Perl 6 - the future is here, just unevenly distributed

IRC log for #moarvm, 2016-04-13

| Channels | #moarvm index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
01:24 diakopter timotimo: there used to be an instruction that encoded that, but (ironically, perhaps?) it might be optimized away
01:48 ilbot3 joined #moarvm
01:48 Topic for #moarvm is now https://github.com/moarvm/moarvm | IRC logs at  http://irclog.perlgeek.de/moarvm/today
06:50 geekosaur joined #moarvm
06:50 domidumont joined #moarvm
06:55 domidumont joined #moarvm
07:33 zakharyas joined #moarvm
07:41 jnthn timotimo: On callee side, you can look at the callsite to validate it
08:15 dalek MoarVM: ca1a21a | jnthn++ | src/core/frame.c:
08:15 dalek MoarVM: Fix fragility in VMNulling work registers.
08:15 dalek MoarVM:
08:15 dalek MoarVM: It could potentially use the spesh num_local_types when not using the
08:15 dalek MoarVM: spesh local_types.
08:15 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/ca1a21abf0
08:18 jnthn Unfortunately, this (well, actually the original patch that it makes more robust) accounts for a huge slowdown
08:21 jnthn Well, maybe not huge, but 10s or so on the Rakudo build
09:40 zakharyas joined #moarvm
09:41 ingy1 joined #moarvm
09:42 ilmari_ joined #moarvm
10:29 Unavowed joined #moarvm
10:34 domidumont joined #moarvm
11:35 ggoebel113 joined #moarvm
11:36 vendethiel joined #moarvm
11:43 psch joined #moarvm
12:23 lizmat joined #moarvm
13:37 orbus_ joined #moarvm
13:42 domidumont joined #moarvm
13:57 timotimo that's bad :|
13:58 jnthn Yeah, though we'll figure something out on it
13:58 jnthn Today I've been digging into the long-awaited invocation/frames re-design.
13:58 timotimo sounds exciting :)
13:59 jnthn Yeah...I think I can save us a decent bit
13:59 jnthn Along various axes
13:59 jnthn And make that refs_frames thing go away
14:09 mtj_ joined #moarvm
14:15 nwc10 and a pony?
14:15 timotimo refs_pony = 1?
14:18 zakharyas joined #moarvm
14:19 jnthn Well, it maybe prepares for the way for a pony ;)
14:31 domidumont joined #moarvm
15:33 ilmari joined #moarvm
15:35 zakharyas joined #moarvm
15:41 ggoebel113 joined #moarvm
15:47 geekosaur joined #moarvm
15:50 geekosaur joined #moarvm
16:05 domidumont joined #moarvm
16:07 domidumont joined #moarvm
16:30 zakharyas joined #moarvm
16:49 pyrimidine joined #moarvm
16:53 dalek MoarVM/reframe: 9103b61 | jnthn++ | src/6model/6model.h:
16:53 dalek MoarVM/reframe: Add collectable flag for frames.
16:53 dalek MoarVM/reframe: review: https://github.com/MoarVM/MoarVM/commit/9103b6147d
16:53 dalek MoarVM/reframe: 3daded3 | jnthn++ | src/core/frame.h:
16:53 dalek MoarVM/reframe: Give MVMFrame an MVMCollectable header.
16:53 dalek MoarVM/reframe: review: https://github.com/MoarVM/MoarVM/commit/3daded370d
16:53 dalek MoarVM/reframe: 95c87b2 | jnthn++ | src/gc/allocation. (2 files):
16:53 dalek MoarVM/reframe: Add function for GC-allocating a heap frame.
16:53 dalek MoarVM/reframe: review: https://github.com/MoarVM/MoarVM/commit/95c87b2975
16:53 dalek MoarVM/reframe: 1ce7271 | jnthn++ | src/core/frame. (2 files):
16:53 dalek MoarVM/reframe: Pull frame destroy out of reference dec.
16:53 dalek MoarVM/reframe: review: https://github.com/MoarVM/MoarVM/commit/1ce727182f
17:15 dalek MoarVM/reframe: d2591ad | jnthn++ | src/ (19 files):
17:15 dalek MoarVM/reframe: [WIP] Switch frames to be collectables.
17:15 dalek MoarVM/reframe:
17:15 dalek MoarVM/reframe: Danger: highly explosive branch; very unlikley to build NQP, though
