🦋 Welcome to the IRC channel of the core developers of the Raku Programming Language (raku.org #rakulang). This channel is logged for the purpose of history keeping about its development | evalbot usage: 'm: say 3;' or /msg camelia m: ... | log inspection situation still under development | For MoarVM see #moarvm
Set by lizmat on 22 May 2021.
Xliff p6-Suite timings have once again been cancelled this week, due to: github.com/rakudo/rakudo/issues/4655 05:25
lizmat Files=1349, Tests=117874, 307 wallclock secs (34.83 usr 9.15 sys + 4304.02 cusr 343.92 csys = 4691.92 CPU) 09:48
timo i wonder if we would be able to put retries in for precomp for cases where we're running up against some kind of resource limitation 17:24
.. are we currently doing in-process precomp or out-of-process precomp ...
ah yes, a tree of processes 17:30
nine To me it'd make more sense to fix the actual bugs 18:21
timo yes, true 18:22
gist.github.com/timo/a34409256a10f...0adf221771 - does this veeeery sloooowly leak for anybody else? 18:25
releasable6 Next release in ≈4 days and ≈23 hours. 4 blockers. Please log your changes in the ChangeLog: github.com/rakudo/rakudo/wiki/ChangeLog-Draft 19:00
MasterDuke timo: after how many iterations? i did it for 1_000 and heaptrack didn't think it leaked anything 20:28
timo imgur.com/FzJvjfl 20:33
does that link actually work?
yeah it does 20:34
MasterDuke i see the same thing in my heaptrack recording. i guess they're getting cleaned up on exit
so it's an hll leak, not a vm one i guess
timo i mean, this isn't really a program that should be using more data after the first minute 20:35
yeah i wasn't expecting moar to leak something here
MasterDuke looks like it's likely repr_clone -> copy_to 20:36
timo which repr are we cloning? quite possibly cloning MVMCode probably 20:37
MasterDuke dunno, heaptrack doesn't know that
it was only 849 allocations, but 442mb 20:38
timo i see clonings of scalar and IterationBuffer 20:41
MasterDuke then i suspect it's the IterationBuffer 20:42
huh. i ran for 100 iterations and see 4300 MVMCodes and 770 P6opaques, no iterationbuffers 20:43
timo 483328 ssize, could be the array of words we pick from 20:44
MasterDuke what if you shorten @w to just a few elements 20:45