Testing  of ZEO is hanging in Python2.5. When i traced i think it is
failing to connect and is attempting to reconnect continously in client.py.
Does this happen due to changes in socket module or due to in
threading module ?

FWIW, when I looked at this about a year ago, the immediate cause was
that Python 2.5 changed exceptions from old-style classes to new-style
classes.  As a result, a bit of low-level code in ZEO's lowest-level
communication layer wasn't recognizing exceptions correctly.

Don't know whether that's still the case.  IIRC, it's at least part of
the problem if the find_global() function in ZEO/zrpc/marshal.py still
contains this block:

   # TODO:  is there a better way to do this?
   if type(r) == types.ClassType and issubclass(r, Exception):
       return r

types.ClassType is the type of an old-style class:

class C: pass  # old-style class
<type 'classobj'>
class C(object): pass  # new-style class
<type 'type'>


<type 'classobj'>


<type 'type'>

Of course it would be wise to review all uses of types.ClassType in ZODB.
For more information about ZODB, see the ZODB Wiki:

ZODB-Dev mailing list  -  ZODB-Dev@zope.org

Reply via email to