Perl 6 - the future is here, just unevenly distributed

IRC log for #moarvm, 2017-12-03

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

All times shown according to UTC.

Time Nick Message
00:35 dalek joined #moarvm
00:35 synopsebot joined #moarvm
00:37 Geth joined #moarvm
00:37 SourceBaby joined #moarvm
00:57 patrickz_ joined #moarvm
00:59 evalable6 joined #moarvm
02:58 ilbot3 joined #moarvm
02:58 Topic for #moarvm is now https://github.com/moarvm/moarvm | IRC logs at  http://irclog.perlgeek.de/moarvm/today
04:26 reportable6 joined #moarvm
04:51 reportable6 joined #moarvm
08:16 AlexDaniel joined #moarvm
08:21 AlexDaniel joined #moarvm
09:03 robertle joined #moarvm
09:11 domidumont joined #moarvm
09:19 domidumont joined #moarvm
10:44 reportable6 joined #moarvm
10:45 reportable6 joined #moarvm
11:06 MasterDuke joined #moarvm
11:09 Ven`` joined #moarvm
13:03 brrt joined #moarvm
13:22 AlexDaniel joined #moarvm
13:31 reportable6 joined #moarvm
13:44 timotimo every few months i seem to fail at shrinking phi nodes
13:47 reportable6 joined #moarvm
13:54 moritz try psi nodes next time :-)
14:16 geospeck joined #moarvm
14:18 robertle joined #moarvm
14:59 geospeck joined #moarvm
15:19 timotimo the first difference that seems wrong appears in the spesh log at line 17.6k
15:20 timotimo hm, but what if it's actually correct, and the phi shrinking actually manages to do something good?
15:20 timotimo and i've just been staring at perfectly benign code ...
15:43 Ven`` joined #moarvm
16:02 Voldenet joined #moarvm
16:02 Voldenet joined #moarvm
16:08 lizmat joined #moarvm
16:09 dogbert17 timotimo: are there any 'easy' optimization wins left in MoarVM now?
16:13 dogbert17 clickbaits lizmat: https://irclog.perlgeek.de/perl6-dev/2017-12-03#i_15525445
16:18 * lizmat looks
16:19 lizmat ah, test-t < 3 ?   yeah cool!
16:20 dogbert17 jnthn and timotimo at work
16:21 * dogbert17 oO (stumbles upon a valgrind error in a Proc::Async-test)
16:22 dogbert17 ==20866== Conditional jump or move depends on uninitialised value(s)
16:22 dogbert17 ==20866==    at 0x418355B: optimize_throwcat (optimize.c:1728)
16:22 dogbert17 ==20866==    by 0x418355B: second_pass (optimize.c:2348)
16:22 dogbert17 ==20866==    by 0x41834F6: second_pass (optimize.c:2356)
16:23 jnthn Poor cat...
16:24 dogbert17 indeed :)
16:24 dogbert17 was going to look for potential memory leaks but got distracted by this
16:26 dogbert17 test 36 in t/spec/S17-procasync/basic.t ('Process that doesn't output anything will not emit')
16:28 dogbert17 should be this piece of code: https://github.com/MoarVM/MoarVM/blob/master/src/spesh/optimize.c#L1728
16:30 jnthn Could well be
16:30 jnthn Oh, I wonder
16:30 jnthn It could be the new inline boundary indicators I stuck into the handlers table
16:31 jnthn That probably doesn't set action
16:31 jnthn But since it's just an integer compare, and even if it accidentally matches then the line below won't match anyway, it'll be benign
16:31 jnthn Still, worth squishing the warning
16:31 dogbert17 so you've already solved it, impressive. Do you want an issue?
16:32 jnthn https://github.com/MoarVM/MoarVM/blob/master/src/spesh/inline.c#L549
16:32 jnthn Adding inliner->handlers[*inline_boundary_handler].action = 0; here will likely clean up the warning
16:33 jnthn Yeah, issue, or if you're up for trying out adding that line and seeing if it helps and sending a PR, that also works :)
16:33 dogbert17 I could do that unless you'd like to do it yourself :)
16:33 jnthn I can't right now :)
16:34 dogbert17 ok, you can see a PR in your future :)
16:34 jnthn Cool, thanks :)
16:34 * jnthn away for a bit
16:46 vendethiel- joined #moarvm
16:54 Geth ¦ MoarVM: dogbert17++ created pull request #760: Impl. of the new inline boundary indicators forgot to set action
16:54 Geth ¦ MoarVM: review: https://github.com/MoarVM/MoarVM/pull/760
17:03 wander joined #moarvm
17:19 Ven`` joined #moarvm
18:12 patrickz joined #moarvm
18:16 Geth ¦ MoarVM: 41050d1894 | (Jan-Olof Hendig)++ | src/spesh/inline.c
18:16 Geth ¦ MoarVM: Impl. of the new inline boundary indicators forgot to set action
18:16 Geth ¦ MoarVM:
18:16 Geth ¦ MoarVM: The newly implemented inline boundary indicators omitted to set
18:16 Geth ¦ MoarVM: the action field. This led to a warning from valgrind. Although
18:16 Geth ¦ MoarVM: the warning was benign in this case, it should be fixed anyway.
18:16 Geth ¦ MoarVM: jnthn++ for providing the fix.
18:16 Geth ¦ MoarVM: review: https://github.com/MoarVM/MoarVM/commit/41050d1894
18:16 Geth ¦ MoarVM: 3c7b49e8f0 | (Jonathan Worthington)++ (committed using GitHub Web editor) | src/spesh/inline.c
18:16 Geth ¦ MoarVM: Merge pull request #760 from dogbert17/fix-uninit-action
18:16 Geth ¦ MoarVM:
18:16 Geth ¦ MoarVM: Impl. of the new inline boundary indicators forgot to set action
18:16 Geth ¦ MoarVM: review: https://github.com/MoarVM/MoarVM/commit/3c7b49e8f0
19:35 releasable6 joined #moarvm
19:36 unicodable6 joined #moarvm
19:36 bisectable6 joined #moarvm
19:36 committable6 joined #moarvm
20:33 benchable6 joined #moarvm
21:40 greppable6 joined #moarvm

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