> I don't think mandating that the username we use for SASL is the unix user name > the client runs as makes a lot of sense. Imo it would be better if we always asked > for username/password when SASL asks for it, and prefill the username field > with the local unix user name. Exactly. I can implement that if you want?