Perl 6 - the future is here, just unevenly distributed

IRC log for #moarvm, 2017-10-20

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

All times shown according to UTC.

Time Nick Message
01:18 nativecallable6 joined #moarvm
01:18 quotable6 joined #moarvm
01:18 committable6 joined #moarvm
01:18 bloatable6 joined #moarvm
01:18 bisectable6 joined #moarvm
01:18 greppable6 joined #moarvm
01:18 releasable6 joined #moarvm
01:18 benchable6 joined #moarvm
01:18 evalable6 joined #moarvm
01:18 coverable6 joined #moarvm
01:18 unicodable6 joined #moarvm
01:18 statisfiable6 joined #moarvm
01:18 squashable6 joined #moarvm
01:20 lizmat joined #moarvm
01:56 ilbot3 joined #moarvm
01:56 Topic for #moarvm is now https://github.com/moarvm/moarvm | IRC logs at  http://irclog.perlgeek.de/moarvm/today
02:16 evalable6 joined #moarvm
04:02 evalable6 joined #moarvm
06:02 domidumont joined #moarvm
06:38 nativecallable6 joined #moarvm
06:38 greppable6 joined #moarvm
06:38 evalable6 joined #moarvm
06:38 committable6 joined #moarvm
06:38 quotable6 joined #moarvm
06:38 coverable6 joined #moarvm
06:38 unicodable6 joined #moarvm
06:38 bisectable6 joined #moarvm
06:38 bloatable6 joined #moarvm
06:38 releasable6 joined #moarvm
06:38 benchable6 joined #moarvm
06:38 squashable6 joined #moarvm
06:38 statisfiable6 joined #moarvm
06:52 Ven joined #moarvm
07:10 domidumont joined #moarvm
07:27 Ven_ joined #moarvm
07:57 domidumont joined #moarvm
08:03 Ven joined #moarvm
08:03 rba joined #moarvm
08:13 zakharyas joined #moarvm
08:15 robertle joined #moarvm
08:34 rba_ joined #moarvm
09:00 rba joined #moarvm
09:01 rba timotimo: I have now build MoarVM on Solaris with --debug=3 and --optimize=0. Got todays MoarVM and nqp from git and still get same test errors:
09:26 rba timotimo: t/nqp/059-nqpop.t and t/nqp/104-method-cache.t failed
09:26 rba not ok 92 - nqp::pow_n(1, nqp::nan)
09:26 rba not ok 93 - nqp::pow_n(1, nqp::inf)
09:26 rba not ok 94 - nqp::pow_n(1, nqp::neginf)
09:27 rba # ./nqp-m t/nqp/104-method-cache.t
09:27 rba 1..16
09:27 rba ok 1 - calling a method found in cache
09:27 rba ok 2 - nqp::can on a method found in cache
09:27 rba ok 3 - nqp::can on a method not in cache
09:27 rba ok 4 - ...find_method is not called when the cache is authoritative
09:27 rba Segmentation Fault (core dumped)
09:31 rba bash-4.3# ./nqp-m -e 'say(nqp::pow_n(1, nqp::nan))'
09:31 rba NaN
09:31 rba bash-4.3# ./nqp-m -e 'say(nqp::pow_n(1, nqp::inf))'
09:31 rba NaN
09:31 rba bash-4.3# ./nqp-m -e 'say(nqp::pow_n(1, nqp::neginf))'
09:31 rba NaN
09:44 rba About the segmentation fault:
09:44 rba https://gist.github.com/b83f789c58dcfda29a53e682e2f4edd0
09:46 rba If someone could adivise me how to "cook" the gdb to get valuable information to dig to the root cause, let me know. dbg session still open...
