Perl 6 - the future is here, just unevenly distributed

IRC log for #moarvm, 2016-12-19

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

All times shown according to UTC.

Time Nick Message
00:33 pyrimidine joined #moarvm
00:47 pyrimidine joined #moarvm
01:39 pyrimidi_ joined #moarvm
02:26 pyrimidine joined #moarvm
02:48 ilbot3 joined #moarvm
02:48 Topic for #moarvm is now https://github.com/moarvm/moarvm | IRC logs at  http://irclog.perlgeek.de/moarvm/today
04:15 dalek joined #moarvm
04:23 stmuk_ joined #moarvm
07:45 pyrimidine joined #moarvm
07:55 domidumont joined #moarvm
07:58 pyrimidine joined #moarvm
08:00 domidumont joined #moarvm
08:20 pyrimidine joined #moarvm
08:31 pyrimidine joined #moarvm
08:33 dalek MoarVM: f777352 | niner++ | build/setup.pm:
08:33 dalek MoarVM: Remove seemingly unneeded and harmfull ldrpath from build
08:33 dalek MoarVM:
08:33 dalek MoarVM: The @prefix@/share/perl6/site/lib directory does not even exist on a normal
08:33 dalek MoarVM: installation of rakudo, nqp and moarvm. "@/libdir" does not seem to be needed
08:33 dalek MoarVM: either as even with it removed, everything seems to work just fine.
08:33 dalek MoarVM: Removing because rpath is prohibited on Fedora and strongly discouraged on
08:33 dalek MoarVM: openSUSE.
08:33 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/f777352eba
08:39 pyrimidine joined #moarvm
08:47 nwc10 niner: that does break "the world" when it's installed to a private path
08:47 nwc10 $ ldd /home/nicholas/Sandpit/moar-san/bin/moar linux-vdso.so.1 =>  (0x00007fffc5fa8000) libasan.so.1 => /usr/local/gcc49/lib64/libasan.so.1 (0x00007feb5e6d2000) libubsan.so.0 => /usr/local/gcc49/lib64/libubsan.so.0 (0x00007feb5da09000) libmoar.so => not found
08:47 nwc10 ...
08:48 nwc10 (oh, that got eaten
08:48 nwc10 libmoar.so => not found
08:48 nwc10 there^^
08:48 nwc10 the relevant library is in /home/nicholas/Sandpit/moar-san/lib/libmoar.so
08:51 travis-ci joined #moarvm
08:51 travis-ci MoarVM build failed. Stefan Seifert 'Remove seemingly unneeded and harmfull ldrpath from build
08:51 travis-ci https://travis-ci.org/MoarVM/MoarVM/builds/185085866 https://github.com/MoarVM/MoarVM/compare/0595ff82f3c1...f777352ebae2
08:51 travis-ci left #moarvm
08:52 nwc10 I think that it needs this: http://paste.scsys.co.uk/540273
09:03 pyrimidine joined #moarvm
09:06 samcv hi having a problem. getting, can't find moar.so
09:06 samcv used perl Configure.pl --prefix="$HOME/perl6"
09:06 samcv pretty sure that's what i used before, i mean that's where i have all my moar/nqp/rakudo things
09:07 samcv am I doing something wrong?
09:08 samcv nqp said my moar was too old, so i pull'd and then ran make again... and then broke :(
09:10 samcv nine, you broke my Moar ._.
09:10 samcv reverted that last commit. and works fine
09:13 samcv just read through all the scrollback hah. looks like others had problems
09:13 pyrimidine joined #moarvm
09:23 pyrimidine joined #moarvm
09:29 dalek MoarVM: 71bbf62 | lizmat++ | build/setup.pm:
09:29 dalek MoarVM: Revert "Remove seemingly unneeded and harmfull ldrpath from build"
09:29 dalek MoarVM:
09:29 dalek MoarVM: This reverts commit f777352ebae29349d1bfa9e6c03c12f32f2ec689.
09:29 dalek MoarVM:
09:29 dalek MoarVM: It appears to cause grave problems
09:29 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/71bbf628ff
09:29 lizmat samcv nwc10 ^^^
09:29 samcv cool
09:29 samcv thank you :)
09:30 lizmat .tell niner I've reverted f777352 so people can continue on, see backlog
09:30 yoleaux2 lizmat: I'll pass your message to niner.
