On 18/10/06, Trustin Lee <[EMAIL PROTECTED]> wrote:

Congratulations for the incubation, and I wish you good luck.  I hope we
collaborate closely on MINA.

Thank you.

I certainly hope that some of our performance work can feed into MINA.
We are doing another cycle of performance tests at the moment.

That's exciting.  Can we adopt the code after we move to Java 5?  We will
move to Java 5 soon.

Sure. In fact the current code would work fine with the backport too.

Why don't you create a demultiplexing IoFilter which decorates a thread pool
filter?  The filter could make the messages with higher priority bypass the
thread pool.  Otherwise, you could have two thread pool filters in it and
choose which pool to use, although it doesn't guarentee the thread safety at
all.

Yes, we could certainly do it in a separate filter. It just seemed
simpler when writing our pool to make it decide what gets added to the
session event queue.

RG

Reply via email to