17:15 dalek MoarVM/reframe: passes all but 2 of the NQP test suite. Pushed simply for backup.
17:15 dalek MoarVM/reframe:
17:15 dalek MoarVM/reframe: The eventual goal is to manage frames that are not referenced at all
17:15 dalek MoarVM/reframe: by other collectables in their life time using a traditional-style
17:15 dalek MoarVM/reframe: call stack, and promote them to the heap as normal GC-able objects
17:15 dalek MoarVM/reframe: upon escape. As a stepping stone, we'll allocate all frames on the
17:15 dalek MoarVM/reframe: heap, to help uncover missing mark bugs and the like, as well as to
17:15 dalek MoarVM/reframe: break the big change into steps between which stuff should entirely
17:15 jnthn dinner :) &
17:16 jnthn (btw, I've a list here of stuff to fix still on that branch, plus expect to rebase it further locally)
17:17 jnthn Also, I doubt this branch will merge until the stack part is done, as my guess would be that it'll represent a sizable regression.
17:17 jnthn (Performance wise)
17:17 jnthn Until that is in place.
17:17 jnthn Really dinner. &
17:18 timotimo <3
17:18 timotimo have a good food!
17:27 lizmat joined #moarvm
17:48 nwc10 jnthn: most recent commit on that branch SEGVs in NQP build (looks like a NULL pointer dereference)
17:48 nwc10 building previous commit to see how far I get
18:05 zakharyas joined #moarvm
18:39 jnthn nwc10: It'd be a near micracle if it worked, given there's still missing write barriers to add and some places it's not marking frames at all 'cus it got away with it before
18:40 jnthn (HEAD, that is)
18:40 jnthn The commits leading up to HEAD shoulda been benign refactors though
18:40 leedo joined #moarvm
18:42 nwc10 ASAN "likes" the commit before HEAD
18:43 cognominal joined #moarvm
19:07 patrickz joined #moarvm
19:33 dalek MoarVM/reframe: da21d2f | jnthn++ | src/ (3 files):
19:33 dalek MoarVM/reframe: Add some of the missing write barriers.
19:33 dalek MoarVM/reframe: review: https://github.com/MoarVM/MoarVM/commit/da21d2f5f8
19:33 dalek MoarVM/reframe: 17adaa4 | jnthn++ | src/ (3 files):
19:33 dalek MoarVM/reframe: Fix some missing frame marking.
19:33 dalek MoarVM/reframe: review: https://github.com/MoarVM/MoarVM/commit/17adaa4f49
19:33 dalek MoarVM/reframe: 6f61939 | jnthn++ | src/6model/ (44 files):
19:33 dalek MoarVM/reframe: Remove now-unused refs_frames.
19:33 dalek MoarVM/reframe: review: https://github.com/MoarVM/MoarVM/commit/6f61939df1
19:35 jnthn Innerestin'. If you disable spesh it passes all the NQP tests now
19:35 jnthn heh, and if you enable spesh but not inlining you get more fail
19:37 jnthn I've a horrible feeling this trail might lead to deopt.c
19:46 nwc10 doesn't build Rakudo, though
19:46 nwc10 failed to load library 'dynext/libperl6_ops_moar.so'   at gen/moar/m-BOOTSTRAP.nqp:1066  (<ephemeral file>:)
19:46 nwc10 ...
19:46 nwc10 however, that's the failure. not ASAN
19:57 orbus_ joined #moarvm
19:57 arnsholt joined #moarvm
19:57 _longines joined #moarvm
19:57 jnthn joined #moarvm
19:57 khagan_ joined #moarvm
20:04 _longines_ joined #moarvm
20:09 orbus_ joined #moarvm
20:09 arnsholt joined #moarvm
20:09 jnthn joined #moarvm
20:09 khagan_ joined #moarvm
20:17 jnthn joined #moarvm
20:17 arnsholt joined #moarvm
20:17 khagan joined #moarvm
20:26 orbus_ joined #moarvm
20:39 Util_ joined #moarvm
20:40 arnsholt_ joined #moarvm
20:44 harrow` joined #moarvm
20:53 ingy joined #moarvm
21:41 Util joined #moarvm
22:11 diakopter joined #moarvm

| Channels | #moarvm index | Today | | Search | Google Search | Plain-Text | summary