Dominic J. Eidson writes:
> On Mon, 23 Apr 2001, Joel Burton wrote:
>
> > pg_user holds users
> >
> > (passwords in pg_shadow)
>
> I doubt the -hackers people would let me add SPI_* stuff into libpq, just
> to retrieve whether a user exists or not..
You wouldn't have to do that. There are better ways to read system tables
in the backend. See FAQ_DEV.
> My first thought was to check
> the existance of users against $PGDATA/pg_pwd... One question I'd have
> there, is whether pg_pwd always exists (or, can be relied upon existing.)?
No it doesn't and no you can't.
The best way to verify a user's existence in the context of a new
authentication method is to not do that at all. None of the other methods
do it, the existence of a user is checked when authentication has
completed and the backend starts.
--
Peter Eisentraut peter_e@gmx.net http://funkturm.homeip.net/~peter
