Re: OID wraparound: summary and proposal

On Wed, 1 Aug 2001, Tom Lane wrote:

> Based on the discussion so far, here is an attempt to flesh out the
> details of what to do with OIDs for 7.2:
>
> 1. Add an optional clause "WITH OIDS" or "WITHOUT OIDS" to CREATE TABLE.
> The default behavior will be WITH OIDS.
>
> Note: there was some discussion of a GUC variable to control the default.
>
> Note: an alternative syntax possibility is to make it look like the "with"
> option clauses for functions and indexes: "WITH (oids)" or "WITH (noOids)".
> This is uglier today, but would start to look more attractive if we invent
> additional CREATE TABLE options in the future --- there'd be a place to
> put 'em. Comments?

I think a fixed default and placing it in parentheses are probably good
ideas.

> 3. For a table without OIDs, no entry will be made in pg_attribute for
> the OID column, so an attempt to reference the OID column will draw a
> "no such column" error. (An alternative is to allow OID to read as nulls,
> but it seemed that people preferred the error to be raised.)

Okay, at least the foreign key stuff will need to change (since it does a
select oid), but I don't think it ever does anything with that except
check for existance, so I could probably make it select 1 as reasonable
replacement.