On Thu, Jun 16, 2016 at 7:40 PM Mat Booth <fedora@matbooth.co.uk> wrote:
On 16 June 2016 at 20:16, Christopher <ctubbsii@fedoraproject.org> wrote:
I'm trying to revive rpms/hadoop, but am running into some failures[1] on i686 in koji that I cannot reproduce locally in x86_64. The thing is... I don't really have any good 32-bit environment to test this locally. Hadoop isn't really suitable for 32-bit architectures (IMO) and I doubt it's a well-tested upstream arch.

Any suggestions for how I should proceed? Is it really necessary that *all* packages support i686 arch, even when it doesn't make sense for the application's users?

[1]: http://koji.fedoraproject.org/koji/taskinfo?taskID=14521689



Actually nothing to do with arch, all arches would fail, it just failed on i686 first and koji cancelled the others ;-)

The maven-jar-plugin was updated to new version in Fedora, which has this new behaviour, but the error message should be self-explanatory: "You have to use a classifier to attach supplemental artifacts to the project instead of replacing them." It is due to the jar:jar goal being executed twice for the same artifact for some reason. Try eliminating the second invocation from the pom file of the broken module.


Yeah, I saw that error message and behavior, but I thought I might be able to rule out the possibility that it was a simple maven problem, since it built perfectly fine locally in mock. I figured it might be a problem with maven executing different tasks under different arch's, but didn't have a way to easily test a 32-bit arch locally to reproduce.

I'll try playing with fedpkg mock-config to see if I can do a mock build with i686 arch locally.