SQLObject and Python 3
There are a few changes in the behaviour of SQLObject on Python 3, due to the changed stings / bytes handling introduced in Python 3.0.
In Python 3, BLOBCol now accepts and returns bytes, rather than strings as it did in Python 2.
In Python 3, StringCol now accepts arbitrary Unicode strings.
The dbEncoding parameter to UnicodeCol has no effect in Python 3 code. This is now handled by the underlying database layer and is no longer exposed via SQLObject. The parameter is still available for those writing Python 2 compatible code.
SQLObject is tested using mysqlclient as the database driver on Python 3. Note that the default encoding of MySQL databases is latin1, which can cause problems with general Unicode strings. We recommend specifying the character set as utf8 when using MySQL to protect against these issues.
For most cases, things should just work as before. The only issues should around UnicodeCol, as how this is handled has changed.
The Python 3 sqlite driver expects Unicode columns to be encoded using utf8. Columns created using the default encoding on Python 2 should work fine, but columns created with a different encoding set using the dbEncoding parameter may cause problems.
Postgres' behaviour is similar to sqlite. Columns created using the default encoding on Python 2 should work fine, but columns created with a different encoding set using the dbEncoding may cause problems.
For MySQL, the results depend on whether the Python 2 database was using MySQLdb's Unicode mode or not.
If a character set was specified for the database using the charset parameter, such as:
Things should work provided the same character set is specified when using Python 3.
If a character set wasn't specified, then things may work if the character set is set to match the dbEncoding parameter used when defining the UnicodeCol.