github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm
Set by AlexDaniel on 12 June 2018.
jnthn morning o/ 09:25
Altai-man_ o/ 09:37
raku-bridge <Rustem B.> o/ 09:45
Geth MoarVM/master: 14 commits pushed by (Daniel Green)++, Unknown++, (Jonathan Worthington)++
review: github.com/MoarVM/MoarVM/compare/6...94d8e1db02
10:26
MoarVM: a4b1959365 | (Jan-Olof Hendig)++ | 3rdparty/libuv
Update libuv to version 1.38.0

Clean stresstest under Linux (Mint)
MoarVM: b9fa480ad1 | (Jonathan Worthington)++ (committed using GitHub Web editor) | 3rdparty/libuv
Merge pull request #1302 from dogbert17/update-libuv-2-v1.38.0

Update libuv to version 1.38.0
lizmat jnthn: should I bump NQP and Rakudo for that to give a more spin ? 10:31
jnthn MOAR SPIN! 10:32
Yes, please ;) 10:33
*:)
lizmat oki
Geth MoarVM/new-disp: 70c0f8e66d | (Jonathan Worthington)++ | src/core/callsite.c
Correct named args copying logic

Not sure how it ended up this way, and it didn't matter until we started doing arg drops and insertions; now it is a problem, however.
12:34
Geth MoarVM: dogbert17++ created pull request #1311:
Add missing concreteness check
13:52
MoarVM/new-disp: 9938865825 | (Jonathan Worthington)++ | 3 files
Move some frame exit logic to callstack unwind

Eventually, more will move in this direction, hopefully with an amount of streamlining and simplification. For now, however, this removes us ending up with exceptions in boot-* dispatchers ending up in panic due to tc->cur_frame getting out of sync during dispatcher delegation.
14:05
Geth MoarVM/new-disp: 6574457be7 | (Jonathan Worthington)++ | src/core/args.c
Implement handling slurpy nameds for new call conv
14:39
MoarVM/new-disp: fec687d635 | (Jonathan Worthington)++ | src/disp/inline_cache.c
Give dispatch with flattening its own unlinked

Which for now just points out that we can't do flattening with dispatch yet, but will eventually be the entrypoint to pre-flattening the args ahead of the dispatch program (which in turn will lift the limitation on spesh of things called with flattening args).
14:52
Geth MoarVM/new-disp: 13c24f6d3a | (Jonathan Worthington)++ | src/disp/program.c
Mark dispatch recording attributes
16:09
Kaiepi how do i debug gc issues again? 20:23
iirc there were some macros i need to change somewhere in src/gc
timotimo first, make sure you're not experiencing what masterduke is fixing right now 20:27
after that, yeah there's a GC_DEBUG one
probably also depends on what exactly is going wrong 20:28
MasterDuke MVM_GC_DEBUG
for some reason i always forget it has MVM_ at the start. but luckily you can still find it easily 20:29
Kaiepi ah, i thought there were other macros besides that that needed setting 20:36
timotimo there are different memory-debug related things 21:01
you can reduce the size of the nursery to force gc runs to happen more often
MasterDuke any reason not to merge github.com/MoarVM/MoarVM/pull/1311 now? 21:54
timotimo does IS_CONCRETE do a null check first? 22:02
MasterDuke `#define IS_CONCRETE(o) (!(((MVMObject *)o)->header.flags & MVM_CF_TYPE_OBJECT))` 22:04
guess it should be `if (is->invocation_handler && IS_CONCRETE(is->invocation_handler))` 22:05
timotimo: how does gist.github.com/MasterDuke17/b966b...31feda0b31 look? 22:14
timotimo i don't know how MVM_HASH_BIND_FREE works exactly 22:18
so can't speak to the change in sc.c
MasterDuke it just runs the block it gets passed 22:19
like MVMROOT
timotimo ugh, i wish i could click "expand" on gisted patch files 22:20
MasterDuke ha 22:21
can you make your greasemonkey script linkify the filenames? 22:22
timotimo oh, hm, in theory i guess i can 22:23