RE: ISSUE-74: Relative URIs in targetNamespace schemaLocation

Hi Paul

> Am doing what I can to keep up with what you are working on.  

Great! 

We've four deliverables:

1) Basic Patterns for Databinding specification
    (in preparation for Last Call)

2) Advanced Patterns for Databinding specification
    (first Working Draft almost ready)

3) a tool for detecting patterns in schemas and WSDLs
    (about to be published)

4) a test suite to demonstrate support of the patterns in databinding tools
    (example documents wrapped up into WSDLs and
     fired at tools. Reports about to be published)

> I don't know the details of this thread, 

It's a new Issue, so no thread, yet. Apart from this one :-)

> but I will say that FWIW, I don't think relative
> URIs should be classified "advanced".  

Ah, I'll explain our criteria:

The goal of Basic is for someone authoring a schema who wants to 
reach as wide a market place as possible.

For a pattern to be "Basic" it has to have very widespread demonstrable
interoperability using today's "state of the art" toolkits. Interest 
and experience within in the Working Group covers several older technologies
and databindings for non-OO mappings, so that bar is very low. 
Basic will indeed be very basic.

The goal of Advanced to promote better support of commonly used features
in implementations and improve "user experiences" of using standard
schemas with tools.

Patterns with demonstrable use and utility in schemas in the wild
but which don't work well with today's databinding tools are "Advanced".

We have still to agree on why we wouldn't document a pattern
as Advanced under ISSUE-62:
http://www.w3.org/2002/ws/databinding/issues/62/

> In my view of many libraries of
> schemas, some open standards and some not, a strong majority of them utilise
> relative paths in the schemaLocation.  This is often because they want their
> libraries to be self contained and available offline (like trying to edit a
> library of schemas on a plane, which I am oft to do...).  The relative paths
> allow them to connect entire libraries and ship them around to colleagues
> and business partners who may or may not be online.  

So unless we can demonstrate uniform support for xml:base, 
I still see this as "advanced".

> My 0.02.

Valued more than that :-)

Paul

Received on Thursday, 19 October 2006 07:27:55 UTC