Perl 6 language and compiler development | Logs at colabti.org/irclogger/irclogger_log/perl6-dev | For toolchain/installation stuff see #perl6-toolchain | For MoarVM see #moarvm
Set by Zoffix on 27 July 2018.
00:31 b2gills left
Geth rakudo: vrurg++ created pull request #3076:
Better support for EXPORT with Stash
00:54
00:59 ufobat__ joined 01:02 ufobat_ left 01:09 b2gills joined 02:30 lucasb left 04:38 pamplemousse joined 07:07 pamplemousse left 07:37 patrickb joined 07:42 ufobat__ left
lizmat Files=1268, Tests=108561, 191 wallclock secs (26.62 usr 7.76 sys + 2685.23 cusr 255.49 csys = 2975.10 CPU) 08:00
08:15 nebuchadnezzar left, nebuchadnezzar joined 09:03 AlexDaniel left 10:31 Kaiepi left
kawaii any word from the star team about their next release? 10:41
nine Looks like there's a segfault in threaded code using assign meta ops in github.com/rakudo/rakudo/blob/mast....pm6#L1688 11:55
github.com/rakudo/rakudo/blob/mast....pm6#L1660 could use some protection from concurrent access
timotimo oh snap 11:58
perhaps atomic ops would be good here
since atomic ops excell when contention is very low, i think? 11:59
nine Shouldn't the bind to $METAOP_ASSIGN be atomic already? Then all it'd need would be some temporary variable in INSTALL-CORE-METAOPS for setting up and do the bind in the last step 12:01
timotimo perhaps that'd be enough 12:02
nine It's hard to say with such things if it's really fixed, but at least the likelyhood is greatly reduced 12:15
timotimo is the point release already done? %) 12:16
i don't think it is. this fix should perhaps go in as well :)
Geth rakudo: 60cfbb39e9 | (Stefan Seifert)++ | src/core/Rakudo/Internals.pm6
Fix (or greatly reduce chance of) segfault when using meta ops in threaded code

Different threads trying to set up $METAOP_ASSIGN at the same time and binding to its keys caused a segfault. Fix by changing it to set up a hash in a local variable and only binding that to the shared variable as the last step.
12:17
timotimo .o( find more instances of this pattern across the codebase? ) 12:18
12:21 pamplemousse joined 12:35 AlexDaniel joined
lizmat re 60cfbb39e9 I wonder why that isn't done at compile time 12:38
ah, the comment above it explains why: 12:42
# Method for lazily installing fast versions of METAOP_ASSIGN ops for
# core infix ops. Since the compilation of &[op] happens at build time
# of the setting, we're sure we're referring to the core ops and not one
# that has been locally installed. Called by METAOP_ASSIGN. Please add
# any other core ops that seem to be necessary.
pamplemousse weekly: yakshavingcream.blogspot.com/2019/...oh-my.html
notable6 pamplemousse, Noted!
lizmat I was wondering why "make spectest" was doing so well the past days, and it turns out that somehow, Inline::Perl5 is not tested anymore 13:05
even though I have Inline::Perl5 installed
13:09 Kaiepi joined
|Tux| Rakudo version 2019.07-86-g60cfbb39e - MoarVM version 2019.07-17-ga7563e71b
csv-ip5xs0.668 - 0.726
csv-ip5xs-205.491 - 5.496
csv-parser22.578 - 23.242
csv-test-xs-200.431 - 0.440
test6.849 - 7.186
test-t1.686 - 1.692
test-t --race0.785 - 0.788
test-t-2027.915 - 28.860
test-t-20 --race9.021 - 9.031
13:18
lizmat notable6: weekly 13:28
notable6 lizmat, 16 notes: gist.github.com/8569dbdaa4a4f5f167...2546eef98a
AlexDaniel weekly: 2019.07 is out! We will likely see 2019.07.1 this week with some minor fixes. 13:29
notable6 AlexDaniel, Noted!
AlexDaniel weekly:
notable6 AlexDaniel, 17 notes: gist.github.com/b8a66e506975ce4d01...20f35a8c1e
AlexDaniel lizmat: :)
13:32 pamplemousse left
lizmat hmmm... looking at Changelog: ReifiedListIterator is now a PredictiveIterator [9107215] 13:34
that one got reverted
nine *sigh* still more packaging issues with rakudo 2019.07: 13:42
[ 49s] + perl6 /usr/share/perl6/bin/install-perl6-dist --to=/home/abuild/rpmbuild/BUILDROOT/perl6-OO-Monitors-1.1-1.130.x86_64/usr/share/perl6/vendor --for=vendor --from=.
