Stephen Frost <sfrost@snowman.net> writes:
> Perhaps I'm missing something obvious (and if so, I'm sorry) but
> couldn't we just build up the character array in PQsetdbLogin to be
> passed in to connectOptions1?
That's a possibility too, though by the time you've finished building
that string (with appropriate quoting) and then tearing it apart again
in conninfo_parse, it's likely that you've eaten the cycles you hoped to
save ... especially since that processing would happen for all uses of
PQsetdbLogin, whether or not they could avoid calling pg_fe_getauthname.
I'm starting to feel that it's not worth messing with, which is
obviously the same conclusion we came to last time round ...
regards, tom lane