09:49 rba https://gist.github.com/21673ab1df34016a06c5128910b67203
10:01 Ven joined #moarvm
10:23 Ven_ joined #moarvm
11:01 Ven joined #moarvm
11:16 rba joined #moarvm
11:42 patrickz joined #moarvm
11:49 timotimo rba: if the debug session for the seg fault is still open, please "frame 5" and "print c_name"
11:57 samcv yhello guys
11:57 yoleaux 19 Oct 2017 20:53Z <AlexDaniel`> samcv: https://irclog.perlgeek.de/moarvm/2017-10-19#i_15327049
11:57 samcv so i volunteered for something :) ok
11:57 samcv oh yes i did
11:57 AlexDaniel` hehe :D
11:58 samcv i've been needing something to do anyway
11:58 samcv releasable6, status
11:58 releasable6 samcv, Next release in 1 day and ≈7 hours. 1 blocker. Changelog for this release was not started yet
11:58 releasable6 samcv, Details: https://gist.github.com/37d496766d04eaa162c794138697e98a
11:58 samcv ok cool
11:58 samcv AlexDaniel`, there's a list right? of things to be done before release?
11:59 AlexDaniel` samcv: https://github.com/MoarVM/MoarVM/blob/master/docs/release_guide.md
11:59 samcv ok cool
11:59 AlexDaniel` samcv: arguably the most time consuming part is the changelog, so that's probably something you want to start with early
12:00 samcv that's what I was thinking too
12:01 AlexDaniel` samcv: if you have any extra resources, there's one thing that can be improved in the process (and in that document) – signing
12:01 AlexDaniel` I've noticed that your commits are signed so it shouldn't be too hard for you
12:02 AlexDaniel` samcv: rakudo release guide talks about signing (e.g. step 16) https://github.com/rakudo/rakudo/blob/nom/docs/release_guide.pod
12:03 AlexDaniel` but depending on how you have your git set up it may actually happen automatically
12:03 AlexDaniel` then you'll just need to sign the tarball (like in step 17), and that's basically it
12:05 samcv i sign all my commits
12:06 samcv only things i don't sign is doing merges using github
12:06 AlexDaniel` yeah, me too
12:14 Ven joined #moarvm
12:47 zakharyas joined #moarvm
13:33 rba joined #moarvm
13:36 rba timotimo: output from "frame 5" and "print c_name": https://gist.github.com/38c6b0898d6dd2fd40076d0fda2ab330
13:37 timotimo ok, what's STABLE(obj)? and what's STABLE(obj)->debug_name?
13:37 timotimo (again with print)
13:38 timotimo i wonder if debug_name is a null pointer and that simply crashes on freebsd whereas it just prints an empty string on other systems?
13:39 timotimo ah, yes, passing NULL there is "officially undefined"
13:42 timotimo i'll build something
13:42 rba timotimo: https://gist.github.com/f9738f74c0358fc45ba61e2f6d22a26b
13:42 Ven joined #moarvm
13:43 timotimo yup, that's it
13:43 timotimo might be time to run moarvm with ubsan at one point
13:44 rba timotimo: what can I do now?
13:44 timotimo wait for my patch
13:44 rba timotimo: nice!
13:50 arnsholt joined #moarvm
13:55 timotimo so, so, so many usages of debug_name everywhere
14:09 rba joined #moarvm
14:12 Geth ¦ MoarVM/debug_name_safe_null: 51158fa1b3 | (Timo Paulssen)++ | 21 files
14:12 Geth ¦ MoarVM/debug_name_safe_null: don't printf NULL via an unset debug_name
14:12 Geth ¦ MoarVM/debug_name_safe_null:
14:12 Geth ¦ MoarVM/debug_name_safe_null: it's undefined behaviour and actually segfaults
14:12 Geth ¦ MoarVM/debug_name_safe_null: under solaris.
14:12 Geth ¦ MoarVM/debug_name_safe_null: review: https://github.com/MoarVM/MoarVM/commit/51158fa1b3
14:12 timotimo rba: try this branch please
14:15 rba timotimo: building, stand by
14:18 jnthn timotimo++
14:19 timotimo i wonder if we should probe for "printf null is dangerous" during configure and make this function cheaper otherwise ...
14:19 jnthn Not really; we only use it in error reporting
14:19 timotimo true
14:20 timotimo and spesh dump :P
14:20 jnthn And the compiler knows its builtins, it can strip the check out anyway
14:20 jnthn Oh, maybe not
14:20 jnthn But still...
14:20 timotimo it's not out of the question that gcc might be smart enough for this
14:20 timotimo MSVC, though ... nope :)
14:21 domidumont joined #moarvm
14:21 timotimo it took me a minute or two to notice sed without /g at the end will only replace stuff once per file :)
14:22 ilmari once per line, no?