[ 49s] Unhandled exception: While looking for 'ModuleLoader.moarvm': no such file or directory
[ 49s] at <unknown>:1 (/usr/share/perl6/runtime/perl6.moarvm:<dependencies+deserialize>)
patrickb: looks like a relocatability issue ^^^ 13:43
patrickb nine: I'll have time in the evening to look deeper. Which perl6 executable does this use? The installed one or the one in the build dir? 13:45
nine The installed one
From this package: build.opensuse.org/package/show/de...rl6/rakudo
nine@sunshine:~> /usr/bin/perl6 -e '' 13:47
Unhandled exception: While looking for 'ModuleLoader.moarvm': no such file or directory
patrickb Is the install-perl6-dist script available somewhere?
Ah! Unrelated to anything else there! 13:48
This is a normal build installed to /usr and that's it?
nine The file is /usr/share/perl6/lib/Perl6/ModuleLoader.moarvm but rakudo looks for it in /usr/share/nqp/lib/ModuleLoader.moarvm, /usr/share/perl6/lib/ModuleLoader.moarvm, /usr/share/perl6/runtime/ModuleLoader.moarvm and ModuleLoader.moarvm (!) 13:49
patrickb: the spec file tells the story: build.opensuse.org/package/view_fi...c?expand=1
In essence, it's perl Configure.pl --prefix=/usr && make && make install
lizmat reportable6: 2019-07-15T00:00:00 2019-07-22T00:00:00Z 13:53
reportable6 lizmat, OK, working on it! This may take up to 40 seconds
lizmat, gist.github.com/3471b932ac770917e5...3b3fbff11d 13:54
patrickb nine: Thanks for the info. I'll see what I can do. 14:03
nine: Are you sure it's missing the Perl 6 module loader and not the NQP one? 14:17
I think the problem actually is, that the nqp build installed the build files into /usr/lib64 instead of into /usr/share/nqp/lib 14:18
That's where the rakudo build expects the files. 14:19
Do you know how the files ended up in /usr/lib64/nqp?
lizmat notable6: weekly reset 14:21
notable6 lizmat, Moved existing notes to “weekly_2019-07-22T14:21:03Z”
patrickb nine: I think it's the --libdir option passed to the NQP build. 14:28
nine patrickb: checking... 14:48
patrickb nine: I seem to recall we talked about this effect of SuSE putting stuff in different folders some time ago. 14:50
nine: gist.github.com/patzim/710baab43da...s-approach 14:51
nine patrickb: yep, removing --libdir seems to fix the issue 14:53
Or rather work around it. An option that leads to broken software when used is clearly still an issue that needs a real fix 14:54
patrickb nine: We did talk about this specific problem. I'm not entirely sure anymore what the result was. Whether we said we don't want to support separate lib dirs or thought of a soution? I'll see if I can find the conversation again... 14:55
lizmat and another Perl 6 Weekly hits the Net: p6weekly.wordpress.com/2019/07/22/...-released/ 15:00
15:55 patrickb left 16:10 patrickb joined 16:14 chloekek joined
patrickb nine: I didn't find much except for: colabti.org/irclogger/irclogger_lo...-01-20#l33 16:39
nine: I think there is some more conceptual work necessary to solve the problem of distros sometimes changing the directory layout of the installation. 16:43
The way it currently is, rakudo and nqp provide no means to have a different directory layout in PERL6_HOME (currently placed into share/perl6) or NQP_HOME (currently placed into share/nqp). Those folders could be put somewhere else though. 16:45
The most important question in that regard: Is the general idea of a single folder where all stuff NQP and all stuff perl6 is in fitting or not? 16:47
If that design decision turns out to be bad, it's back to the drawing board... 16:48
16:52 pamplemousse joined
masak lizmat++ # weekly 16:56
17:27 patrickb left 17:28 robertle joined 17:30 patrickb joined
Kaiepi are pullreqs that could possibly break stuff going to have to wait for 2019.07.1 before they can be merged? 17:53
AlexDaniel no 17:54
patrickb Kaiepi: No. Actually quite a lot of explosivey stuff is already in master...
