Hiroshi Inoue <Inoue(at)tpf(dot)co(dot)jp> writes:
> Tom Lane wrote:
>>
>> Given Hiroshi's objections, and the likelihood of compatibility problems
>> for existing applications, I am now thinking that it's not a good idea to
>> turn off OID generation by default.
> Would OIDs be globally unique or per table ?
Same as now: if you have a unique index on 'em, they're unique within a
table; otherwise, no guarantee at all (once the system wraps around).
We should document this state of affairs better, of course, but I'm not
proposing to change it. The point here is just to let people suppress
OIDs for tables that don't need them, and thereby postpone OID wraparound.
regards, tom lane