Refactor #132
Updated by bmbouter almost 10 years ago
**The Problem**
Many tests rely on the state leftover from other tests. This causes some tests to pass when run with ./run-tests.py but fail when run individually. All tests should be able to run individually in both platform and all plugins.
%{color:red}This story is expected to be done after #131%
**Which tests need fixing?**
Install "nosepipe":https://pypi.python.org/pypi/nosepipe/ and modify pulp.devel.test_runner.run_tests to include the --with-process-isolation option. Run the tests on platform or your plugins using ./run-tests.py. They will take a long time. Look at the output for the failed tests; those are the ones that are not independent.
**Expected problems**
* There will likely be many non-initialized database errors from test code attempting to use the database but relying on a previous test to configuring it. This can be solved by having run_tests() connect to the database once and for all and removing calls to initialize() throughout the test code. A consistent base class used by all unit tests is also an option.
.