold and rainy day | |||
| jnthn | It became rainy here in the afternoon and cooler than it has been | 23:51 | |
| timotimo | nice | ||
| jnthn | Absolutely hammered it down, filled the drains, and the water in my office toilet started trying to escape :P | 23:52 | |
| (It failed, thankfully.) | |||
| timotimo | oh yikes | ||
| jnthn | Plenty of gurgling/splashing noises, but no mess. | ||
| sleep time o/ | 23:53 | ||
| timotimo | gnite jnthn! | 23:54 | |