github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm
Set by AlexDaniel on 12 June 2018.
timotimo wakelift.de/2018/11/10/rakudo-perl...ss-report/ - lol i blogged a second time! 00:23
hoelzro timotimo: FWIW, the fastest I can seem to get a tree-building query for that DB you sent me is about 7s 17:40
it's been a fun little exercise, though =)
timotimo 7s for the entire thing? 17:41
or for one step? :D
hoelzro just to build the tree - no writes 17:42
I ended up making an index on parent_id to help a bit with the run time 17:43
but I'm guessing your code that you added is going to be more efficient
timotimo yeah, index on parent_id helps tremendously 17:46
Geth MoarVM: 1bf132125c | (Stefan Seifert)++ | src/6model/reprs/VMArray.c
Fix uninitialized values reported by valgrind.

  Thanks to dogbert11++ for reporting!
17:53
travis-ci MoarVM build passed. Stefan Seifert 'Fix uninitialized values reported by valgrind. 18:17
travis-ci.org/MoarVM/MoarVM/builds/453334521 github.com/MoarVM/MoarVM/compare/2...f132125ce4
dogbert11 nine++ 20:03
that fix seems to have unbusted the 32 bit build 20:04
there's still the problem with 'Register types do not match between value and node' in t/02-rakudo/12-proto-arity-count.t though. Seems to (possibly)) be expr jit related. 20:28
nine Oh that's a relief :) 20:59
timotimo hum. what conclusion did i last reach when i wondered how best to log what kind of wrongness happened to cause a deopt ... 21:24
maybe i'll just duplicate all the guard* ops under prof_ 21:28
it's only 9 anyway, 10 if you count sp_rebless, but for a start i'll only do sp_guard* 21:30