If you look at the python agent test in
"qpid/cpp/bindings/qmf/tests/python_agent.py", you will see that the
schema is built up programatically. You can add new elements to the
schema as you go along. If you always add content (don't delete or
rename), you can minimize the disruption experienced by your console code.
QMF will handle multiple different definitions for the same schema
classes (it creates a "class-key" that consists of package name, class
name, and a hash (MD5) of the schema). This ensures that a console that
is working with both versions will not misinterpret the data.
-Ted
Ming Zhang wrote:
Hi All
If I use QMF schema to describe a host, it is quite possible that I
will add cpu statistics today, and memory data tomorrow. this means i
have to change the schema from time to time. what is the good practice
for this? Do I have to recompile the schema, change the code, and
shutdown my current agent and start a new one? or i have a way to
dynamically add some as a plugin like architecture? Thanks.
Ming
---------------------------------------------------------------------
Apache Qpid - AMQP Messaging Implementation
Project: http://qpid.apache.org
Use/Interact: mailto:[email protected]
---------------------------------------------------------------------
Apache Qpid - AMQP Messaging Implementation
Project: http://qpid.apache.org
Use/Interact: mailto:[email protected]
