versioning scheme for tests

[miabbott]

We have encountered situations where users have run the tests from git master against older releases and the tests do not run successfully. This is usually because the tests assume certain packages will be installed or certain files will be present.

This seems to push us in the direction of needing tags/releases for our tests.

The big question is how to setup a versioning scheme that works for the multiple streams we claim support for.

Do we use time based releases? A monthly release would let us claim support for the AH releases that landed during that month.

Or tag according to well defined AH releases? Tagging a release is relatively inexpensive, so we could create tags for each RHELAH release or each Fedora AH release.

Looking for ideas and input, please feel free to comment.