Fire is step THREE! | github.com/perl6/toolchain-bikeshed | Channel logs: irclog.perlgeek.de/perl6-toolchain/today | useful prior art: metacpan.org/pod/CPAN::Meta::Spec
Set by moderator on 10 September 2017.
01:55 ilbot3 joined
moderator Fire is step THREE! | github.com/perl6/toolchain-bikeshed | Channel logs: irclog.perlgeek.de/perl6-toolchain/today | useful prior art: metacpan.org/pod/CPAN::Meta::Spec
05:22 domidumont joined 05:27 domidumont joined 06:21 domidumont joined 06:29 domidumont joined 08:08 tony-o joined 08:33 leont joined 10:35 jonathon joined 10:53 llfourn joined 11:47 awwaiid joined
jonathon Hi! I've got a "weird" issue with `zef install Linenoise`; custom zef deb package; zef appears to be picking up the flags moarvm was compiled with; bpaste.net/show/cecf8244d91e 13:10
just need to know if it should be doing that so i need to add the dependencies to the package, or if it shouldn't be doing that
(lines 32, 35, 36 are the key ones) 13:11
/usr/bin/ld: cannot find -latomic_ops; /usr/bin/ld: cannot find -luv; 13:12
packages here: launchpad.net/~jonathonf/+archive/ubuntu/perl6 13:16
leont I don't know much about how moarvm extensions work, but I don't think they should need those flags on Linux (they would on Windows) 13:38
nine Well there are no real MoarVM extensions. LibraryMake just picks the compiler and linker flags used to compile the VM and replaces variables in Makefile templates. Clearly a broken assumption. 13:42
jonathon Is it an issue/bug with LibraryMake or intended behaviour? If it's intended behaviour, I'll just add the library packages as dependencies for my zef package, then it should be happy 13:46
Sounds like intended behaviour: "This is effectively a small configure script for a Makefile. It will allow you to use the same tools to build your native code that were used to build perl6 itself." 13:47
nine jonathon: I see LibraryMake as a stopgap measure until we have a better auto configuration system. And one that has mostly outlived its usefulness. 13:52
leont Yeah, that sounds rather suboptimal
.oO(so many things to do, so little time)
13:53
14:08 timotimo joined
timotimo heyo 14:08
how should we handle brian d foy shipping a Test.pm6 in one of his repos and zef suggesting you install that if rakudo for some reason fails to find/load the Test.pm6 we ship? 14:09
we can potentially give the fudger the task to put a :auth<rakudo> to every "use Test" in the spec test files
but then literally every file will say "fudged" at the end unless we special case it 14:10
which of course we could
nine Well if we mean a certain Test.pm6, we should really say so 14:11
At the same time, why does he have a Test.pm6 in a repo?
timotimo yeah, but we shouldn't put "use Test:auth<rakudo>" into the spec tests :)
it's a fork of rakudo's Test.pm6 because rakudo's has some bug or something 14:12
it would have been lovely if there were a link to a ticket from there
jonathon why is zef picking up two versions of the same module?
nine And he has that in a public repo and provides that? A bit antisocial I dare say
jonathon (e.g. why does it suggest a non-canonical module?) 14:13
nine jonathon: there are no canonical modules
timotimo: does roast work with his Test.pm6?
timotimo i haven't tried it yet 14:14
i should probably pull it and get a diff
nine Because if it does work, there's nothing to fix
timotimo yesterday there was someone having a problem where Test.pm6 was not found and zef suggested to install PrettyDump to get it 14:16
brians test.pm6 changes how the output io variables are set up and used 14:19
among other things it decides to output failure stuff to OUT instead of ERR 14:20
lizmat has commented on github.com/briandfoy/perl6-PrettyD...1a832a5e1b 14:23
nine Okay, if it's incompatible (and timotimo's description sounds like), it's definitely antisocial to just add it to the ecosystem. 14:26
In his defense, we obviously as a community need to find some rules and procedures for dealing with such situations.
We replaced technical barriers by the need for arbitration and haven't followed up on that so far. 14:27
timotimo not necessarily incompatible 14:28
14:40 domidumont joined 14:52 domidumont joined
timotimo the Test.pm from that repo only fails tests in S24-testing 15:03
3-output, 12-subset-todo, 8-die_on_fail, and line_numbers
leont TBH, the behavior of Test.pm6 in that regard is rather unfortunate 16:22
In subtests is plain dumb, in normal tests is often far to load
*loud
16:25 leont left 16:26 leont joined
leont I looked into changing its behavior, but hit the failing spectests -_- 16:27
IMNSHO, the test is way to specific on where what output goes, it's the oks/not oks that should matter not *comments*
ugexe those are the types of changes that should be ok to make to the roast 16:29
leont What brian really should have done though is «use lib "t/lib";», just as in perl5. Then again in perl5 he also likes doing toolchainy things in his own idiomatic way. 16:30
16:36 domidumont joined 19:20 hoelzro joined 20:07 perlpilot joined 20:19 Brock joined 20:31 JimmyZ joined 21:18 ugexe joined 21:28 perlpilot_ joined