MasterDuke hm, guess i shouldn't use asprintf for just some of the no inlining reason cases 00:31
well, i could just add `if (no_inline_reason) free(no_inline_reason)`, right? 00:33
jnthn I didn't really worry about memory management there, but yeah, we'll have to if it was to, for example, store the data for profiling purposes 00:37
(Which would be nice) 00:38
MasterDuke well, i was just adding some sizes to the error messages 00:39
`asprintf(no_inline_reason, "bytecode (%d) is too large to inline, max is %d", cand->bytecode_size, MVM_SPESH_MAX_INLINE_SIZE);`
jnthn oh, I see 00:41
MasterDuke but it's actually kind of annoying, since then sometimes no_inline_reason is just directly assigned to, e.g., `no_inline_reason = "reason"` and sometimes it's asprintf'ed
ugh, could asprintf all the cases, then always freeing is fine 00:42
jnthn or strdup the constants :) 00:44
MasterDuke could add a #define for MVM_LOG_INLINES to inline.c and only assign to no_inline_reason inside an #if 00:46
jnthn: afk for a bit, but do you have a preference? and, think adding those values to the log message is useful enough for the bother? 00:47
jnthn MasterDuke: I think eventually we'll want it to send those messages to be included in profiler output; I'd ask timotimo about how that might best be done. 01:09
jnthn afk for sleep :) o/
MasterDuke .tell timotimo some conversation starting here irclog.perlgeek.de/moarvm/2018-03-18#i_15934422 where your input is requested 01:54
yoleaux MasterDuke: I'll pass your message to timotimo.
02:58 ilbot3 joined 03:41 releasable6 joined 07:18 SmokeMachine joined, committable6 joined, SmokeMachine joined 07:21 evalable6 joined 07:28 unicodable6 joined, notable6 joined, committable6 joined, reportable6 joined 07:30 samcv joined 08:10 domidumont joined 08:18 nativecallable6 joined, benchable6 joined 08:19 statisfiable6 joined, domidumont joined 11:03 shareable6 joined 12:38 yoleaux joined 12:49 AlexDaniel joined 14:20 brrt joined 15:24 brrt joined
timotimo MasterDuke: it'll be interesting to have all this info, but i'm thinking what collects it for the profile will go more of a "datastructure, flags, numbers" approach than "formatted strings" 16:11
17:22 squashable6 joined 20:41 evalable6 joined 20:44 unicodable6 joined
samcv timotimo and jnthn, do you want to write some text to include in the changelog for the new debug server? 20:52
jnthn Remote debugging support, optionally enabled with the --debug-port option to MoarVM 21:08
AlexDaniel fwiw it's also mentioned in rakudo changelog: github.com/rakudo/rakudo/wiki/ChangeLog-Draft 21:09
jnthn Or some such...I guess we could put in more detail but that's something :)
AlexDaniel because I thought it's notable
22:10 lizmat joined 22:25 L84DJMmarcusw joined
samcv timotimo: so do we have multithreading profiler support? 22:40
timotimo more or less. it's not entirely stable, i.e. it sometimes crashes, but the results are conveyed 22:42
samcv so I should add it to the changelog? 22:43
what should I say about it, should I mention it's not entirely stable?
timotimo yeah, mention that 22:51
japhb
.oO( Volatile/May Explode )
22:53
22:57
One of my favorites in that (hilarious) series
Geth MoarVM: 55061b58ae | (Samantha McVey)++ | docs/ChangeLog
Update ChangeLog for release
23:25
samcv changelog is up if someone can look at it and see if all is fine 23:26
or any changes I should make
MasterDuke samcv: `Fix CALL's size parameter in existskey template` probably could be excluded since the fix went in along with adding the existskey template 23:32
timotimo the strtoi64 thing is also a change that came in with the debug support, and only pertained to debugserver.c iirc 23:37
also ,the deadlock in unblocking therad GC thing is a thing that debugserver introduced and then fixed, so i would leave it out of the chaneglog 23:39
samcv thanks 23:48