You can clone with
HTTPS or Subversion.
I noticed the UID from OAuth requests are being stored as an Integer (cast via #to_i). I am not familiar enough with mongo to know whether there is significant performance gain by using an Integer over a string. However, this forces providers to use integer IDs for users, which is unfortunate (eg: UIDs are not an option).
I don't remember why this decision was made.
Sure, it's realistic. It's a document store, there's no schema! Seems fine.
There is theoretically no significant difference in storing the integer or the string version.
Right. I guess my question was brought up because I am interested in making it easy to integrate with other OAuth providers, and having this requirement requires all providers to use integers as their unique identifier for users. So, for example, a provider is not able to use UUIDs instead, because when it is cast via #to_i, it results in an ID of '0'...which is problematic.
I don't see why not.