On 29 Mar 2007 00:19:03 -0700, Mark Baker <[EMAIL PROTECTED]> wrote: > > > > > > > On 3/22/07, Gregg Wonderly <[EMAIL PROTECTED]> wrote: > > I think about it in terms of the data passed. With HTTP, you open a port, > with > > a particular transport. You send an operation string of the form > > <OPER> <RESOURCE>. You then follow that up with headers for http 1.1 and > later, > > a blank line, and then any associated content, which is specific to the > type of > > operation you are invoking. You then might wait for a reply depending on > what > > type of operation you invoked. > > > > With RMI, the client end point object opens a port using a configured > transport. > > The operation is uniformly and always "invoke". The resource is indicated > by > > sending the object id and method name, and the associated content, the > > parameters to the call, are sent. You always wait for a reply, and > operations > > continue. > > > > This is how I see it anyway. > > Ok, well it seems clear that you're using different terminology than > the rest of us in this conversation, as we see operation and method as > the same thing. But to put things in your terms, the uniform > interface constraint would mean that every service offered the same > set of *methods*. > > Does that help?
To clarify Mark, just so I'm clear as Crystal here. Are you saying that every service has only the HTTP "methods"(/operations) and that there is no conceptual framework within which anything else can be considered as a method or operation. > > Mark. >
