Saturday 27 December 2003 13:50
> This analysis is nonsense ... system catalog operations do not do the
> same kinds of permission checks as user queries do. Furthermore, if
> he'd not had permissions to create users, the initial CREATE USER
> command would have failed, and so would CREATE DATABASE (since it would
> then be specifying a nonexistent owner name).
The CREATE USER *did* fail, didn't it? I'm sorry if I was inaccurate in my
response, but I don't believe I was - I think maybe we're just interpreting
the E-Mail two different ways. Devrim, if you could copy and paste exactly
what you're doing and exactly where the errors appear in a psql session, that
would be most helpful.
I did nat say that the user had to be a superuser, only that the user had to
have createuser permission in order to CREATE USER. And the last sentence of
the first paragraph may not be syntactically accurate because I am not
familiar with PostgreSQL's internals, but that is the effect of creating a
user, as I said in the second paragraph.
> My guess is that Devrim interpreted your suggestion as telling him to
> make the created user (tdmsoftmailserveruser) a superuser, which would
> naturally suppress any and all permissions failures for operations
> executed by that user. That's hardly a reasonable answer to his problem
> though.
That is *NOT* what I suggested at all.
> As to what his real problem is, I dunno, but I'd like to find out.
Sure.
Vertu sæll,
--
Sigþór Björn Jarðarson (Casey Allen Shobe)
cshobe@softhome.net / http://rivyn.livejournal.com
Jabber: sigthor@jabber.org; ICQ: 1494523; AIM/Yahoo: SomeLinuxGuy
Free development contributor of:
> KDE toolbar icons
> Kopete user interface, usability, and testing
> X11 Icelandic Dvorak keymaps
> Reporting of over 100 Kopete bugs