Yes, it makes sense, but what about the other issue - multiple messages sent
together? Can I decode a few messages and let MINA invoke my IoHandler's
messageReceived method with all of them somehow?

On 4/1/06, peter royal <[EMAIL PROTECTED]> wrote:
>
> On Apr 1, 2006, at 2:39 PM, Zohar Amir wrote:
> > 1. What's so bad about StreamIoHandler?
>
> its best used for integration with "legacy" code that is stream-
> based. When working in MINA, you'll get the best performance by not
> dropping back to blocking streams.
>
> > 2. I've searched in
> > mina-0.9.2\examples\src\main\java\org\apache\mina\examples for
> > ProtocolCodecFilter but didn't find anything - can you point at the
> > right
> > example?
>
> the 'sumup' example.. http://svn.apache.org/viewcvs.cgi/directory/
> trunks/mina/examples/src/main/java/org/apache/mina/examples/sumup/
>
> > 3. Just from looking at the javadoc, it seems like messageReceived is
> > invoked with a message - how is it created? Can this message be
> > created with
> > data from several packets arriving (cellular networks are sometimes
> > slow)?
> > What if several protocol messages are sent together?
> > Thanks again,
>
> You provide the ProtocolCodecFilter with a ProtocolEncoder and a
> ProtocolDecoder. Encoder is simple, it receives an application-level
> message (some sort of Object) and should turn it into a ByteBuffer.
>
> For the Decoder, I highly recommend extending
> CumulativeProtocolDecoder. It will help your packet problem. You
> subclass 'doDecode', and if you don't have a full packet in the
> ByteBuffer, just return false, and nothing further will happen. The
> next received packet will be appended, and it will be called again,
> Just decode a single message, return true, and if there is more data,
> your decoder will be called again.
>
> Then, in your IoHandler / filters after the ProtocolCodecFilter, the
> object in messageReceived will be the object that you decoded, rather
> than a raw ByteBuffer.
>
> Make sense?
>
> -pete
>
>
>
> --
> [EMAIL PROTECTED] - http://fotap.org/~osi
>
>
>
>
>

Reply via email to