At 09.35 13/11/2008 +1100, you wrote:
That is another clue that makes me think the marshaller might be involved.
Have you changed pippo to only support out-of-process registration?
Cheers,
Mark
Sorry for replying late, I had some problems and I could not do tests on
pippo_server before.
Well
I'm afraid that with the exact same process (although the 'print' says
'print QI, iid' - using python 2.6 on vista) I see exactly:
| Object with win32trace dispatcher created (object=None)
| Object with win32trace dispatcher created (object=pippo_server.CPippo
instance at 0x02AD8210)
| QI
Thanks Mark for the test. I'm running the test on Windows XP and
Windows 2k,
both with python 2.5.2. The problem is that the QI with value to 0 is
raised somewhere in the pythoncom code
I'm pretty sure pythoncom never (explicitly) QI's for IID_NULL. It may QI
for one of its private
The result in PythonWin trace Collector is:
Object with win32trace dispatcher created (object=None)
Object with win32trace dispatcher created (object=pippo_server.CPippo
instance at 0x015DF210)
{----}Traceback (most recent call
last):
File