> It's probably best if you use separate cursors anyway. Say you have
> two methods with a shared cursor:
>
> def iter_table_a(self):
> self.cursor.execute("SELECT * FROM TABLE_A")
> yield from self.cursor
>
> def iter_table_b(self):
> self.cursor.execute("SELECT * FROM TABLE_B")
> yield from self.cursor
>
> If the client code calls iter_table_a and partially iterates over the
> result, and then needs to use iter_table_b for some reason, then when
> it gets back to the table A results the data will be gone. You can
> only safely share cursors when you can guarantee that each query will
> be completely processed before the next query is run, and you can't
> guarantee that if you're just returning iterators.
Oops, I admit I never tested that scenario before and you are right, its
not a sane assumption.
Thanks, I'll rewrite that section.
jlc
--
http://mail.python.org/mailman/listinfo/python-list
