|
github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm Set by AlexDaniel on 12 June 2018. |
|||
|
00:01
lizmat joined,
p6bannerbot sets mode: +v lizmat
02:06
Kaiepi left
02:07
Kaiepi joined,
p6bannerbot sets mode: +v Kaiepi
02:34
saigel_ joined
02:38
saigel_ left
02:51
fake_space_whale left
03:13
kurbasUB joined,
kurbasUB left
03:14
grim_radical joined,
grim_radical left
04:03
Zoffix joined,
p6bannerbot sets mode: +v Zoffix
|
|||
| Zoffix | New blog post: "A Request to Larry Wall to Create a Language Name Alias for Perl 6": perl6.party/post/A-Request-to-Larr...for-Perl-6 | 04:33 | |
|
04:46
Zoffix left
05:03
kc2uez joined,
p6bannerbot sets mode: +v kc2uez
05:05
kc2uez left
05:47
TimToady left
06:01
dwellerqS joined,
dwellerqS left
06:32
TimToady joined,
p6bannerbot sets mode: +v TimToady
06:49
domidumont joined
06:50
p6bannerbot sets mode: +v domidumont
07:06
mingus joined
07:07
p6bannerbot sets mode: +v mingus
07:08
mingus left
|
|||
| lizmat | .tell jnthn s/construts/constructs/ | 09:29 | |
| yoleaux | lizmat: I'll pass your message to jnthn. | ||
|
09:57
brrt joined
09:58
p6bannerbot sets mode: +v brrt
|
|||
| brrt | \o | 09:58 | |
| Kaiepi | o/ | 10:06 | |
| brrt | ohai Kaiepi | 10:07 | |
| I think you have some PRs open right | |||
| Kaiepi | yeah | ||
| they're not quite finished yet | 10:08 | ||
| brrt | ok :-) | ||
| timotimo | what was the last status of the investigation into the problem with asyncsocket? | ||
| brrt | I may have the ivnoke sequence almost ported to the expr jit | 10:09 | |
| timotimo | ooooh | 10:10 | |
| brrt | was simpler than expected actually | 10:11 | |
| it may be fathers' pride, it may be stockholm syndrome, but I kind of feel like the expr JIT language is not so bad.... | |||
| Kaiepi | i'm working on making it so fds are immediately available to both servers and clients as opposed to what i had finished before timo | 10:12 | |
| moar wasn't too bad to get working, but rakudo needs a lot of refactoring | 10:13 | ||
|
10:44
brrt left
11:10
robertle joined,
p6bannerbot sets mode: +v robertle
|
|||
| timotimo | cool | 11:49 | |
|
12:16
lizmat left
12:18
lizmat joined,
p6bannerbot sets mode: +v lizmat
14:28
lizmat left
14:39
Kaiepi left
14:40
Kaiepi joined
14:41
p6bannerbot sets mode: +v Kaiepi
15:20
buggable joined
|
|||
| Geth | MoarVM/nqp-mbc: 2f6b37e8f3 | (Stefan Seifert)++ | lib/MAST/Nodes.nqp Get rid of the remaining MAST nodes - bytecode is now written directly |
15:20 | |
| MoarVM/nqp-mbc: 2bcd8f389c | (Stefan Seifert)++ | 2 files Make bytecode dumper more resilient against invalid bytecode |
|||
|
15:21
p6bannerbot sets mode: +v buggable
15:32
fake_space_whale joined
15:33
p6bannerbot sets mode: +v fake_space_whale
15:51
lizmat joined,
p6bannerbot sets mode: +v lizmat
|
|||
| nine | So now that this thing is working correctly, I just need to shave off some 10s to get it competitive | 15:54 | |
| lizmat | nine: want me to look at some code ? | 15:55 | |
| timotimo | it'll probably also be interesting to read the spesh output | 15:58 | |
| like, does nqp turn every number into a float and then into an int? :P | |||
| nine | First I need to get rid of a lot of instruction list assembly. They are not used any more but I tried to keep the changes as small as possible. They still were large enough. | 15:59 | |
| timotimo | that does sound like a potential win | 16:00 | |
| nine | lizmat: I'll be sure to invoke the lizmat once I'm through with my plan :) | 16:02 | |
