Perl 6 - the future is here, just unevenly distributed

IRC log for #native, 2016-05-04

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

All times shown according to UTC.

Time Nick Message
13:38 plicease I am going to start working on Alien::Builder again next week sometime.
13:38 plicease Probably do some initial releases to CPAN.
13:38 plicease between now and then is a good time to comment on
13:38 plicease https://github.com/Perl5-Alien/Alien-Base/issues/157
13:38 plicease if you have any opinions.
13:39 plicease Also planning on creating Alien::MSYS2 and consider using it as the default for windows instead of Alien::MSYS for Alien::Builder.
13:43 plicease sivoais: any thought you have on msys vs msys2 are welcome.  I think since there is a 64bit version of msys2 there may be a slight advantage there.
13:44 plicease even though you can build 64bit using the strawberry 64 gcc and the 32 bit msys
14:12 jberger plicease: I wonder if you might consider targeting leont's Dist::Builder that is nearly ready
14:48 plicease where can I find it?
14:48 plicease MM is in core which is an advnatage
14:48 plicease and the idea of A::Builder is that it can be retargeted pretty easily.
14:48 plicease I'm happy to target both actually that would prove the concept
14:58 jberger https://github.com/Leont/dist-build
15:00 plicease thanks.
16:15 sivoais I really do like MSYS2 due to the availability of particularly hard to install packages as binaries. And it handles all the deps further down.
16:16 sivoais And with that in mind, for an Alien::MSYS2, it would also be good for modules that use it to be able to check/request that particular packages be installed.
16:17 plicease that would be useful but a bit tricky.
16:18 plicease if we are using the "system" MSYS2 then philosophy / policy is not to install packages
16:19 sivoais And we would always need the path to MSYS2 set up in order to link dynamically
16:19 plicease if we are using the share MSYS2 then we'd need to update a msys2 root inside a share directory after Alien::MSYS2 has been installed.
16:20 plicease checking for an existing package is np though.
16:20 plicease maybe we stick to that?
16:21 sivoais Yeah, that seems reasonable.
16:34 plicease sivoais: do you know if you can move a MSYS2 root?
16:37 sivoais hmm, I have not tried that. I suppose I could create a branch of appveyor test to try how well it works.
16:37 sivoais according to <https://sourceforge.net/p/msys2/mailman/message/34292001/>, it is relocatable with some caveats
16:38 sivoais for packages that need absolute paths, it seems
16:40 sivoais Useful thread since it talks about how to build an installer using MSYS2 packages.
16:40 plicease thanks.
16:41 plicease that is pretty much the same for us.
19:37 plicease sivoais: I think the trick to probing for MSYS2 is going to have PkgConfig.pm hook into Alien::MSYS2 once it gets off the ground.
19:38 plicease I think you might have even asked for that at one point but I wasn't sure how to do it.
19:38 plicease But I think I am closer.
19:47 sivoais Ah, yes, I had tried to do something with linking Strawberry Perl's pkg-config to the PKG_CONFIG_PATH for MSYS2. I didn't continue down that line because I got hit with some static linking issues.
19:48 sivoais But I've figured the linking issues now that I have things working under Appveyor.
20:13 plicease I think part of the problem is msys2 uses /msys64 as the root in the .pc files
20:13 plicease pkg-config that comes with MSYS2 knows how to translate that.
20:14 plicease but PkgConfig.pm (strawberry's pkg-config) doesn't yet.
20:14 plicease but now that I am writing some code to probe for MSYS2 it should be easy to get PkgConfig.pm to take advantage of that.
21:02 sivoais Oh, I changed that by adding --define-variable=prefix=/my/prefix/path . Perhaps using that along with cygpath.exe ?
21:03 sivoais Some .pc files might not use prefix= , but those are arguably broken? I'm not sure if that is a defined standard.
21:49 plicease I think most .pc files do not use prefix=
21:49 plicease but anyway, when I get to it I think I can figure it out.
21:52 jberger btw, I'm 100% ok throwing out my half-arsed pkgconfig thingy in AB
21:52 jberger I think I've said that before
21:53 jberger but making sure it is remembered
21:54 plicease I believe the issues we are discussing are at the PkgConfig.pm level not Alien::Base::PkgConfig.
21:54 plicease If the latter starts to cause headaches I will keep that in mind :)
22:26 sivoais I wrote up the steps I took to get the Appveyor build on <https://github.com/project-renard/curie/pull/86> working at <https://project-renard.github.io/doc/development/meeting-log/posts/2016/05/03/windows-build-with-msys2/>
22:27 sivoais Going to also share on #toolchain, since there was some EUMM hackery involved and I'd like to have a better solution.
23:48 preaction joined #native

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