Geth MoarVM: MasterDuke17++ created pull request #746:
Remove duplicate env variable usage line		 00:30
Geth MoarVM: d2a7fe3002 | MasterDuke17++ (committed using GitHub Web editor) | src/main.c
Remove duplicate env variable usage line
 01:26
MoarVM: 7f6dfabf20 | (Zoffix Znet)++ (committed using GitHub Web editor) | src/main.c
Merge pull request #746 from MasterDuke17/patch-1

Remove duplicate env variable usage line
brrt good * #moarvm 08:26
chrisseaton.com/rubytruffle/jokerconf17/ did anyone read this? 08:27
i thought it was pretty cool
about graal
so, two more random thoughts 08:36
- i still want to reduce the node size for the expression nodes to 32 bits, and 08:37
- i want to encode the number-of-args in an operator directly in the operator
so i can use the lower 16 bits as the operator and the upper as the number-of-args and maybe some other flags 08:38
maybe [operator (16), argnum(8), flags(8)]
jnthn brrt: Yes, I read that one 09:55
nwc10 good *, * 10:05
jnthn o/ 10:16
dogbert2 jnthn: how's the cold? 10:53
jnthn Clearing up, thanks, though I still feel oddly tired 11:08
dogbert2 running perl6 -e '' leaves 13 fd's open according to valgrind # guess that's ok 12:43
running perl6 -e 'await (^5).map({start { say qqx{echo -n foo $_} } })' leaves 54 fd's open # is that to be expected ? 12:44
lizmat dogbert2: maybe that includes threads ? 12:48
and the handles for async stdout/stdin/stderr 12:49
dogbert2 probably
lizmat you could try an END { nqp::force_gc } and see if that makes a difference 12:50
dogbert2 but shouldn't they be cleaned up at some point or is the OS supposed to do it
ilmari the OS cleans it up at exit, of course
dogbert2 lizmat: like this ? 'use nqp; await (^5).map({start { say qqx{echo -n foo $_} } }); END { nqp::force_gc }' 12:54
ilmari it's a bunch of pipes and eventpoll and eventfd handles
paste.scsys.co.uk/565690 12:55
lizmat dogbert2: yeah that looks like it
dogbert2 ilmari: couldn't this become a problem for a more long running, heavily threaded prg
lizmat: Unhandled exception: getexpayload needs a VMException, got P6opaque (X::AdHoc)
lizmat whee! 12:56
dogbert2 :)
lizmat that happens if you get an exec error in an END block
I guess we need some catcher there
dogbert2 lizmat: tried this before 'await (^5).map({start { say qqx{echo -n foo $_} } }); say "test"' I got the impression that the program hang always came after having written "test" 13:00
highly unscientific ofc :) 13:01
it's like the code suddenly fails to terminate properly 13:03
Zoffix Wasn't that one golfed down to a nqp proc assync start op? 13:05
dogbert2 does anyone happen to know how to get debug symbols for stuff like /lib/i386-linux-gnu/libc.so.6 and /lib/i386-linux-gnu/librt.so.1 13:06
Zoffix: did I miss that ?
Zoffix dogbert2: github.com/rakudo/rakudo/issues/12...-339720430 13:07
dogbert2: that was found by sticking debug prints and seeing where it hangs and eventually it showed it hangs on nqp::spawnprocasync op 13:08
dogbert2 aha, cool
dogbert2 tries the example
the annoying thing, for me at least, is that I only get hangs when I want SEGV's :)
Zoffix: you code hangs after a while like you wrote. If I leave it hanging I find 400+ lines, when running ps auxww, with the text '... [echo] <defunct>' 13:12
heh, the code only seems to hang, in fact it continues to run albeit superslow 13:18
dogbert17 new libuv 1.16 out, github.com/libuv/libuv/blob/v1.x/ChangeLog 17:40
Zoffix buggable: speed tests 18:06
buggable Zoffix, Known performance tests: I'll, csv, csv-R, csv-c, csv-c-20, csv-cc, csv-easy-pp, csv-easy-pp-pi, csv-easy-xs, csv-easy-xs-20, csv-go, csv-ip5pp, csv-ip5xs, csv-ip5xs-20, csv-ip5xsio, csv-java10, csv-java6, csv-java7, csv-java8, csv-java9, csv-lua, csv-parser, csv-pegex, csv-php, csv-python2, csv-python3, csv-ruby, csv-rust-csvrdr, csv-rust-libcsv, csv-rust-qckrdr, csv-test-pp, csv-test-xs, csv-test-xs-20, csv-xsbc, csv_gram, first, ip5xs, test, te
Zoffix wc
timotimo I'll 18:08
Zoffix That's from rogue entries in the source data :) "017-10-20 09:00:25 I'll try to remember to run the timings later today. System is rather busy and 3.253" 18:12
japhb timotimo: Message already sent to [Tux] about the rogue lines. :-)
timotimo heh.
samcv how do i induce crashes 19:49
i tried to get t/spec/S32-str/CollationTest_NON_IGNORABLE-3.t and t/spec/S32-str/encode.t to crash but couldn't even running them on loop
also tried setting MVM_SPESH_NODELAY=1 and MVM_SPESH_BLOCKING=1 anything else i can do?
nine, said he saw them under heavy load, but i have 8 threads so i guess i could try to do that but 19:50
samcv sounds like a pain is there an easier way 19:50
timotimo you can reduce the size of the nursery in src/gc/collect.h, and you can crank up the memory debug in src/gc/debug.h or what it was 19:54
nine samcv: are you running with MVM_SPESH_NODELAY=1 MVM_SPESH_BLOCKING=1? 20:19
oh, you did
samcv yes that's what i said above yeah 20:20
nine is just starting to read the backlog
dogbert17 samcv, nine: gist.github.com/dogbert17/f7c41756...8931eba85a 21:12
btw, the same valgrind complaint shows up in t/spec/S32-str/CollationTest_NON_IGNORABLE-3.t as well 21:29
samcv dogbert17++ 21:38
thank you
dogbert17 hi samcv
MasterDuke samcv: fwiw, i saw a fail in t/spec/S32-str/CollationTest_NON_IGNORABLE-3.t yesterday during one of my many spectest runs, but didn't try to reproduce
samcv others have been seeing it but i haven't been able to reproduce locally. but if there's valgrind issue that will make it much easier 21:39
samcv yeah it shows up only with gcc not with clang 22:03
Geth MoarVM: 11b12a268e | (Samantha McVey)++ | 2 files
Fix uninitialized variable access in MVM_unicode_get_property_str

With the more recent changes the variable doesn't get set if the codepoint is unknown. Make sure we initialize it so we don't get any segfaults or other issues.		 22:09
lizmat and another Perl 6 Weekly hits the Net: p6weekly.wordpress.com/2017/11/06/...tablished/ 22:39

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!