|
00:15
vendethiel joined
01:04
vendethiel joined
02:47
ilbot3 joined
02:51
vendethiel joined
03:31
colomon joined
07:19
ggoebel joined
07:36
rurban joined
07:46
Ven joined
07:52
Ven joined
07:54
FROGGS joined
08:00
Ven joined
08:11
Ven joined,
zakharyas joined
08:41
donaldh joined
08:57
virtualsue joined
09:03
virtualsue_ joined
09:29
kjs_ joined
10:35
colomon joined
10:50
Ven joined
11:06
vendethiel joined
12:22
vendethiel joined
12:53
virtualsue joined
13:03
vendethiel joined
13:25
brrt joined
13:44
vendethiel joined
15:02
njmurphy joined
15:12
Ven joined
|
|||
| dalek | arVM: ec72fc0 | coke++ | docs/release_guide.md: Add a note about macport updates to the guide But seriously, just ping me for now, that's fine. |
15:43 | |
| arVM: e8d891c | coke++ | ports/macports/Portfile: track latest changes to make portfile function Ionic++ |
|||
|
16:27
Ven joined
17:15
kjs_ joined
17:30
Ven joined
18:02
colomon joined
|
|||
| dalek | arVM: f1c138f | timotimo++ | src/6model/reprs/P6opaque.c: bindattrs_s can be spesh'd, too. |
18:13 | |
|
18:30
colomon joined
18:53
colomon joined
19:02
FROGGS joined
19:34
kjs_ joined
19:38
kjs_ joined
19:44
colomon joined
20:18
virtualsue joined
20:42
vendethiel joined
22:10
kjs_ joined
22:31
kjs_ joined
|
|||
| timotimo | killing unused guards also doesn't free up space in the deopt table (which is MVM_alloc'd, so a potential win to reclaim stuff there) or the annotations list (where each annotation is spesh_alloc'd, so no win to be had) | 22:32 | |
| and removing the annotations when killing a log guard that belongs to it would require going through all annotations and fixing up the indices to deopt table entries | 22:33 | ||
| hm, my measurements already show that there exist cases where we end up removing more than 10 log-based guards, but that doesn't count guards that get thrown out because whole BBs have been eliminated | 22:45 | ||
|
23:37
lizmat joined
|
|||
| timotimo | huh, something's weird about this piece of dump ... | 23:54 | |
| gist.github.com/timo/bc2b12f65bbced5a7d8a - how does BB 2 end up with "2" as one of its successors? | |||