+1. The proposed changes are good and they are fairly isolated.
Thanks,
Raymond
--------------------------------------------------
From: "Simon Laws" <[email protected]>
Sent: Wednesday, March 25, 2009 7:06 AM
To: "tuscany-dev" <[email protected]>
Subject: [1.x] API change for TUSCANY-2931?
I have a fix for http://issues.apache.org/jira/browse/TUSCANY-2931 but
it implies an API change. In particular to support the separation of
input from output wrappers :
org.apache.tuscany.sca.interfacedef.Operation
getWrapper() -> getInputWrapper() & getOutputWrapper()
setWrapper() -> setInputWrapper() & setOutputWrapper()
isWrapperStyle() -> isInputWrapperStyle() & isOutputWrapperStyle()
setWrapperStyle() -> setInputWrapperStyle() & setOutputWrapperStyle()
The WrapperInfo utility class changes to represent only one wrapper at
once but there is not separate interface for this class. Of course
there are various changes across the runtime to fix up things that
refer to the old methods and to perform separate processing for input
and output wrappers.
Looking back at this I could have taken the slightly less pleasing but
smaller impact route of just changing the WrapperInfo class to
represent separate input and output data bindings which would have
left the getWrapper() method unchanged but would still need the
wrapper style change. We could still go this route if people are
concerned as haven't committed yet.
There is also an addition to allow the inputs and outputs of an
interface to be set independently based on some other interface.
Interface
public void resetInterfaceInputTypes(Interface newInterface);
public void resetInterfaceOutputTypes(Interface newInterface);
But as these are just used from the JMS binding at the moment they
should not affect existing code.
Any concerns about these changes?
Regards
Simon