09:30 lizmat .tell nine I've reverted f777352 so people can continue on, see backlog
09:30 yoleaux2 lizmat: I'll pass your message to nine.
09:43 pyrimidine joined #moarvm
09:46 travis-ci joined #moarvm
09:46 travis-ci MoarVM build passed. Elizabeth Mattijsen 'Revert "Remove seemingly unneeded and harmfull ldrpath from build"
09:46 travis-ci https://travis-ci.org/MoarVM/MoarVM/builds/185096120 https://github.com/MoarVM/MoarVM/compare/f777352ebae2...71bbf628ff3c
09:46 travis-ci left #moarvm
09:52 nine Ah, damn. It probably worked on my machine because I still had the moarvm rpm installed when testing install to a private path.
09:52 yoleaux2 09:30Z <lizmat> nine: I've reverted f777352 so people can continue on, see backlog
09:53 nine Sorry folks :/
09:53 timotimo oops! :)
09:54 dogbert17_ heh, "Must not GC when in the specializer/JIT"
09:55 dogbert17_ this I haven't seen before
09:56 jnthn Yeah, that was added last week
09:56 jnthn To panic instead of corrupt memory :)
09:57 jnthn Also, the stack trace of the point in panics will be useful for fixing it :)
09:57 dogbert17_ jnthn: will this do as report material? https://gist.github.com/dogbert17/4c7d21ce1f075473c1ab0411cf3c3260
09:58 timotimo oooh, demand_extop, eh?
09:58 timotimo because it locks + unlocks the mutex?
09:59 timotimo i expect we'd mark a thread blocked before and unblocked after spesh anyway, right?
09:59 timotimo at least ... we probably should? except spesh is supposed to be blistering fast anyway :P
10:00 jnthn dogbert17_: Yup
10:01 dogbert17_ cool, I'll add it as a MoarVM issue
10:05 pyrimidine joined #moarvm
10:10 dogbert17_ btw, are there any other flags apart from FSA_SIZE_DEBUG and MVM_GC_DEBUG that might be useful to turn on for bug hunting purposes?
10:12 timotimo depends on what kind of bug you're hunting
10:13 timotimo if there is/you're suspecting a spesh bug, there's a tool that'll help identify the offending frame for you
10:13 jnthn There's also various others that turn on extra debug output
10:13 jnthn But you need to know what kind of problem you're looking at for them to be much use
10:14 jnthn Doubly so given extra output introduces timing effects that can hide certain problems you might be hunting
10:14 dogbert17_ thx for the explanantion, I guess I'll stick with the ones above for the time being then :-)
10:15 jnthn For example, there's ways to turn on logging output for GC orchestration and deopt
10:16 jnthn Though both produce a decent amount of into that is likely to be meaningless for most bug hunts
10:17 dogbert17_ ok, I'll definitely stick to the above two then, at least until they are exhausted :-)
10:24 timotimo oh, hm, it's more nuanced than setting gc to blocked when spesh runs
10:25 timotimo because we do have many spots that have interactions with the gc, such as log entries
10:25 timotimo and spesh slots
10:28 jnthn Far more
10:28 jnthn The spesh rule is that GC never takes place while it's running
10:30 timotimo right. so we'll need an API (or maybe only usable by spesh? though extops can also define spesh) that locks and unlocks mutexes without doing the blocked/unblocked thing
10:31 jnthn It's not that simple
10:31 jnthn The holder of the mutex might have decided to do GC
10:31 jnthn Then deadlock
10:33 timotimo ugh.
10:46 timotimo .o( but i thought acquiring a mutex marks a thread as blocked anyway? )
10:49 pyrimidine joined #moarvm
10:52 brrt joined #moarvm
10:52 brrt \o #moarvm
10:52 timotimo o/
10:54 brrt \o timotimo
