Greg Stark <gsstark@mit.edu> writes:
> Stephan Szabo <sszabo@megazone.bigpanda.com> writes:
>> I'd expect copy to be a single command, no matter how many rows were
>> copied.
> It might prevent you from using pg_dump --inserts ?
Not even that, unless you *also* modified the dump output to wrap
BEGIN/END around it. Otherwise each INSERT is a separate xid.
(Of course you could definitely take a pretty long coffee break while
waiting for a 4-billion-row table to be restored with INSERTs. Also
I think it would be necessary to run VACUUM partway through to avoid
transaction wraparound issues. pg_autovacuum could be expected to
take care of that for you, if it were running. But in practice anyone
sane would use COPY for this, anyway.)
regards, tom lane
