Thank you so much Ajith, your answer is really valuable to me. Now I have some more questions about the implemention of MEP. Since a MEP is a one-to-one binding to an operation and a MEP can hold one or more messages, how a MEP is mapped to an operation when the MEP contains more than 2 messages?
An operation is naturally in-out(provider view) or out-in(requester view), and either input or output can be optional. So the eight MEPs mentioned in WSDL 2.0 are very clear and natural. But what if a MEP contains more than 2 messages? Can an operation in WSDL 2.0 define more than one input or output? The operations in business logic such as a Java class or an EJB are always one-input-type and one-output-type. So what is the mapping rule? Regards, HaoRan Zheng On 8/15/05, Ajith Ranabahu <[EMAIL PROTECTED]> wrote: > Hi, > Since you seem to be an enthusiastic guy I'll try to answer your questions > to the best of my knowledge. > > 1) Must all messages in a MEP be exchanged between exactly two > entities? Can a MEP intersect with more than one service? > > Well, There's no such limitation. The MEP talks about the server side of > one entity (how many coming in,how many going out and in what order) there's > no mentioning about the other entity. Infact when WS-addressing is enabled, > the messages can be sent through to many different entities. > > 2)Is MEP and operation a one-to-one binding? > > Yes > > 3) If the answer of the previous question is negative, can we say a > MEP links multiple operations? Then how a MEP is bound or waved with > multiple concrete operations in one or several service? > > The answer is positive :) > > 4) Also, if the operation and MEP is not a one-to-one binding, the way > to invoke a web services seems to be changed. Before, we invoke a > service by invoking one of its operation; but now we invoke a service > by starting a MEP. Am I right? > > Well, an operation and a MEP *is* a one-to-one binding. A given operation > can have only one MEP. > > > -- > Ajith Ranabahu
