Also, scanning through the backup file for BLOBs used in the table to be
restored will be slow, to say the least. It can be done in one pass by
building an indexed temporary table with all OID fields from the target
table, but it is definitely getting messy. Especially since Jan was talking
It recently came to my attention that, when dumping a single table and
using the --blobs option, pg_dump will dump all BLOBs (not just those
referenced by the table). Similarly, restoring a single table from a backup
will restore all the BLOBs.
This is (at least) non-intuitive.
Unfortunately,
