Re: BUG #15025: PSQL CLI - inconsistency when both -d and -Usupplies a username

On Sun, Jan 28, 2018 at 02:38:46PM -0500, Tom Lane wrote:
> Bruce Momjian  writes:
> > 	psql -d "postgresql://user2@momjian.us/test" -U user1
> > -->	Password for user user1:
> > but it wants the user2 password.
> 
> > It doesn't matter whether -U is first or last, it always prompts for the
> > -U user, but connects as the -d user.
> 
> Bleah.
> 
> > Because the URI is parsed by libpq, I don't think we can do any better
> > than just suppress the user name in the password prompt when a URI is
> > used.
> 
> That doesn't seem very user-friendly at all.  I think that the most
> correct behavior in this case would be to throw an error because of the
> conflicting command line parameters.  Now admittedly, psql doesn't throw
> an error for
> 
> 	psql -U alice -U bob
> 
> so maybe just silently making a choice is OK, but we need to be clear
> as to which choice we made.  Isn't it possible to get the URI parse
> results back out of libpq?

Well, there is PQuser(), but you need to pass a connection struct to
that, and before you connect you don't have one.  libpq's
conninfo_uri_parse_options() is a static function so that can't be used.

I don't see any libpq documentation saying that URIs override
command-line specifications.

-- 
  Bruce Momjian          http://momjian.us
  EnterpriseDB                             http://enterprisedb.com

+ As you are, so once was I.  As I am, so you will be. +
+                      Ancient Roman grave inscription +