Re: AW: AW: Reimplementing permission checks for rules

Zeugswetter Andreas SB <ZeugswetterA@wien.spardat.at> writes:
> I don't know, but imho one field for all permissions would have been
> better, like we discussed for the permissions system table, since
> there are more rights in SQL than read/write (e.g. write is separated
> into insert, update and delete)

Not really necessary in the current implementation.  checkForWrite
essentially identifies the target table for the operation, and then
the query's commandType is used to decide exactly which flavor of
write access to check for.

IIRC, the ACL code doesn't have the right set of primitive access types
anyway to match the SQL spec's requirements, but that's a task for
another day.
        regards, tom lane