First cut at a rules processor, separate from the disposition of rule hits.

The basic idea is that we process rules on a mlist, message, metadata triplet
making a list of all rules that hit.  Then a different part of the system will
decide on the disposition of a message based on which rules hit and their
priority.

The doctest and plugin architecture is in place, including the tie-in to
setup.py.  Ported the first rule -- emergency.py -- to the new rule
processor.

We no longer need SQLAlchemy as a requirement, and the setuptools/bzr plugin
name has changed.

1 files changed, 6 insertions, 1 deletions

diff --git a/Mailman/interfaces/__init__.py b/Mailman/interfaces/__init__.py
index c4094e365..20dff7fdf 100644
--- a/Mailman/interfaces/__init__.py
+++ b/Mailman/interfaces/__init__.py
@@ -46,7 +46,12 @@ def _populate():
                 is_enum = issubclass(obj, Enum)
             except TypeError:
                 is_enum = False
-            if IInterface.providedBy(obj) or is_enum:
+            is_interface = IInterface.providedBy(obj)
+            try:
+                is_exception = issubclass(obj, Exception)
+            except TypeError:
+                is_exception = False
+            if is_interface or is_exception or is_enum:
                 setattr(iface_mod, name, obj)
                 __all__.append(name)