I'm tracking development by storing pg_dumps in CVS. This is pretty handy.
I even have a script that makes diffs of the versions to help us write
upgrade patches. If anyone has hints or wants build a general solution for
this type of thing, I'm interested.
The problem is this:
We create two tables, each with implicit SEQUENCES by using the SERIAL datatype.
Then:
pg_dump --schema --table=sometable databasename
pg_dump --schema --table=anothertable databasename
The output for sometable includes the "CREATE SEQUENCE
sometable_column_seq" statment, but the output for anothertable does not.
It only referes to that sequence in the column type nextval statement.
How can I reliably pg_dump the sequence with the table?
Mike Adler
