I have decided to place the non blocking http/s transport on-hold for a
short time, as I would like for a Synapse 0.90 to be released without
any further delay. The code I currently have is checked into the
https://svn.apache.org/repos/asf/incubator/synapse/branches/NIO branch
for anyone to look into. Note that this implementation tries not to
buffer the complete message body into memory. Hence, for very large
messages (e.g. with MTOM) the clients may block, but by turing the
buffer sizes appropriately, this could be minimized for the more common
case where messages may range from around 4 ~ 16k.
The class ReactorTester
(http://svn.apache.org/viewvc/incubator/synapse/branches/NIO/modules/niohttp/src/org/apache/axis2/transport/niohttp/impl/ReactorTester.java?view=markup)
gives some simple cases on how the transport could be used. The
transport has still not been fully integrated as an Axis2 transport..
The plan is to complete this work on the branch, and to donate the code
to the Axis2 project around December to get feedback from a much wider
audience and get the code tested once its better, and then use it from
within Synapse. This is the same approach we followed when we developed
the JMS transport for Synapse, and then move it into Axis2..
Hope you all would agree with me on this, as it seems like we need to
nail down the many features we have already implemented in Synapse and
make a release first, without getting tied down right now..
thanks
asankha
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
