sandesha client side usage

[aditya vikram <ad...@yahoo.com>]

Hi,

Is it possible to use sandesha on the client side
without having to modify the client code? 

Currently, it seems the client code has to be changed
with atleast the following:

1. add a call to: RMInitiator.initClient() method

2. set transport with: call.setTransport(new
RMTransport(targetURL, ""));

Is it be possible to do the above in the client-side
deployment descriptor so that RM is completely
pluggable and existing client code does not need to be
modified ?

Another question is about using sandesha with stub
based invocations as opposed to DIIs using
Call.invoke(). Can this be done? or is sandesha
designed to be used only with DIIs ?

Thanks,

Aditya



		
__________________________________ 
Do you Yahoo!? 
Yahoo! Small Business - Try our new resources site!
http://smallbusiness.yahoo.com/resources/ 

RE: sandesha client side usage

[Jaliya Ekanayake <ja...@opensource.lk>]

Hi Aditya,

Sorry for the late reply.

Here are my thoughts for the questions you raised.

Sandesha uses the "RMInitiator.initClient()" method to initialize the client
side listener and the queue. This can be done without coding it in the
client.

However we need a point where we can give the control back to the client
application, after sending all the messages. This can be done using some
polling mechanism inside the RMSender.

Setting of the transport can also done using some handler which can be
deployed using the deployment descriptor.

Setting of message numbers can also be done by the RMSender, but then we
assume that the order of the messages is the same as the order in which they
are received by the RMSender.(we will not be able to use threads in the
client code)

Setting of LastMessage should be done by the client, since RMSource (All the
RM related stuff in the client side) does not know about the last message if
the client does not mark it as a last message.

So the current implementation was decided after considering all these and
sticking into one configuration point, that is the client code itself. If we
can find a clear solution for all of the above then Sandesha configuration
will simply be on a config file.

Currently all the samples use DII instead of Stubs, Sandesha can use with
stubs, however you have to modify the stub code, so that properties that we
set to the "call" object is set inside the stub as well. We can easily use
stubs if we can find a good solution to the above problems (message number
and the last message)


If you have any suggestions please send them as well, so we will have a
better product.

Thanks,

Jaliya

-----Original Message-----
From: aditya vikram [mailto:adityavkrm@yahoo.com] 
Sent: Monday, March 21, 2005 3:19 AM
To: fx-dev@ws.apache.org
Subject: sandesha client side usage

Hi,

Is it possible to use sandesha on the client side
without having to modify the client code? 

Currently, it seems the client code has to be changed
with atleast the following:

1. add a call to: RMInitiator.initClient() method

2. set transport with: call.setTransport(new
RMTransport(targetURL, ""));

Is it be possible to do the above in the client-side
deployment descriptor so that RM is completely
pluggable and existing client code does not need to be
modified ?

Another question is about using sandesha with stub
based invocations as opposed to DIIs using
Call.invoke(). Can this be done? or is sandesha
designed to be used only with DIIs ?

Thanks,

Aditya



		
__________________________________ 
Do you Yahoo!? 
Yahoo! Small Business - Try our new resources site!
http://smallbusiness.yahoo.com/resources/