10:54 brrt i've figured out a new plan for the template pointer removal thingy
10:55 brrt first add all the new stuff, then remove the old stuff
10:56 jnthn o/ brrt
10:56 zakharyas joined #moarvm
10:57 brrt \o jnthn. nice post as usual :-)
11:01 jnthn Glad you liked it :-)
11:14 pyrimidine joined #moarvm
11:18 lizmat Unexpected named argument 'exists' passed   # /me is puzzled
11:19 lizmat ah, subs
11:37 lizmat joined #moarvm
11:38 pyrimidine joined #moarvm
11:58 pyrimidine joined #moarvm
12:15 dalek MoarVM: fb6cc14 | niner++ | build/setup.pm:
12:15 dalek MoarVM: Remove at least the obsolete path from ldrpath
12:15 dalek MoarVM:
12:15 dalek MoarVM: As it seems like we still need the @libdir@ rpath, the non-existing
12:15 dalek MoarVM: @prefix@/share/perl6/site/lib can at least still be removed so it will
12:15 dalek MoarVM: not cause any future confusion and to solve at least half the problem.
12:15 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/fb6cc147ad
12:23 pyrimidine joined #moarvm
12:26 lizmat joined #moarvm
13:09 pyrimidine joined #moarvm
13:16 pyrimidine joined #moarvm
13:23 pyrimidine joined #moarvm
13:25 SmokeMachine Any one have any idea what could be causing this? https://gist.github.com/FCO/bb1bb1c0285fcfa12d8f2d8789beb21f
13:26 [Coke] open an RT with both the source and the bt.
13:27 SmokeMachine [Coke]: but the source is too big... I'm trying to reduce it...
13:28 timotimo SmokeMachine: att he point of an explosion, you can "call MVM_print_backtrace(tc)" (or what it's called) and get a stack trace from inside your program
13:28 timotimo that can help guide the golfing effort
13:37 SmokeMachine timotimo: No symbol "MVM_print_backtrace" in current context.
13:37 timotimo then it must be MVM_dump_backtrace
13:37 nine yep it's MVM_dump_backtrace
13:38 nine also you may have to up a couple of times
13:38 timotimo not in that stack trace, though, as it's inside a gc_mark function
13:42 pyrimidine joined #moarvm
13:42 SmokeMachine :) https://gist.github.com/FCO/c879dbd3506ca9265cc9dde7a685d4c0
13:44 SmokeMachine timotimo: worked! \o/ ^^
13:53 SmokeMachine looks that's an error on (|) op
14:15 timotimo it's quite unlikely that it has anything to do with (|) itself
14:15 timotimo probably something between then and the previous GC run?
14:38 pyrimidine joined #moarvm
14:54 colomon joined #moarvm
15:11 pyrimidine joined #moarvm
15:30 pyrimidine joined #moarvm
16:10 pyrimidine joined #moarvm
16:50 zakharyas joined #moarvm
17:32 SmokeMachine Maybe...
17:34 FROGGS joined #moarvm
17:36 pyrimidine joined #moarvm
17:59 stmuk joined #moarvm
18:09 pyrimidine joined #moarvm
18:29 lizmat joined #moarvm
18:32 pyrimidine joined #moarvm
18:56 pyrimidine joined #moarvm
19:00 notviki joined #moarvm
19:14 pyrimidine joined #moarvm
19:31 pyrimidine joined #moarvm
19:45 yoleaux2 joined #moarvm
19:49 pyrimidine joined #moarvm
20:06 pyrimidine joined #moarvm
20:17 pyrimidine joined #moarvm
20:28 pyrimidine joined #moarvm
21:03 pyrimidine joined #moarvm
21:04 camelia joined #moarvm
21:09 pyrimidine joined #moarvm
21:35 dalek joined #moarvm
21:36 pyrimidine joined #moarvm
21:53 pyrimidine joined #moarvm
22:15 pyrimidine joined #moarvm
22:32 colomon_ joined #moarvm
22:45 pyrimidine joined #moarvm
23:07 stmuk_ joined #moarvm
23:10 pyrimidine joined #moarvm
23:16 dalek Heuristic branch merge: pushed 20 commits to MoarVM/line_based_coverage_5 by timo
23:41 colomon joined #moarvm
23:51 pyrimidine joined #moarvm
23:55 pyrimidi_ joined #moarvm

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