need to provide totally offline ./run_tests.sh to support downstream packaging

what about having python-keystoneclient as a git submodule?

i'd be mostly okay with that, but it will be our first submodule stuff in the project

I'm not totally conversant on submodules - and I'm not sure that will actually solve their problem. My impression (not knowledge) is that they're working from a tarball that we generate and packaging with that independent of any network access.

I guess as long as the git submodule work populates data that rolls into that tarball, we'll be fine. If we try and invoke it during the process flow under ./run_test.sh, it's going to present the same issues for them.

Yeah, that would not work. keystone tarball should just contain keystone, not a copy of python-keystoneclient. The need for Ubuntu (and most other distros) is to build a package from a source tarball + other packages pre-installed as build dependencies, and no network access, while running a reasonable amount of tests (the idea being to be able to repeat the build in a controllable environment).

So they either need to:
* be able to run the full tests without network access, using just keystone source tarball and other packages installed as build dependencies (including python-keystoneclient if that's needed)
or
* be able to disable the specific tests that require networking

Fix proposed to branch: master
Review: https://review.openstack.org/5192

Reviewed: https://review.openstack.org/5192
Committed: http://github.com/openstack/keystone/commit/259d9380e835d03d7358e4d953404b8207b8e8d7
Submitter: Jenkins
Branch: master

commit 259d9380e835d03d7358e4d953404b8207b8e8d7
Author: Joe Heck <email address hidden>
Date: Fri Mar 9 22:41:47 2012 -0800

    enables run_test option to skip integration

    * fixes bug 948495

    Change-Id: I274bfe9611d677c44117a0d9ff67394790794fc4