|
03:01
jimmy_ joined
03:48
JimmyZ_ joined
|
|||
| hoelzro | should Moar bugs be reported on GH or RT? | 04:34 | |
| moritz | hoelzro: github | 05:36 | |
|
07:22
FROGGS joined
08:33
jimmy_ joined
09:15
zakharyas joined
11:40
woolfy joined
12:56
brrt joined
|
|||
| brrt | \o | 12:57 | |
| i just realised something | 13:00 | ||
| spesh can specialize escape analysis | |||
| like if you have $a.foo($b) - given if you know $a (and $a.foo), you can actually determine if $b will escape | 13:01 | ||
| a static compiler cannot | |||
| also: gist.githubusercontent.com/nitsanw...tfile1.asm - wtf, shl rbp ? | 14:21 | ||
| shift left base pointer? but why? | |||
| does jvm do this | 14:22 | ||
| scrap that i don't get that code *at all* | 14:26 | ||
|
16:02
FROGGS joined
16:13
rurban joined
18:03
vendethiel joined
19:10
FROGGS joined
|
|||
| timotimo | a perf report over setting compilation seems to think we spend 5% of time in MVM_gc_add_roots_to_worklist | 21:23 | |
| for some reason, MVM_JIT_DISABLE seems to have lost its effect | 21:29 | ||
| i don't think this report is correct. if the gc times grow and grow and grow, that ought to show a lot of time spent in gc stuff | 21:30 | ||