Hi, I believe SQLAlchemy's behavior when creating primary_key integer columns is incorrect, at least against PostgreSQL, but before opening a ticket I'd like to hear what others think about it.
If you define a column as Integer or SmallInteger (Smallinteger) and primary_key=True, SA assumes it requires an associated SQL sequence, i.e., a SERIAL (or auto-increment) datatype. In fact, in the latter case, it overrides the SmallInteger specification and creates an Integer column because (at least in PG) SERIAL implies INTEGER. I have at least two tables in an existing database that have SMALLINT primary keys and the values are *not* sequential (one of them is a calendar year and not all years are logged in the table, the other has arbitrary codes). I also have a table that is one-to-one with the principal table (which does have a SERIAL key), but for operational reasons (it keeps track of activity against the other) it's kept separate from it, so it's INTEGER primary key does not have an associated sequence (except by derivation). I don't think my examples are unique. I'm pretty sure there are many systems where identifiers are assigned non-sequentially, e.g., bank account numbers with check digits. Granted that these will tend to be legacy systems, but I thought SA was supposed to cater to those. Joe ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys -- and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV _______________________________________________ Sqlalchemy-users mailing list Sqlalchemy-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/sqlalchemy-users
