I'd like to wrap a bit of code around every request.  It should look like:

save some stuff
process request
do something to session based on saved stuff
save session

If I put this into process_request, I'm too late to save the session.
If I put it around try_publish I'm still too late because
finish_successful_request is called before try_publish exits.  It
makes more sense to me to pull finish_successful_request out of
try_publish and stick it in process_request, like so:

*** publish.py.old      2005-09-09 15:37:34.000000000 -0700
--- publish.py  2005-09-09 15:39:52.000000000 -0700
***************
*** 250,259 ****
          assert path[:1] == '/'
          # split path into components
          path = path[1:].split('/')
!         output = self.root_directory._q_traverse(path)
!         # The callable ran OK, commit any changes to the session
!         self.finish_successful_request()
!         return output

      def filter_output(self, request, output):
          """Hook for post processing the output.  Subclasses may wish to
--- 250,256 ----
          assert path[:1] == '/'
          # split path into components
          path = path[1:].split('/')
!         return self.root_directory._q_traverse(path)

      def filter_output(self, request, output):
          """Hook for post processing the output.  Subclasses may wish to
***************
*** 279,284 ****
--- 276,284 ----
          except:
              # Some other exception, generate error messages to the logs, etc.
              output = self.finish_failed_request()
+         else:
+             # The callable ran OK, commit any changes to the session
+             self.finish_successful_request()
          output = self.filter_output(request, output)
          self.logger.log_request(request, start_time)
          if output:


This puts the three "post request" methods in one place and
incidentally lets me wrap try_publish and do "pre-post request" stuff
there.