| jnthn | nine: Did you add the new MoarVM ops for writing binary data efficiently? :) | 16:11 | |
| nine++ for getting it this far already :D | |||
| nine | jnthn: the ones I needed, yes | ||
| Though they can be further optimized for sure | 16:12 | ||
| jnthn | Yeah, I was thinking of the integer ones, which'd avoid a bunch of bitshifting :) | 16:13 | |
| nine | Well writenum was even necessary because I had no idea how to implement that one in NQP | 16:18 | |
| jnthn | nine: btw, if not yet winning on speed, I'd expect your branch might already win on memory :) | 16:52 | |
| (since it's not constructing a giant tree) | 16:53 | ||
| timotimo | .o( borealectomy ) | 17:35 | |
| nine | I'd guess so, yes :) | ||
| lizmat | .oO( nine - the borealectomy guy ) |
17:53 | |
| nine | That....is so not my usual stance on trees :D | 17:57 | |
| lizmat | yet, perl 5 is known for its Swiss Army Chainsaw :-) | 17:58 | |
|
18:03
MasterDuke joined,
p6bannerbot sets mode: +v MasterDuke,
MasterDuke left,
MasterDuke joined,
herbert.freenode.net sets mode: +v MasterDuke,
p6bannerbot sets mode: +v MasterDuke
18:21
ggoebel left
18:36
ggoebel joined
18:37
p6bannerbot sets mode: +v ggoebel
18:41
brrt joined
18:42
p6bannerbot sets mode: +v brrt
|
|||
| brrt | \o | 18:44 | |
|
19:15
domidumont left
|
|||
| jnthn | o/ | 19:36 | |
|
19:40
MasterDuke left
|
|||
| brrt | I may have fixed the prepargs sequence | 19:41 | |
| lizmat | whee! | 19:45 | |
| that should help a lot with speed, no ? | 19:46 | ||
| brrt | no :-P | ||
| well, it could help a little, but I don't expect it to | 19:47 | ||
| look to jnthn++ for things that have to go faster :-P | |||
| lizmat | but it is nice to have anyway, right? otherwise , why do it ? | ||
| brrt | two reasons | 19:48 | |
| - to solve a bug in which an inline_end annotation was moved to a prepargs, where we'd allocate a label for it but not emit it (because the legacy JIT didn't handle annotations in the invoke sequence) | |||
| jnthn | lizmat: fwiw, this is - as I understand it - bug fixing, and separate from brrt++'s work on getting invocation handled in the expression JIT, which does indeed have potential for a bit of a speedup | 19:50 | |
| lizmat | aha! :-) ok | ||
| brrt | - and also just to get rid of legacy complexity | 19:53 | |
|
19:54
AlexDaniel left,
AlexDaniel joined
19:55
p6bannerbot sets mode: +v AlexDaniel
|
|||
| lizmat | m: role A { multi method a { ... } }; A.^pun # jnthn, is that expected ? | 20:06 | |
| camelia | Multi method 'a' with signature :(A: *%_) must be implemented by A because it is required by a role in block <unit> at <tmp> line 1 |
||
| lizmat | ah, I guess it is... duh | ||
| afk for half an hour or so | 20:07 | ||
| jnthn | Yes :) | ||
| brrt afk | 20:09 | ||
|
20:16
brrt left
|
|||
| nine | With a rather unscientific measurement I got 1.256G of RAM peak usage during settings compilation on nqp-mbc | 20:41 | |
|
21:19
AlexDaniel left,
robertle left
21:20
AlexDaniel joined,
p6bannerbot sets mode: +v AlexDaniel
|
|||
| japhb | nine: Compared to ...? | 21:54 | |
|
22:05
dogbert21 joined,
dogbert21 left,
dogbert2_ left
22:13
lizmat left
23:01
lizmat joined,
p6bannerbot sets mode: +v lizmat
23:24
MasterDuke joined,
p6bannerbot sets mode: +v MasterDuke
23:28
MasterDuke left,
MasterDuke joined,
herbert.freenode.net sets mode: +v MasterDuke,
p6bannerbot sets mode: +v MasterDuke
|
|||