Jan Wieck wrote:
> > Maybe we should prevent shadowing of a table that is used as a foreign
> > key.
>
> And open another DOS attack possibility? Each user could
> potentially create a permanent table of a name used as a temp
> table somewhere inside of an application. Then create another
> referencing it and boom - the application goes down. Not that
> good IMHO.
>
> I think we need to hand OID's to the RI triggers and have
> some special syntax to tell the parser that a table or
> attribute is named by OID in the actual query.
>
> Will cause alot of headaches for the next pg_dump/restore
> cycle.
>
> (Bruce: still waiting for your call)
>
> Jan
Just curious, but couldn't this also be done in the TRUNCATE
TABLE case? Where the consensus was to disallow TRUNCATE TABLE on
a table whose attribute was the foreign key of another relation?
It seems that proper implementation of SCHEMA and a finer
granularity of GRANT/REVOKE is need to address these issues.
Mike Mascari