Re: FW: LC Comments: Web Method Feature

Hey again,

On Wed, Jul 03, 2002 at 07:08:53PM +0100, Williams, Stuart wrote:
> > Hmm, that wouldn't be my preference.  My preference would be that it is
> > possible for there to be a mismatch between a set of features/MEPs,
> > and a protocol binding.  i.e. "sorry, there's no reasonable way to
> > implement this feature/MEP with this underlying protocol"
> Ok... in our current style that would likely be expressed as a bunch of
> precondition wrt to the use of a feature or features in combination. But we
> are back to the optional/mandatory nature of the use rather than provision
> of features.
> I guess what your saying is... the binding user didn't set webmeth:Method,
> the binding goes 'barf... you didn't set webmeth:Method' and the binding
> user scratches its head... but I know what I expect Request-Response to
> do... if its lucky someone says... just plug in POST it'll work, all will be
> fine... so I use that recipe with no more or less understanding of who to
> organise and identify resources than I had before.

No, I'm saying that a designer of a binding, who wants to incorporate a
list of MEPs, might do lots of work only to discover that she can't
implement one or two of them in a reasonable way.  So she does the best
she can and then specs out that binding, listing only the MEPs that she
was able to incorporate.  i.e. it's a design-time mismatch, not a
runtime one.

> >  [1]
> Well I think that leaves things horribly twisted... but as possibly a non-
> or an anti- layerist... I doubt that would bother you ;-)

We didn't start the fire! 8-)  I'm just trying to make the best of a
sub-optimal situation.

Mark Baker, CTO, Idokorro Mobile (formerly Planetfred)
Ottawa, Ontario, CANADA.     

Received on Wednesday, 3 July 2002 16:02:33 UTC