Falby, Walter wrote:
I'm new to MINA and have only recently looked at the Reactor pattern. It
seems that MINA is a replacement for that pattern.
It's not a replacement, it implements this pattern (more or less ...)
I'm confused about
the threading in MINA. In the tutorial on "Configuring Thread Model", it
states that it is only for MINA 1 and that later versions will not have
the ThreadModel. I don't understand how MINA can perform asynchronously
without threading.
MINA is using threads. But it does not anymore (in 2.0 branch) defines a
threading model. (Thread Model refers to the way we manage threads in an
Executor. We don't allow the user to define this anymore, just because
it's too tricky)
The application I'm developing runs on UNIX System Services (USS), which
is UNIX on an IBM z/OS machine. I'm using Java 5, derby, and log4j. An
unknown number of clients can connect to this server application. In the
initial release, there will be two different connection formats and
message formats. The code that receives a message determines if the
message is written to the database or processed immediately. In the
majority of cases, there is a reply for every request. There is a thread
for processing messages written to the database. There is also a thread
to receive messages from the z/OS console. The Reactor pattern looked
promising because I could start a short-lived thread to process received
messages. The acceptor would be available to accept another message and
start a message processor. How does MINA keep the acceptor available
without threading?
it is using threads. You can still inject an ExecutorFilter in the chain
if you need.
I've looked for some examples and found some at mvnbrowser.com. I
downloaded the samples for MINA version 1.1.7, or so I think. I've not
found anything on threading in them.
Sure, some samples would help :)...
Hope it helps !
--
--
cordialement, regards,
Emmanuel Lécharny
www.iktek.com
directory.apache.org
