QA (public) » openQA Project (public)

what is the exact plan here? I suppose a local script that accesses the database directly is the safest choice?

That is the fallback solution, my original plan was to detect first run of the application with uninitialized database and instead of script do web wizard setting stuff up - mainly as opportunity to figure out how Mojo works, also would be more convenient.

Ok, couldn't figure out how to tap into the web thing and discovered how ugly Perl is, so giving up for now, going with the simple script.

Rewrote based on comments, will play with test case

To create the test, I would need to simulate OpenID. Tried playing with it, nothing usefull that I could decipher from OpenID package test-suite. Option would be to start slightly modified OpenID server on localhost and try to authenticate against it and make it part of the tests, but sounds like an overkill :-/

Could move check somewhere else but running it just after successful authentication makes sense. Elsewhere it would get run multiple times. Other option would be refactoring whole login process and moving account creation and permission setup out of OpenID handle. That would help us with testing, but not sure whether it is worth complicating things so much :-/

Now, the first user logging in gets admin superpowers. That partially fix the problem, but several manual steps in the web interface are still required to get a fully functional system (promoting a user to be operator, creating an API key, copying the key and secret in the proper file...). Can we really consider to issue to be 100% done?