RE: [wsrf] Singleton Resource Pattern - meaning of "implied"

[Tim Banks <tim_banks@uk.ibm.com>]

Hi Savas,


"Savas Parastatidis" <Savas.Parastatidis@newcastle.ac.uk> wrote on
12/07/2004 18:53:44:

<snip>

> What's bad, in my
> opinion, is that as a requestor you have to reason about it, you have to
> be aware of what you are talking to (i.e., the resource). At the
> architecture level, you are sending messages to the resource and not to
> the service (the service just happens to receive the messages on the
> resource's behalf).

<snip>

Bad? The concept of directing operations at a resource (thing, object) has
been popular in many programming languages. Why is it bad? Isn't this a
matter of personal preference, influenced by the way the service/resource
in question is going to be used? If resources identifiers need to be
surfaced for human consumption (like your own name) then, yes, one would
naturally make them obvious in a document/parameter list. When dealing with
more mechanical resources (such as components of a parallelised
computation) the approach of sending a message to a 'thing' seems more
obvious, to me.
<snip>
> The service will analyse the document and figure
> out what information it needs from the entire document (e.g., the order
> number plus my name) to establish the execution context. We do this
> today already and it's called service-orientation.
<snip>

Yes, the term "Service Oriented" means that, but it implies other
attributes (such as loose coupling) which can be applied independently of
the way parameters are used to express the intent of the interface. I guess
the jury is out on whether WSRF has extended the notion so far as to need a
new term.


Best wishes, Tim Banks