As per Eloy comment in MDL-64655:

B) A new issue is created about to document the rationale and expectations from all the xxxx_access_information functions. Only caps, also completion rules (y/n), maybe settings (y/n) or other calculated canxxxx based in 1 or more of the previous (y/n)... right now we have implemented combinations with all those (and surely more), so it's hard to guess which one is the correct approach to follow. Just aiming to have an official one defined.

We should try to "harmonise" the different functions.