On 7/20/06, Christopher Mihaly <[EMAIL PROTECTED]> wrote:
Ok, I agree I may not be using it the best way. But, it doesn't seem correct to me that I can close a connection and then open a connection with the same name and have it fail. If I close a connection, and the function returns successfully, I would think I should be able to turn around and open the connection again and have it succeed (assuming no other process comes in trying to open the connection with that ID, which it cannot in this scenario). Why would this not be considered a bug?
Even though there is no reason at all to ever close a connection, particularly as ActiveMQ supports auto-reconnection to deal with, we should be able to handle code that does this. So it does sound like a bug to me - please raise an issue. Just of interest, how long does it take to close the connection? In previous versions of ActiveMQ we used to wait until a connection was properly closed (blocking the client side until it was done). But sometimes when folks were using embedded brokers or had transport issues that could block the client unnecessarily causing shutdown to hang. So we added a timeout now which defaults to 15 seconds. I have an idea what the issue might be - am about to raise a thread on the dev list to investigate further -- James ------- http://radio.weblogs.com/0112098/
