Re: Candidate Requirement for XML Schema 1.1

At 04:41 PM 8/10/2002 -0400, Dave Peterson wrote:

>This to me implies that no new features can be added, since they
>automatically have not yet been implemented.  E.g., precisionDecimal
>must then wait for 2.0.  I don't think this is what people want.

For PR status, you have to demonstrate this anyway. Here is the 
requirement, as stated in the process document:

         each feature of the technical report has been implemented.
         Preferably, the Working Group should be able to demonstrate
         two interoperable implementations of each feature.

>If an existing feature has not been implemented across the board,
>perhaps it is not that it is inherently too complicated, but that we
>did not describe it properly.  Seems a shame to rule it out then.

Across the board would be too high a standard. Given the importance of 
interoperability at the schema level, I think that asking for 3 compatible 
implementations is not.

>I thought one of the purposes of 1.1 was to better describe the
>features so that incompatabilities between implementations could
>be eliminated.  This seems to make more sense than simply saying
>"if we didn't get it right in 1.0, it's obviously not useful and
>hence is hereby deprecated".

I didn't say that these features needed to be interoperable in 1.0 - it's 
rather too late for that. I said they had to be interoperable in 1.1. If 
each successive version of XML Schema continues to lack interoperability, 
it's going to be pretty hard to defend in the user community.

So what I am saying is this: If you can't get something fixed to the point 
that it is compatibly implemented, or if people aren't seeing the value of 
implementing it, then the feature should be deprecated.

Beyond that, I want a well defined language that simply works across at 
least three implementations. Given the number of implementations of XML 
Schema and the role it plays, is that really too much to ask?

Jonathan

Received on Saturday, 10 August 2002 22:30:50 UTC