Re: Umbrella Specifications Discussion

At 06:21 PM 6/15/2005 +0200, Dominique Hazaël-Massieux wrote:
>Le mercredi 15 juin 2005 à 10:12 -0600, Lofton Henderson a écrit :
> > I don't think "different advancement" is necessarily the issue.  The issue
> > still exists when they (multiple parts) are all done, if either:
> > -- they contain multiple conflicting conformance bits;
> > -- or, they contain no clear conformance bits at all, that allow you to
> > draw conformance conclusions about the collection as a whole or common
> > conformance concepts that apply to multiple parts.
> >
> > "Umbrella specification" is a handle for dealing with that.
>
>OK, if I understand correctly, you're saying that ViS should (or may?)
>address the need for a technology developed as a collection of documents
>to define in a well-known document (the umbrella specification) how the
>documents in this collection interact conformance-wise?

Well, I guess ViS may be an odd place for it.  It does not pertain to "how 
conforming implementations may vary amongst themselves".

To vastly oversimplify, I think "umbrella specifications" is no more than a 
term that we invented and used to cover some cases of our requirement, "You 
must have a conformance clause".

-Lofton.


>In other words - and this is consistent with what Karl was doing as well
>-, we would introduce this concept as a way to name the results of what
>we think is a good practice (usability-wise for Karl, conformance-wise
>for you). I guess I'm not sure there is much benefit in introducing the
>term, but there is no big harm either, so I wouldn't object to that.
>
>Then the question is: does it fit in ViS? and if so, where? I'm pretty
>confident that if we want to keep it there, it shouldn't stay where it
>currently is (between levels and deprecated features).
>
>Dom
>--
>Dominique Hazaël-Massieux - http://www.w3.org/People/Dom/
>W3C/ERCIM
>mailto:dom@w3.org
>

Received on Wednesday, 15 June 2005 16:42:25 UTC