Re: [ws-caf] the issue on Context type identification

[Kevin Conner <Kevin.Conner@arjuna.com>]

Furniss, Peter wrote:
> Very briefly, the reason for wanting the ability to "segment the context
> space" was to make the marking for recognition for processing orthogonal
> to whatever the context is really about. This would seem to be directly
> following the pattern of use (or capability) you are proposing for
> contexts as a whole - i.e. you are asking to have the ability to
> recognise a context as a context so that a system can do something with
> all contexts, regardless of what they are really about. (or at least all
> contexts that are only recognisable as contexts and not as their
> specific form).

Only that I'm not arguing for segmentation but generic identification.

> Your proposal says there is something special about contexts-in-general
> that would make it worth identifying such for special processing by
> systems that are neither the originator or the consumer of the context
> as a header (though they may be consumers of the body of the message). I
> can't see why the mere fact that someone derived their specification
> from WS-Context (with no specification in ws-context of what processing
> applies)should be the marker for treating the headers in a particular
> way.   I can see (some) utility in having markers for headers,
> orthogonal to their primary use (and where such primary use is opaque to
> a system) which  could be lead to locally-defined, but
> publically-announced processing [ of which propagation is an example ].

I am not arguing that there is something special about context, just that someone else may think so :-), and neither do I believe that identifying a context as such is orthogonal to context.  This is a facility that existed until the last change and one that does have its use (however minor it may turn out to be).

	Kev

-- 
Kevin Conner
Arjuna Technologies Ltd.