"Dieter Maurer" <dieter@handshake.de> writes:
Ross Patterson wrote at 2008-11-3 16:04 -0800:
... If memory serves, a *functional* test fixture calls the Zope publisher when a testbrowser request is made. The publisher opens a new transaction, as it should.
Really?
When I wrote a functional test browser (years before "zope3"), I emulated the Zope publisher rather than calling it directly in order to have control how transactions are handled.
Especially, I used subtransactions instead of top level transactions to avoid the bug reported by Peter.
I'm not completely sure, but I'm *pretty* sure. I remember stepping through the code before and seeing a new transaction start. As for "as it should", it just seems reasonable to me that a functional test fixture would try to use the real application behavior as much as possible. Ross