[DB-SIG] DB API compliance test--clarification

[M.-A. Lemburg]

David Rushby wrote:
> Code snippet from Stuart's DB API compliance suite:
> 
> -----------
> def test_commit(self):
>     con = self._connect()
>     try:
>         # Commit must work, even if it doesn't do anything
>         con.commit()
>     finally:
>         con.close()
> -----------
> 
> Is it reasonable to commit a transaction without ever having created
> it?

You create the transaction implicitly by connecting to the database.

> It seems to me that this behavior is likely to give rise to subtle
> errors, as when the programmer thinks he's made some changes to the
> database without having actually done so (without even having
> established a transaction), and the commit() call doesn't complain
> about the lack of an active transaction.
> 
> kinterbasdb (Interbase/Firebird driver) currently raises the following
> exception in this situation:
> 
> ProgrammingError: (-901, 'There is no active transaction to commit.
> Consider using begin() to explicitly start a transaction in advance of
> this commit() call.')
> 
> If allow-commit-of-a-nonexistent-transaction is really the convention
> in the Python DB API community, I'll change kinterbasdb to conform to
> it, but I don't like it.

Opening a connection implicitly starts a transaction and all
calls to .commit() and .rollback() do the same. There never is
a state which you could call "non-existent transaction" (and that's
good :-).

-- 
Marc-Andre Lemburg
eGenix.com

Professional Python Software directly from the Source  (#1, Feb 17 2003)
 >>> Python/Zope Products & Consulting ...         http://www.egenix.com/
 >>> mxODBC, mxDateTime, mxTextTools ...        http://python.egenix.com/
________________________________________________________________________
Python UK 2003, Oxford:                                     43 days left
EuroPython 2003, Charleroi, Belgium:                       127 days left