|
00:45
cognome joined
00:48
cognome joined
01:19
FROGGS_ joined
01:45
cognome joined
02:45
cognome joined
03:45
cognome joined
04:45
cognome joined
05:37
FROGGS_ joined
05:45
cognome joined
05:55
FROGGS_ joined
06:38
FROGGS joined
06:45
cognome joined
06:56
camelia joined
07:10
zakharyas joined
07:28
camelia joined
07:45
cognome joined
08:04
kjs_ joined
08:45
cognome joined
08:58
FROGGS joined
09:30
camelia joined
09:45
cognome joined
09:48
ggoebel1111119 joined
10:03
JimmyZ joined
10:07
JimmyZ_ joined
10:09
JimmyZ__ joined
10:23
cognome joined
10:33
JimmyZ_ joined
10:34
JimmyZ__ joined
10:50
cognome joined
11:05
cognome joined
11:48
kjs_ joined
12:10
brrt joined
13:03
JimmyZ joined
13:09
LLamaRider joined
13:27
JimmyZ_ joined
13:51
kjs_ joined
13:57
JimmyZ joined
|
|||
| hoelzro | jnthn: I don't know how much time you have for Perl 6 right now, but I found a moar bug and I was hoping you could perhaps point me in the right direction as to how to fix it | 14:44 | |
| the bug is here: rt.perl.org/Ticket/Display.html?id=122773 | |||
| and the offending commit is 7a722dc4018ce1b99bbd4abea198e242d7da0157 | |||
| (if you have time, of course) | |||
| timotimo | jnthn: could you or masak give instructions on what exactly the escape analysis annotations mean and maybe i'll be able to annotate some more ops? | 14:45 | |
| jnthn | hoelzro: Quite time-constrained at the moment... | ||
| timotimo | so that you can get started doing the actual escape analysis code instead of slogging over the opcode annotations? | ||
| hoelzro | ok, don't worry about it =) | ||
| jnthn | timotimo: I'd not spend itme on that so far, we're not convinced we have the API right yet. | ||
| timotimo | OK | 14:46 | |
| jnthn | We've been discussing algorithm stuff around it over dinner a few times :) | ||
| timotimo | good to hear :) | ||
| i'm also wondering if removing boxing/unboxing redundancies is even possible before escape analysis exists | 14:47 | ||
| and if so: how to handle the relevant data | |||
| i suppose we could have known type + type == Int box type, but we don't store anything like "the instruction that boxes the value" or "the register and version the datum comes from" | 14:48 | ||
| (and in addition, i haven't actually identified any actual example code where it would make a noticable difference to throw out boxing/unboxing) | |||
|
15:00
woolfy1 left
15:02
woolfy joined
15:24
flussence joined
15:25
flussence joined
15:55
FROGGS joined
16:58
vendethiel joined
18:03
Ven joined
18:53
zakharyas joined
19:02
brrt joined
|
|||
| brrt | \o | 19:03 | |
| timotimo: dunno if you were talking about it, but it doesn't seem to me that the right place for escape analysis is moar | |||
| timotimo | oh? | 19:04 | |
|
19:04
Ven joined
|
|||
| brrt | moar basically only has a linear sequence of register ops | 19:04 | |
| nqp / perl6 compilers have a whole tree | |||
| timotimo | masak and jnthn are planning to implement it based on the ops | 19:05 | |
| brrt | welll | ||
| masak and jnthn are both much smarter in this regard, so they're probably right :-) | |||
| timotimo | hehe :) | 19:06 | |
| well, the tree can still have some transformations applied to them | 19:07 | ||
| and we may also only want to subject hot code to this analysis | |||
| depending on how much work it is | |||
| brrt wonders if we can get moarvm to run on minix | 19:12 | ||
| timotimo | that'll help us get mad street cred | ||
| brrt | i say | 19:14 | |
| timotimo | i'm now almost sure i'll go to the APW :) | 19:17 | |
| how do you feel about that? | |||
| FROGGS | +1 from me | 19:18 | |
| :o) | |||
| brrt | you should do that | ||
| multiboot has gotten so uninteresting since virtualisation | 19:29 | ||
|
19:32
camelia joined
|
|||
| brrt | i can't imagine going through the trouble of setting it up again | 19:32 | |
|
19:42
ggoebel11111110 joined
|
|||
| brrt | ugh... i'm actually forced to compile moarvm with clang | 19:50 | |
| ain't that an epic fail | 19:53 | ||
| libuv doesn't seem to support minix | |||
| wait, it's prettier than that | 19:54 | ||
| i can't checkout libuv because minix-git doesn't know how to deal with https urls | |||
| TimToady | .oO(the course of true minimalism never did run smooth) |
19:55 | |
| brrt | i'm guessing nobody's been dogfooding minix | 19:56 | |
| TimToady | obviously something named "moar" is against the spirit of something named "minix" :) | ||
| was minix ever more than a toy? | |||
| brrt | it's burned over 2,5 million euros of EU grant money over the past 5 years | 19:58 | |
| FROGGS | so it is a toy with bling bling now? :o) | 19:59 | |
| TimToady | that's chump change compared to the investment in linux | ||
| brrt | maybe so | 20:05 | |
|
20:07
camelia joined
20:36
cognome_ joined,
ggoebel11111110 joined
|
|||
| [Coke] wishes perl6 could get some of that money. :P | 21:26 | ||
| FROGGS | I would whish parrot and jvm wouldn't be that damn slow | 21:29 | |
| I would clearly suffer less | |||
|
23:05
cognome joined
|
|||