IRC channel logs


back to list of logs

<mark_weaver>jxself: are you working on updating linux-libre to 3.17.3, or should I?
<mark_weaver>jxself: I'll work on it. doing a test build now.
<jxself>mark_weaver: Yep, I build them for my repository first. Updates are typically 24 hours after the release of a new version.
<jxself>This allow for compiling, testing, making configuration changes in the rare cases it's needed, etc. :)
<jxself>"Officially", though I only commit to "within a few days" so that some slippage is allowed for in cases where it may be needed. :)
<mark_weaver>jxself: if it builds successfully on my i686 system, should I go ahead and commit the update, or should I wait?
<jxself>I did it just a few mimutes ago. It succesfully built in my 32 & 64-bit systems.
<mark_weaver>ah, good, thanks!
<jmd>finally getting some success with an arm build.
<iyzsong`>packaging wine, should I add it to games.scm? :)
<DusXMT>I don't know, wine doesn't have to be used explicitly for games
<DusXMT>For example, I'm building gtk+ with it and mingw atm
<iyzsong`>wow, I just use wine for gaming.
<iyzsong`>then, I'll just add wine.scm.
<DusXMT>Also, there are many non-game programs for wine. For example, I recall using WinImage when I was still new to GNU/Linux, which ran in wine
<DusXMT>Yup, I think that's reasonable
<iyzsong`>I agree :)
<roptat_>there's a package I compiled that comes with its own custom (free) license. How do I write this in the recipe?
<jmd>roptat_: I think there is a fsf-free procedure
<jmd>roptat_: Look how it is done in texlive
<roptat_>thanks, it worked
<jmd>I don't seem to be able to cross build gcc-toolchain:
<jmd>In guix/build-system/gnu.scm:
<jmd> 392: 0 [gnu-cross-build #<build-daemon 256.14 22ab9c0> "gcc-4.9.2" ...]
<jmd>guix/build-system/gnu.scm:392:0: In procedure gnu-cross-build:
<jmd>guix/build-system/gnu.scm:392:0: In procedure #<procedure gnu-cross-build (store name #:key target native-drvs target-drvs guile source outputs search-paths native-search-paths configure-flags make-flags out-of-source? tests? test-target parallel-build? parallel-tests? patch-shebangs? strip-binaries? strip-flags strip-directories phases system imported-modules modules substitutable? allowed-references)>: Odd length of keyword argument
<jmd>It would be nice to have an environment variable like GUIX_FLAGS to save be repeatedly (mis)typing --target=armel-linux-gnueabihf
<davexunit>hmm, yeah that might be a good idea.
*davexunit submits a patch to update ruby
<jmd>Do we run tests for ruby?
<jmd>Hmm. How much patching was necessary?
<davexunit>none. I just submitted a patch for guix to bump the version number.
<jmd>I looked at packaging ruby some months ago.
<davexunit>ruby 2.1.5 was released, so I updated the guix package definition.
<davexunit>pjotrp and I packaged it.
<davexunit>during the hackathon.
<jmd>They had some really stupid tests which had no chance of passing.
<davexunit>check it out in gnu/packages/ruby.scm
<davexunit>I was having trouble too, but pjotrp figure it out
<jmd>Maybe they changed some things.
<jmd>There used to be a test which failed if $HOME did not match that in /etc/passwd.
<jmd>(God knows what the value of that test was)
*davexunit returns to the minetest patch set
<davexunit>should be ready to go
<jmd>If we have ruby now, the libwebkit might be worth trying to package ?
<davexunit>we need to package some of the fundamental gems that most gems depend on.
<davexunit>building ruby gems is sort of a nightmare, though.
<davexunit>haven't figured out a great strategy for that yet
<davexunit>nix's way is no good
<jmd>Frankly I don't see the point of ruby.
<jmd>Its re-inventing the wheel.
<jmd>But they invented a triangular one.
<viric>is there anything good written in ruby?
<jmd>I doubt it.
<ijp>jmd: triangular wheels can work (on a suitable road)
<jxself>I imagine it'd be a bumpy ride.
<ijp>iirc square wheels need a sequence of catenaries
<viric>I doubt such a suitable road can exist
<davexunit>jmd: ruby is actually a pretty nice language
<davexunit>as far as mainstream languages go, it's one of my favorites
<davexunit>I can program in a somewhat functional way, the OOP implementation is pretty good, and there's decent metaprogramming
<davexunit>it's most direct competitor is python, I think, and I'd choose ruby over python any day.
<jmd>Is there a installable image for i386 ?
<davexunit>I think you have to build your own
<roptat_>what is propagated-inputs?
<jmd>roptat_: If A is a propagated-input of B, then if C has B as an input, it implicitly has A as an input too.
<DusXMT>Aren't all inputs propagated by default?
<DusXMT>For example, shared library A will always require shared library X to function properly, and so anything that links against A will link against X, and most libraries are like this, or not?
<jmd>It will link against it. But that library won't be available for any other library to link against, unless it is explicitly listed as an input or propagated by another input.
<roptat_>I have written a recipe for claws-mail that depends on libetpan. Libetpan depends on openssl and cyrus-sasl. libetpan-config --libs says -letpan -lssl -lcrypto -pthread -lsasl2, so I had to use them in the recipe of claws-mail, because it was not found during build
<roptat_>So now I use the propagated-inputs :)
<DusXMT>What will happen if a test suite is run, and the test programs link against A (therefore also against X), does one need to make X a propagated input for that test to not fail?
***v45h` is now known as v45h
***jxself_ is now known as jxself