The Behave testing framework could be a better way to express tests in a human readable format. This could allow end users and developers who have never written tests to describe the tests they want to do in a language that is both human readable and machine executable.

We need a method for capturing the screen shots of failed selenium tests. At the moment we can do this but the screen shot are captured by a private web site. (See the 'after_failure' step at https://github.com/ctsit/redcap-extras/blob/master/.travis.yml#L22). We should investigate the Sauce Labs integration with Travis-CI (see http://docs.travis-ci.com/user/sauce-connect/) to see if we can get this functionality by using Sauce Labs for the selenium tests.

The testing script travis is calling, https://github.com/ctsit/redcap-extras/blob/master/tests/test_redcap_hooks.py, is narrowly focused. It tests only one hook.

I think we need a script that calls a master script for hooks and a master script for plugins. Each of these scripts needs to call a list or directory of lists in its scope. I think it would be better if the master scripts made very few assumptions about what the per-hook and per-plugin tests would do.

I think the hook code removed in commit aaa797b should be reinstated. When it returns each hook should be placed in a directory under ./hooks/library.

I don't think the files should stay in the .hooks/globals folder because this structure assumes the hooks are deployed. I feel the purpose of this repo is to provide a library of hooks that one could deploy. It could provide examples of how these hooks could be deployed, but the repo content itself should not deploy them.

I think we should rename hooks/examples to hooks/library. I would like to see a library of tested, documented hooks. If we provided examples I think they would be examples of how the fully-functional hooks in hooks/library/ could be deployed.

See https://github.com/ctsit/redcap-extras/tree/master/hooks/examples