RE: FW: LC Comments: Web Method Feature

Hi Jacek,

Well Jacek... I think that you've put your finger on the heart of the
problem, although I don't think that your proposed solution 'satisfies' me.

> Say explicitly that the HTTP binding requires the WebMethod 
> feature to be used. 

My expectation of the features/property based Binding framework has always
been as you state:

>  I agree with Stuart though that once we call something a 
> feature, it's not mandatory, it MAY be used to gain something. A 
> binding provides features, it doesn't (so far, AFAIK) require the 
> use of them.

My complaint (if that's what it is) is that in the case of the HTTP binding
that we present in our spec, based on the description that we give, a
binding user cannot make successful use of the SOAP-Response and
Request-Response MEP without also understanding the Web Method feature. This
amounts to it cannot use the HTTP binding based on knowledge of the
supported features it does understand, without knowing that its using the
HTTP binding... which IMO destroys the whole point of having the framework
(which is why I feel quite so passionately about this one).
 
That said, I also agree with Noah on the pragmatics of this [1], in that the
change I have requested [2] will make no practial difference to any
implementations, only on the way we describe it - which is why I would
readily regard it as an editorial change.

Regards,

Stuart
[1] http://lists.w3.org/Archives/Public/xml-dist-app/2002Jul/0034.html
(third bullet).
[2] http://lists.w3.org/Archives/Public/xml-dist-app/2002Jul/0000.html

> -----Original Message-----
> From: Jacek Kopecky [mailto:jacek@systinet.com]
> Sent: 04 July 2002 22:57
> To: Mark Baker
> Cc: Williams, Stuart; 'xml-dist-app@w3.org'
> Subject: Re: FW: LC Comments: Web Method Feature
> 
> 
>  Hi all (the usual suspects), 
>  Stuart pointed me to this thread asking for my thoughts. Again 
> reading it has been enlightening to me on the issues about 
> layering and application protocols.
>  I agree with Mark that the web method specification should be 
> mandatory when using the HTTP binding so that the tunnelists 
> know they are in fact tunneling (by always specifying POST).
>  I agree with Stuart though that once we call something a 
> feature, it's not mandatory, it MAY be used to gain something. A 
> binding provides features, it doesn't (so far, AFAIK) require the 
> use of them.
>  Therefore I see a simple solution like this:
>  Say explicitly that the HTTP binding requires the WebMethod 
> feature to be used. Nothing forbids a binding to do so, I 
> believe. This should satisfy Mark (and Noah) in keeping the HTTP 
> stuff explicit; it should also satisfy Stuart (and Marc) in 
> making the spec clear about the dependencies between parts of it. 
> I think the inconsistency in dependencies was Stuart's original 
> issue. We might also want to say something about the limiations 
> in the possible WebMethod and MEP combinations.
>  Now off to my vacation. 8-)
> 
>                    Jacek Kopecky
> 
>                    Senior Architect, Systinet Corporation
>                    http://www.systinet.com/
>

Received on Friday, 5 July 2002 09:30:02 UTC