00:41 jnap joined 01:26 jnap joined 02:23 ggoebel11118 joined 03:43 woosley joined 04:20 jnap joined 06:34 woosley joined 07:22 jnap joined
dalek arVM: d60a2be | jimmy++ | src/gc/gen2.c:
Minor optimization to MVM_gc_gen2_compact_overflows

ignore moving object until find the first NULL object
07:31
08:08 FROGGS joined 08:14 brrt joined 08:23 jnap joined 08:36 FROGGS joined 08:46 lizmat joined 09:23 jnap joined 09:50 lizmat joined 10:24 jnap joined 10:29 woolfy joined 10:36 woosley joined 10:50 FROGGS joined 11:09 tgt joined, tgt left 11:20 woolfy left 11:25 jnap joined 12:26 jnap joined 12:41 jnap joined 14:10 btyler joined 16:36 cognominal joined 17:41 zakharyas joined 18:09 FROGGS joined 18:35 FROGGS joined 18:38 tgt joined 19:11 _sri joined 20:25 brrt joined
dalek arVM/moar-conc: dfdbdfc | jnthn++ | src/core/threadcontext.h:
Improve comment, layout.
20:32
arVM/moar-conc: edf210d | jnthn++ | src/ (5 files):
Revise work passing handling.

Go with a simpler scheme that is easier to reason about, and seems to be an improvement. Most important in this is giving the coordinator a responsibility to do the final check of in-trays, once everyone else has agreed they're done working.
arVM/moar-conc: f3bc9de | jnthn++ | src/core/threads.c:
Fix thread ID off-by-one.
timotimo jnthn: does that fix the trouble? 20:37
jnthn Not all of it
But it's an improvement.
22:03 brrt left
jnthn And the next trouble is proving hard to hunt down... 22:03
timotimo yay :\ 22:04
masak y = -|x| 22:27
jnthn Latest: it might be a gen2roots cross-thread thing. 22:36
Yeah. I have a Scalar allocated on thread 1, in gen2, that points into thread 3's fromspace. 22:52
m: say 104 +& 32 23:03
camelia rakudo-moar 3699aa: OUTPUT«32␤»
timotimo 8u8< 23:08
huh?
jnthn timotimo: yeah, that's what I'm wondering :) 23:17
23:26 woolfy joined
jnthn It has the "in gen2 root list" flag set but isn't actually in a root list of any thread. Tssk. 23:44
Think that's all I have the energy for tonight. At least I got it narrowed down a good bit. 23:45
23:59 harrow joined