14:22 timotimo it seemed to be once per file
14:22 timotimo but it could have been the other thing
14:23 Ven_ joined #moarvm
14:23 rba timotimo++
14:24 rba timotimo: Confirmed. Error not schowing up any more on Solaris build.
14:24 timotimo cool. now to build rakudo
14:24 timotimo not sure how to handle the mismatch of NaN semantics
14:24 rba Still have the pow_n thing...
14:24 timotimo it's practically harmless
14:25 timotimo only for apps that actually do floating point arithmetic and throw nans around
14:25 timotimo there'll be that kind of program, of course
14:25 lizmat .oO( hmmmm , nans )
14:25 timotimo but we can at least ignore that until rakudo successfully compiles
14:25 rba Ok. will go ahaed...
14:29 timotimo oh, compiler warnings about missing casts
14:31 Geth ¦ MoarVM/debug_name_safe_null: 4faa2ba88a | (Timo Paulssen)++ | 3 files
14:31 Geth ¦ MoarVM/debug_name_safe_null: add a few missing casts to MVMObject *
14:31 Geth ¦ MoarVM/debug_name_safe_null: review: https://github.com/MoarVM/MoarVM/commit/4faa2ba88a
14:32 domidumont joined #moarvm
14:32 timotimo no compiler warnings any more
14:34 jnthn .tell samcv To do the final release step (uploading the tarball), commit the .tar.gz into https://github.com/MoarVM/moarvm.org/tree/master/releases (I gave you permissions on it). The change will be pulled within a minute.
14:34 yoleaux jnthn: I'll pass your message to samcv.
14:35 jnthn That also means changes to the website will appear automatically within a minute too :)
14:36 timotimo oh wow i get a crash immediately with this moarvm
14:36 timotimo in nqp's configure.pl
14:42 timotimo oh, got it
14:44 Geth ¦ MoarVM/debug_name_safe_null: cdd0fb45b6 | (Timo Paulssen)++ | src/core/interp.c
14:44 Geth ¦ MoarVM/debug_name_safe_null: setdebugtypename must use debug_name directly
14:44 Geth ¦ MoarVM/debug_name_safe_null: review: https://github.com/MoarVM/MoarVM/commit/cdd0fb45b6
14:57 timotimo rba: without this commit moarvm will most probably immediately crash for any invocation
15:05 Geth ¦ MoarVM: 1a9be0ad48 | (Stefan Seifert)++ | src/core/nativecall.c
15:05 Geth ¦ MoarVM: Revert "JIT compile native calls with at most 1 string argument"
15:05 Geth ¦ MoarVM:
15:05 Geth ¦ MoarVM: This reverts commit 4eadf94599cc021ec7a9e0e49e198f5861468dc1 which is not yet
15:05 Geth ¦ MoarVM: ready for prime time. Seems to cause RT #132328
15:05 Geth ¦ MoarVM: review: https://github.com/MoarVM/MoarVM/commit/1a9be0ad48
15:05 synopsebot RT#132328 [new]: https://rt.perl.org/Ticket/Display.html?id=132328 [SEGV][REGRESSION] DBIish tests are failing spectacularly (JIT compilation of native calls)
15:07 rba timotimo: rakudo (latest git) whent fine, test has small errors: https://gist.github.com/42a73d311c589991c588618d151b1e29
15:16 timotimo oh? solaris doesn't have ssize_t?
15:16 timotimo oh, interesting, that's not a standard type
15:31 timotimo jnthn: i ran spectest and stresstest and nqp test with debug_name_safe_null; i'm not sure how well the calls are covered, but i assume it's safe to merge now
15:32 jnthn timotimo: Yeah, it's probably pretty safe
15:32 jnthn And fixes a blow-up on solaris, so it's nice to have in
15:33 rba timotimo: may I send you a patch for memmem.h? https://gist.github.com/29520a3c47d7b3d1d14c14db415de9c6
15:34 timotimo want to put a mention of solaris into the comment above, too?
