Hello.

I've been searching for information about the possible drawbacks of using a message selector to consume some specific messages from a queue. I haven't found any clear answer to my doubts.

I'm thinking in two scenarios:

- Creating just one queue and configuring some mdbs with different message selectors to process every kind of message. In this case I'll set 2 or 3 properties to configure the message selectors.

- Creating a sort of "failed to process message" queue. When any mdb fails to process its message, it will produce a new message to this error queue, setting some properties to contain error related information. In this case threre will be about ten properties.

This error queue will be processed on demand. I will call a synchronous consumer, configuring a message selector with the appropiate property values to read just a subset of the messages on the error queue.

Regarding to performace, the question is: is it appropiate to use the message selector facility for the shown purposes?

Both scenarios can be changed to avoid using a message selector, but paying for some more desing complexity.

Thank you very much.

Juan Pedro Lopez