The other solutions presented thus far sound great.  Obviously the devil
is in the details, but in our situation, we have found using an SQL
storage engine to be most useful.  It does no do RPC per se, but we have
found we can "wakeup" the other side to do something using a simple http
call or trigger or queue.  We examined J2EE and high end transaction
systems and found they had many more cons (overhead, maintenance,
training, speed, learning) than pros.  Since the objects on the
different systems might view the content in different ways, SQL tends to
be more flexible.  We simply created different object types that access
the SQL and present it in an appropriate manner to that system.  When
creating a large system distributed system, reporting is also important,
and SQL combined with Zope does wonders.

Zope3-users mailing list

Reply via email to