unit test ordering differences in generated XML

I'm starting from scratch per documentation:

    git clone https://git.openstack.org/openstack-infra/jenkins-job-builder
    cd jenkins-job-builder
    python setup.py install
    tox -e doc

So far so good.

    tox -e py27

The tests run, but fail because of several ordering differences in the generated XML, e.g.,

            <skipVote>
              <onSuccessful>true</onSuccessful>
    + <onUnstable>true</onUnstable>
              <onFailed>true</onFailed>
    - <onUnstable>true</onUnstable>
              <onNotBuilt>true</onNotBuilt>
            </skipVote>

and

          <hudson.tasks.Ant>
            <buildFile>build.xml</buildFile>
            <antOpts>-ea -Xmx512m</antOpts>
    - <properties>failonerror=True
    - builddir=/tmp/
    + <properties>builddir=/tmp/
    + failonerror=True
      </properties>
            <targets>debug test install</targets>
            <antName>Standard Ant</antName>
          </hudson.tasks.Ant>

and

            <targets>
              <hudson.plugins.rubyMetrics.rcov.model.MetricTarget>
                <metric>TOTAL_COVERAGE</metric>
                <healthy>80</healthy>
    + <unhealthy>0</unhealthy>
                <unstable>0</unstable>
    - <unhealthy>0</unhealthy>
              </hudson.plugins.rubyMetrics.rcov.model.MetricTarget>
              <hudson.plugins.rubyMetrics.rcov.model.MetricTarget>
                <metric>CODE_COVERAGE</metric>
                <healthy>80</healthy>
    + <unhealthy>0</unhealthy>
                <unstable>0</unstable>
    - <unhealthy>0</unhealthy>
              </hudson.plugins.rubyMetrics.rcov.model.MetricTarget>
            </targets>

This happens on OS X, Python 2.7.5.