Milos Prudek wrote:
> I have a serial autoincrement column called "idmember" in my main table
> (members). This serial column is a key to a second table. A row in
> "members" table corresponds to many rows in the second table.
> My question is: is this the best practice?
>
> Here's an example in Python:
> conn=psycopg.connect(dbconnstr)
> c=conn.cursor()
> # LOOP BEGINS HERE...
> Cmd = "INSERT INTO members ... VALUES (...);"
> c.execute(Cmd, Data)
> Cmd = "SELECT currval('members_idmember_seq') FROM members LIMIT 1;"
A simple "SELECT currval('members_idmember_seq');" will do it. The
sequence isn't part of the table.
> c.execute(Cmd)
> idmember = c.fetchone()[0]
> Cmd = "INSERT INTO msg (idmember,txt) VALUES (%s,%s);"
Alternatively, you could rewrite this query:
"INSERT INO msg (idmember,txt) VALUES (currval('members_idmember_seq'),
%s);"
--
Richard Huxton
Archonet Ltd