Re: Role Attribute Bitmask Catalog Representation

Adam,

* Adam Brightwell (adam(dot)brightwell(at)crunchydatasolutions(dot)com) wrote:
> I am simply breaking this out into its own thread from the discussion on
> additional role attributes (
> http://www.postgresql.org/message-id/20141015052259.GG28859@tamriel.snowman.net
> ).

Makes sense to me, thanks.

> Based on these above I have attached an initial WIP patch for review and
> discussion that takes a swing at changing the catalog representation.

Just a quick initial look, but I don't think we want to #include
parsenodes.h into pg_authid.h. Why not put the #define ROLE_ATTR_* into
pg_authid.h instead? We have similar #define's in other catalog .h's
(PROARGMODE_*, RELKIND_*, etc).

I'm also not a huge fan of the hard-coded 255 for the default superuser.
That goes back to the other question of if we should bother having those
explicitly listed at all, but I'd suggest having a #define for 'all'
bits to be true (for currently used bits) and then a comment above the
superuser which references that #define (we can't use the #define
directly or we'd be including pg_authid.h into pg_proc.h, which isn't a
good idea either; if we really want to use the #define, genbki.pl could
be adjusted to find the #define similar to what it does for PGUID and
PGNSP).

Thanks!

Stephen