|
01:48
ilbot3 joined
06:08
domidumont joined
06:13
domidumont joined
07:31
zakharyas joined
14:14
domidumont joined
14:53
domidumont joined
15:43
lizmat_ joined
15:50
mtj_ joined
15:53
geekosaur joined
15:56
TheLemonMan joined
16:00
domidumont joined
|
|||
| timotimo | darn. | 17:29 | |
| with the line based coverage turned on, some spec tests die like this: | |||
| Spesh: failed to fix up inline 3 | |||
| had to turn the while condition of the "eliminate dead bbs" loops to "cur_bb && cur_bb->linear_next" instead of just "cur_bb->linear_next" | 17:30 | ||
|
17:37
avar joined
|
|||
| timotimo | ideally, a failed spesh would bring us out of spesh, not out of the whole program | 18:14 | |
| and it's kind of weird to get an exception thrown when spesh fails, because ... are you going to react to that? it could happen literally anywhere | |||
|
18:31
domidumont joined
18:42
TimToady joined
20:02
nebuchad` joined
20:11
_longines_ joined,
dalek joined,
synopsebot6 joined
20:12
mtj_ joined
20:59
Ven_ joined
23:00
brrt joined
|
|||
| brrt | timotimo: that is why we have MVM_oops | 23:04 | |
| to die with a stack trace but beyond catchinf | 23:05 | ||
| also when spesh dies it might well be because either therw is a bug in its handling or because memory got corrupted | 23:06 | ||
| and in those cases dying quickly is I think preferable to corruption | 23:07 | ||