ExamSessionManager.create() cannot return next question pool

[mhrivnak]

both create() and resume() are not capable of returning only the next question pool with unanswered questions, but this is the most desirable behavior. I propose that if fetch_all == False, we then return the next question pool only.

Basically, we will always return results from _get_next_questions(), and the passed value of 'include_answered' will be equal to the value of 'fetch_all'.

[jc0n]

Basically, we will always return results from _get_next_questions(), and the passed value of 'include_answered' will be equal to the value of 'fetch_all'.

Are you saying then that fetch_all==False implies that include_answered==False (which is now include_responses to prevent ambiguity) and in the case where it is False, only the next question pool (and its questions) should be returned?

This whole routine might benefit from a makeover as you noted in #86 I believe

[mhrivnak]

I think my quoted assertion is incorrect. I propose this:

• For create(), it will call _get_next_questions(include_responses=False)
• For resume(), it will take a parameter "include_responses" that will be passed to _get_next_questions()
• In all cases, as noted in the original issue description, a call to create() or resume() where fetch_all==False will return the results of a call to _get_next_questions().

Looking at the code, I'm not convinced that fetch_all is doing what is advertised. We should make sure that if fetch_all==True, we get all questions for the whole exam. And if fetch_all==False, only questions for the next incomplete question pool.

This definitely could use a makeover, but I don't think we can invest in that at the moment since this feature is oh-so-close to being done.

[jc0n]

@mhrivnak: Take a look at 42fe761 and let me know if that is what you had in mind. If so, I'll update the tests accordingly.

[mhrivnak]

That looks reasonable to me.