Using curr_val Wisely

Hi all,

I've got one daft question, and one good one...


Good One:

I'm using curr_val('blah_id_seq') in a transaction like so: (in pseudo SQL)

BEGIN;
INSERT INTO blah;
INSERT INTO foo (currval('blah_id_seq'));
COMMIT;

So the second insert requires the primary key of the first table as a
reference.

In the documentation, it says that even when there's multiple users, the
back end essentially guarantees that curr_val will work in this case. How
does it distinguish? Within the transaction, connection or something else?

Just wondering so I don't stuff anything up too badly! :)


Daft One:

This would be asked all the time, I'm sure. If I want to return the primary
key, or oid of the inserted record. The best way I can figure to do that now
is by selecting curr_val('blah_id_seq') at the end of this transaction
(which is why I asked the above question, because I'd like to be sure it
returns the id I think it should!)

For some reason, I can't get the postgres interface in PHP to return the
oid... But I'm more interested in doing this as a transaction anyway. :)


- Jeff


-- jdub@aphid.net --------------------------------- http://linux.conf.au/ --

              linux.conf.au - coming to Sydney in January 2001

    Installing Linux Around Australia - http://linux.org.au/installfest/