AlexDaniel list of fixes that will be in 2019.07.1: github.com/rakudo/rakudo/issues/3067 17:55
Kaiepi perfect i can get underway with my grant finally 17:56
timotimo AlexDaniel: we just got another interesting commit in moarvm
in a PR
AlexDaniel this one? github.com/MoarVM/MoarVM/pull/1148 17:57
timotimo yep 17:58
without it, the debug server cannot be used on windows
AlexDaniel looks harmless to me
.oO( as long as you ifdef stuff to _WIN32, you can do whatever you want :D )
timotimo i barely know how WSAStartup exactly works, but it's safe to call multiple times, so that's good 17:59
and libuv calls it like that, too
AlexDaniel added it to the list 18:00
thanks
Kaiepi can someone take a look at github.com/rakudo/rakudo/pull/3008, github.com/perl6/nqp/pull/558, and github.com/MoarVM/MoarVM/pull/1126 ? 18:01
may or may not be worth including in the 2019.07.1 since it fixes ipv6 IO::Socket::INET sockets on freebsd
AlexDaniel that I'm not that comfortable with 18:02
can be merged now but it will only be part of 2019.09 I'm afraid 18:03
Kaiepi aight
ugexe does our SyncSocket do that? 18:09
since that doesn't use libuv it wouldnt have the WSAStartup thing
also i think doing this breaks when booting to windows safe mode? 18:10
or windows with no network support or whatever
(im recalling that from libuv somewhere...) 18:11
Kaiepi: github.com/libuv/libuv/commit/af31...b9711ee019 18:13
might be worth checking to see if we need something like that too
Kaiepi i'll check if it is needed after the next thing i plan on doing for my grant 18:15
ugexe Kaiepi: oops you weren't the one who did github.com/MoarVM/MoarVM/pull/1148 18:16
Kaiepi oh 18:17
next thing i plan on doing is making it so moarvm tracks whether or not ipv6 actually works so resolving hostnames with AF_UNSPEC doesn't panic if the first result is ipv6 on systems where it's enabled but misconfigured 18:19
Geth rakudo/release-2019.07.1: 540926bf66 | (Patrick Böker)++ (committed by Aleks-Daniel Jakimenko-Aleksejev) | Configure.pl
Don't check for submodules in a release

Releases already have all the necessary submodules included. Also we don't require git to be installed to build releases. Thus don't do any submodule checks.
18:29
rakudo/release-2019.07.1: f41db0445c | (Patrick Böker)++ (committed by Aleks-Daniel Jakimenko-Aleksejev) | tools/lib/NQP/Config/Rakudo.pm
Don't use add any RPATH compile flags when installing to /usr

Do it just the way MoarVM also does it. The RPATH is not necessary in that case, as the libs will be in the library search anyways and having a
  rogue RPATH is bad style and trips up some packaging tools.
nqp/release-2019.07.1: 68e1eabafd | (Patrick Böker)++ (committed by Aleks-Daniel Jakimenko-Aleksejev) | Configure.pl
Don't check for submodules in a release

Releases already have all the necessary submodules included. Also we don't require git to be installed to build releases. Thus don't do any submodule checks.
18:31
20:21 TimToady left 20:25 pamplemousse left 20:35 TimToady joined
TimToady join moarvm 20:36
jnthn I did, before anyone :P
TimToady had the power company swapping his electric meter, and it broke my / key 20:38
samcv AlexDaniel, i have 2019.07.1 staged for release if we don't need anything other than the commits on the ticket 20:43
20:50 pamplemousse_ joined
Geth nqp: patzim++ created pull request #566:
Fix return code of nqp build runner
20:57
rakudo: patzim++ created pull request #3078:
Fix return code of perl6 build runners
20:58
21:15 robertle left 21:17 lucasb joined 21:21 AlexDaniel left 21:22 pamplemousse_ left 21:26 AlexDaniel joined
AlexDaniel timotimo: I guess this needs to redirect too now? web.archive.org/web/20190111152104...perl6.org/ 21:27
I don't quite remember what was there, but debian checks this url: rakudo.perl6.org/downloads/rakudo/ 21:28
adding a redirect won't unbreak that, but it's better than 404
22:08 patrickb left 22:25 pamplemousse_ joined 22:27 patrickb joined 22:28 chloekek left 23:27 patrickb left