15:34 rba timotimo: yes please
15:35 jnthn .tell samcv I'll be around a bit tonight/tomorrow, if you've any questions or need help on the release stuff
15:35 yoleaux jnthn: I'll pass your message to samcv.
15:35 Geth ¦ MoarVM/master: 4 commits pushed by (Timo Paulssen)++
15:35 Geth ¦ MoarVM/master: 51158fa1b3 | don't printf NULL via an unset debug_name
15:35 Geth ¦ MoarVM/master: 4faa2ba88a | add a few missing casts to MVMObject *
15:35 Geth ¦ MoarVM/master: cdd0fb45b6 | setdebugtypename must use debug_name directly
15:35 Geth ¦ MoarVM/master: 8d8a4ce50a | Merge branch 'debug_name_safe_null'
15:35 Geth ¦ MoarVM/master: review: https://github.com/MoarVM/MoarVM/compare/1a9be0ad48...8d8a4ce50a
15:35 jnthn timotimo: You also have access to moarvm.org, fwiw :)
15:35 timotimo rba: i was hoping *you* would add a mention :D
15:35 rba timotimo: will do, give me a sec
15:35 timotimo jnthn: cool :)
15:36 timotimo with my new-found love for javascript, i'll put in some snowflakes if you view the site on christmas or something
15:36 timotimo (surely you can build something like that in pure css)
15:36 jnthn Well, I was more thinking that you can put stuff in the releases/ directory, but sure :P :P
15:36 jnthn Alright, home time :) bbl
15:38 timotimo o/
15:41 rba timotimo: memmem.h patch https://gist.github.com/fc0372dedcc2b3e966dfc882528fb4a3
15:43 timotimo want to tell me what the --author for the commit should be?
16:05 quotable6 joined #moarvm
16:09 xi- joined #moarvm
16:10 robertle joined #moarvm
16:11 domidumont1 joined #moarvm
16:41 domidumont joined #moarvm
16:45 ZofBot joined #moarvm
16:48 Geth ¦ MoarVM: 1952356854 | (Timo Paulssen)++ | src/platform/memmem.h
16:48 Geth ¦ MoarVM: solaris doesn't come with memmem, so use freebsd one.
16:48 Geth ¦ MoarVM:
16:48 Geth ¦ MoarVM: Thanks rba++
16:48 Geth ¦ MoarVM: review: https://github.com/MoarVM/MoarVM/commit/1952356854
16:50 bisectable6 joined #moarvm
16:54 stmuk does moarvm work on solaris?
16:56 timotimo not sure, it does get through nqp's test suite at least, and build rakudo
16:56 timotimo rba asked about how to run spec test but didn't report any results yet
17:07 AlexDaniel` .tell samcv fwiw please take your time. Release happens when it's ready, there's no rush
17:07 yoleaux AlexDaniel`: I'll pass your message to samcv.
17:11 zakharyas joined #moarvm
19:17 zakharyas joined #moarvm
19:26 buggable joined #moarvm
19:46 xi- joined #moarvm
20:07 bartolin wrt failures with nqp::pow_n(1, nqp::nan) on Solaris: we have RT #124147 for similar failures on NetBSD. looks like OpenSolaris and older OpenBSD versions had that problem, too. (just for the records)
20:07 synopsebot RT#124147 [open]: https://rt.perl.org/Ticket/Display.html?id=124147 [MATH] pow edge cases with rakudo-moar on NetBSD
20:18 timotimo bartolin: good catch :)
20:37 samcv AlexDaniel`, thanks!
20:37 yoleaux 14:34Z <jnthn> samcv: To do the final release step (uploading the tarball), commit the .tar.gz into https://github.com/MoarVM/moarvm.org/tree/master/releases (I gave you permissions on it). The change will be pulled within a minute.
20:37 yoleaux 15:35Z <jnthn> samcv: I'll be around a bit tonight/tomorrow, if you've any questions or need help on the release stuff
20:37 yoleaux 17:07Z <AlexDaniel`> samcv: fwiw please take your time. Release happens when it's ready, there's no rush
20:39 samcv gonna eat somethnig then will start doing work towards it
21:42 rba joined #moarvm
22:35 evalable6 joined #moarvm
23:40 evalable6 joined #moarvm

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