- From: Mark A. Jones <jones@research.att.com>
- Date: Mon, 19 Aug 2002 16:08:41 -0400
- To: "Williams, Stuart" <skw@hplb.hpl.hp.com>
- Cc: "'xmlp-comments@w3.org'" <xmlp-comments@w3.org>, "'jacek@systinet.com'" <jacek@systinet.com>, "'marc.hadley@sun.com'" <marc.hadley@sun.com>, "'mbaker@idokorro.com'" <mbaker@idokorro.com>, "'moreau@crf.canon.fr'" <moreau@crf.canon.fr>
Stuart, Sorry if my summary was overly cryptic. I believe the editors will probably be clearer in their prose, but they should also note your comment. My clarifications are in-line below. Williams, Stuart wrote: > Mark, > > >>Regarding issue 227 on the SOAP issues list, the following four >>findings obtain: >>* Bindings may specify that features are mandatory. >> > > The lack of any distinction between mandatory 'provision' and mandatory > 'use' is not addressed by this clause. Mandatory provision has always been > an aspect of the framework, so I assume that this should be taken as > meaning: "Bindings may specify that the *use* of particular features is > mandatory." Is that correct? <maj>correct</maj> > > >>* We need to sweep through the spec to ensure that the above point is >> > clear. > > :-) <maj>:-)</maj> > > >>* We will leave 'web method' as a mandatory feature of the http binding. >> > > The status quo is mandatory provision... which is fine. Mandatory use... I > have seen no justification for such a constraint. <maj>'Mandatory' here means provision. 'use' is discretionary.</maj> > > >>* It is possible for a binding to make all features optional. >> > > Again could be clearer about 'use' or 'provision'. <maj>This was added to ensure that a binding could be specified (provision) to have no obligatory features. In this case, 'use' would have to likewise follow 'provision'.</maj> > > >>Mark Jones >>AT&T >> > > With regard to the first bullet, I am ok with it being part of the framework > that bindings may make the use of particular features mandatory. That > restores the position that binding users can make correct use of a binding > based on knowledge of the framework (any realisation of which would include > provision of a the use of a supported feature to be marked as mandatory) and > the binding supported features alone, without *having* to know in particular > what underlying protocol is being bound to. > > This was my prinicple concern in raising this issue... the undermining of > the intent of the framework. > > If mandatory 'use' of the Web Method feature is what is intended by the 3rd > bullet, then IMO this has not been adequately justified... cf. prinicple of > minimal constraint. > > Best regards > > Stuart > > > > > > -- Mark A. Jones AT&T Labs Shannon Laboratory Room 2A-02 180 Park Ave. Florham Park, NJ 07932-0971 email: jones@research.att.com phone: (973) 360-8326 fax: (973) 236-6453
Received on Monday, 19 August 2002 16:09:13 UTC