these are the first simple fixes for some of the current Buildbot problems (issue #2872). Some more updates/pull requests may follow, but I'd like to give it a shot now and see which fails are left for the next round.
I just gave this a second look. My first assumption was that the TestSCons.machine var was used in some other tests at well, but it isn't.
We could probably move the code to the test framework directly, as you proposed.
There is still one glitch though: RPM uses the "hardware platform" for naming its files, while uname() returns the "machine hardware name". Any ideas how to get at the former one?
I left the machine attribute where it is now, and instead added two new functions to TestCmd---can't be bad---that help in keeping the test for the created RPM files more general.
Hope that's okay with everyone out there.
I also corrected the Java version support and upgraded it to 1.7, now all throughout JavaCommon.py.