Comment 5 for bug 1564209

@fgallina or @kelvin: Can you check the background for these decisions. I'm not 100% but from memory RnR used to use the SSO (well, LP) username when it has one for the user (as SSO users don't always have a username, it depends whether they have an LP account), so I don't know that we can provide a username consistently. But in Robert's case I'd imagine it's for a user who certainly does have a username (his own account?)

The only consistent options I can imagine are:
 1) always return the consumer key - @Robert - can you use that instead to identify (sorry, I don't know the background), and will this break anything (doubt it, if they're not broken already - if it does currently return either)
 2) add a separate consumer_key which always has the consumer key.

Rodney is keen for (1) so that it's consistent with the click rnr api:

21:47 < noodles> dobey: could you use a different key from the result, I mean if consumer_key was added separately?
21:48 < dobey> noodles: well i think we should match what we're doing for click; and i think software-center might be doing this comparison already anyway
21:57 < dobey> noodles: so, imo, always returning the consumer_key value for everyone in reviewer_username there, would be best
21:59 < noodles> dobey: Right, but that's exactly what Robert *doesn't* want, unless he could also identify with the consumer key?
21:59 < dobey> noodles: he can identify with the consumer key