Re: pg_migrator issues

Alvaro Herrera <alvherre@commandprompt.com> writes:
> Getting rid of the need for placeholders is a good idea.  +1 on getting
> TOAST tables created with the correct relfilenode from the start.  I
> don't know that preserving any other relfilenode is useful; however if
> it means you no longer have to rename the files underlying each table,
> it would probably also be a good idea.

I think this is an all-or-nothing proposition: if you try to preserve
only some relfilenodes, you risk collisions with automatically assigned
ones.  It's just like the situation with pg_type OIDs.

I concur that trying to preserve them looks like it would be less work
than the current method.
        regards, tom lane