[DB-SIG] return value of fetchone (DBAPI 2.0)

[Stefan Fleiter]

Hi!

I had a discussion by private mail with Stuart Bishop about the
return value of fetchone() of the DBAPI 2.0.

In his UnitTest (http://zen.freezope.org/Software/DBAPI20TestSuite/)
it is assumed, that
     'cursor.fetchall should return an empty list if '
     'a select query returns no rows'.

In http://www.python.org/peps/pep-0249.html,
you can read in the paragraph specifying fetchone():
     An Error (or subclass) exception is raised if the previous
         call to executeXXX() did not produce any result set or no
         call was issued yet.

Mr. Bishop writes:
 > I'm pretty certain the existing implementation is correct. There
 > is a difference between returning no result set and an empty one.
 > So .fetchall() after .execute('update ...') may raise an exception,
 > but .fetchall() after .execute('select * from foo where 0 = 1')
 > returns an empty list.

I think the pep does not state this and should really be more clear 
regarding this issue.

One thing that comes to my mind now is, that with Oracle and the 
"returning clause" almost every statement, including updates can produce 
a result.

Con someone please clarify on this?

Thanks in advance,
Stefan Fleiter