Currently some of the tests fail when running against the tupelo webapp server. This exhibits the fact that the tests assume that the context is always completely clean. The failures occur because a previous test left triples which conflict with the next test. In the case of the server, the client is a clean context, but it fronts the server context which is not, nor is there an easy way to clean out all the triples before every test.
This is not a widespread issue and should be readily fixable. This would then permit us to run it against more contexts.