Perl 6 - the future is here, just unevenly distributed

IRC log for #native, 2017-01-10

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

All times shown according to UTC.

Time Nick Message
00:45 plicease willthechill: I have used appveyor but I get a lot of false positives and so haven't been using it actively lately.
03:14 dcmertens Whoah, what's the difference between matrix.cpantesters.org and fast-matrix.cpantesters.org?
03:14 dcmertens I've neve seen fast-matrix
04:25 willthechill joined #native
06:42 preaction dcmertens: fast-matrix uses the tail log from metabase.cpantesters.org, so it's basically a quick view of the latest data, but not the whole data
06:43 preaction those sites both exist because no work was being done on cpantesters, which is no longer the case, and i suspect they will end up being replaced by the work we're doing (when we get there)
06:43 preaction well, probably not. but i do have to make a PR to move them to a new, faster, less error-prone API
14:20 awwaiid joined #native
15:53 dcmertens joined #native
17:11 willthechill plicease: okay my question about appveyor, do you know of any way to make appveyor act lilke travis by running both shared and system tests, each in a new VM like travis?
17:24 plicease willthechill: not sure.  If nothing else you can run the build twice in the same run.
17:25 jberger I asked the google
17:25 jberger https://www.appveyor.com/docs/build-configuration/#build-matrix
17:27 willthechill jberger++
17:28 willthechill plicease: next question, astyle has a different source tarball and make instructions for Windows, how can I configure Alien to change it up based on OS detected?
17:51 plicease Just pass in a different values to AB::MB->new depending on the value of $^O
17:52 plicease the build from source works with the linux build instructions though
17:52 plicease i doubt you'll get the official windows builds to work in most cpantester boxes though
17:53 plicease the makefiles are for Visual C++ and the majorty of cpantesters are using strawberry which is based on gcc
17:54 plicease This PR is what fixed windows build for me: https://github.com/wbraswell/alien-astyle/pull/3
17:54 willthechill ah yes okay, so basically I can skip the windows-specific tarball because very few people will ever use VC++ (except for bulkk88)
17:55 willthechill okay great, I'll set up appveyor and we'll see what happens!  :-)
17:55 plicease i am not sure how many people do a VC++ build aside from bulk88
17:55 plicease but yeah my experince is that there are no cpantesters boxes using VC++ perl
17:55 willthechill *nods* yeah I think you're right
17:57 plicease if you do eventually want to support it you can check $Config{cc} i forget off the top of my hand what the name of the compiler is
17:58 plicease but that is how the toolchain differentiates between gcc and VC++ from memory.
17:59 plicease Actually I've done it before
17:59 plicease https://metacpan.org/source/PLICEASE/Alien-FFI-0.14/inc/My/ModuleBuild.pm#L12
18:00 plicease $Config{ccname} =~ /cl(.exe)?$/i is probably better since windows is case sensitive.
18:02 plicease actually i think it may be fine as is.  I am overthinking it.  i think.
18:02 plicease patches welcome :P
18:02 kaniini wot
18:02 kaniini windows is case insensitive
18:02 willthechill well I'd have to team up w/ bulk88 to even test it, so I'll just focus on getting 100% passing on appveyor & CPAN testers first  :-)
18:03 plicease kaniini: correct i typed that wrongly.
18:07 plicease I have a build of VC++ Perl on my PC actually it isn't that hard to do.  A few key things from CPAN don't work so well though becuse it isn't tested as well.
18:09 willthechill nice
18:09 willthechill I only have Linux so I rely on travis, appveyor, and CPAN testers  :-P
19:18 willthechill plicease: a bug in Alien::Base::ModuleBuild?    http://www.cpantesters.org/cpan/report/da23f76a-d717-11e6-9aca-9a525a7089b8
19:18 willthechill (that's the only failure so far for A::astyle v0.004)
19:19 willthechill jberger: same question as above, a bug w/ A::B::MB?
19:29 plicease more likely a solaris weirdism.
19:30 willthechill okay I will try to contact that tester
19:31 willthechill does anybody know Serguei Trouchelle?
19:33 willthechill AKA "stro"
20:12 jberger there are ways to find that out via cpantesters website
20:12 willthechill I sent stro an e-mail
20:12 willthechill ether said he may be in #perlfr but he wasn't
20:13 willthechill I also filed stro's Solaris failure as an A::astyle issue:  https://github.com/wbraswell/alien-astyle/issues/6
20:15 plicease I've debugged a lot of Solaris bugs with a OpenIndiana VM
20:15 plicease Sadly that VM is no more.
20:16 plicease https://www.openindiana.org/
20:16 willthechill that stinks
20:16 plicease hard drives fail sometimes.
20:47 willthechill plicease & jberger: okay I'm doing something wrong with my Build.PL I think, because I have to manually install Alien::Base::ModuleBuild or else we experience a failure, as I am now seeing in AppVeyor because I did not manually install A::B::MB via appveyor.yml
20:47 willthechill perl Build.PL     Can't locate Alien/Base/ModuleBuild.pm in @INC
20:47 willthechill it works on Travis because I made .travis.yml install A::B::MB before everything else
20:48 willthechill why doesn't `cpanm --installdeps .` install A::B::MB?
20:49 willthechill I already have A::B::MB in configure_requires of Build.PL    https://github.com/wbraswell/alien-astyle/blob/master/Build.PL
20:49 willthechill should A::B::MB also be in Build.PL's requires section as well a in configure_requires?
20:49 willthechill (that's just my random guess)
20:53 jberger I've had to play around with some of this before, I forget how it all works
20:53 willthechill okay well it does NOT work to add A::B::MB to requires
20:54 willthechill I'm trying to figure it out...  :-/
20:54 jberger no, that would happen too late
20:54 jberger you need to make it install before configure
20:54 jberger maybe its a cpanm flag?
20:54 willthechill yes it hits Build.PL, which its inc/My/ModuleBuild.pm, which hits A::B::MB and fails
20:55 jberger https://github.com/jberger/Mojo-Phantom/blob/master/.travis.yml
20:55 jberger that's a slightly different case, but I'm guessing your answer is contained in there
20:55 willthechill --with-develop
20:55 willthechill ?
20:56 jberger that's my guess
20:56 willthechill "EXPERIMENTAL: Installs develop phase dependencies"
20:56 jberger hmmm, maybe not
20:57 jberger oh, right that was to install the mb-tiny binary
20:57 jberger hmmmm, then I have no idea
20:57 willthechill okay
20:59 willthechill since this only seems to affect AppVeyor & Travis, and I can overcome it w/ the yml config files, then I will just do so and not consider it to be a "bug" for now
21:07 willthechill plicease & jberger: okay now Alien::gmake is failing to install on AppVeyor...      In file included from subproc/sub_proc.c:31:0:  ../makeint.h:453:18: fatal error: glob.h: No such file or directory
21:07 willthechill https://ci.appveyor.com/project/wbraswell/alien-astyle/build/job/b8551sn9be4mqi6j
21:08 jberger willthechill: I can't step through this atm
21:08 jberger at work
21:08 willthechill okay thanks
21:10 willthechill so this file appears to be from libc6-dev
21:10 willthechill glob.h
22:48 kaniini glob.h is part of POSIX, windows isn't POSIX
22:49 kaniini gmake may require patches on windows

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