Would this allow you to create a tuble with a user-specified oid?
We definitely need this functionality for all tables, not just large
objects.
However you won't be able to discover if the oid is free or not. You
could end up with duplicate oids. Thus anybody who uses the feature
should know what they're doing.
Philip Warner wrote:
>
> I am wondering if anyone can see an immediate problem in creating a function:
>
> heap_create_oid
>
> which is just like heap_create, but takes an oid arg. This oid (assuming
> it's free) would be used to create the object.
>
> This could then bes called from a new
>
> heap_create_with_catalog_oid
>
> which in turn will be called from
>
> inv_create_oid
>
> to ultimately allow pg_restore to recreate BLOBs, via lo_create_oid.
>
> This plan was developed by simply looking at the LO code in total
> isolation, so I am aware it could be totally flawed for the rest of the DB,
> but I would appreciate some input...
