github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm
Set by AlexDaniel on 12 June 2018.
nwc10 good *, #moarvm 05:23
07:47 sena_kun joined 08:27 vrurg left 08:29 vrurg joined 09:12 Altai-man_ joined 09:14 sena_kun left 09:44 MasterDuke joined
nwc10 OK, so I decided to cheat, and built the setting with a known working moarvm. and (after more cheating) I can run -e0 10:02
and t/01-sanity/18-simple-multisubs.t fails
(and then a bunch more)
but it's failing like this
Can't use unknown trait 'is' -> 'onlystar' in a sub declaration.
which looks very similar to the crazy in (failing) to build the setting 10:03
timotimo yeah that's a multi dispatch failure reported as "you wanted something that doesn't exist" 10:07
10:19 vrurg left
nwc10 and the testcase reduces to 10:19
multi sub foo () {}
10:20 vrurg joined 10:21 sena_kun joined 10:23 Altai-man_ left 10:24 vrurg left 10:56 vrurg joined 10:57 vrurg left, vrurg joined
timotimo nice 11:28
i just patched the visualizer to ignore heaps that have a size of 0 in their entry 12:17
though i must assume i'll be missing important data
firefox is having trouble opening the file ..?! 12:19
12:20 Altai-man_ joined 12:23 sena_kun left
timotimo imgur.com/UlSmaN1 12:24
so what does this tell me, i wonder
Red areas are memory locations in use. 12:27
Gray areas are free locations, but not released back to the kernel.
White areas are freed to the kernel.
so i think that means a malloc_trim ought to release about 33.5 megabytes of memory to the kernel? 12:28
the visualizer writes out every little square as its own png, all of them are only a few bytes big, but use 4 kilobytes on disk because reasons 12:29
12:51 lucasb joined 13:13 vrurg left 13:17 vrurg joined
timotimo tried to include github.com/theoricien/ptmalloc2v, but facing a couple of issues there as well, and the output isn't graphical, it's pages and pages and pages and pages and pages of output 14:00
14:21 sena_kun joined 14:22 vrurg left 14:23 vrurg joined, Altai-man_ left 14:27 vrurg left
timotimo you are supposed to call heap_init before calling malloc or calloc, but ltrd calls it already before your own binary even gets to have its entry point invoked 14:30
14:45 vrurg joined 16:20 Altai-man_ joined 16:23 sena_kun left 16:44 vrurg left, vrurg joined 16:48 vrurg left 17:16 vrurg joined 18:22 sena_kun joined, vrurg left 18:23 vrurg joined, Altai-man_ left 18:27 vrurg left 18:41 vrurg joined 19:03 sena_kun left 19:05 sena_kun joined 20:21 Altai-man_ joined 20:23 sena_kun left 21:55 Altai-man_ left 22:06 sena_kun joined 22:21 Altai-man_ joined 22:23 sena_kun left 22:32 vrurg left