Re: Goal: POJO-based

From: Paul Sandoz <Paul.Sandoz_at_Sun.COM>
Date: Mon, 23 Apr 2007 14:54:06 +0200

Stefan Tilkov wrote:
> On Apr 23, 2007, at 9:34 AM, Hao He wrote:
>> hi, Jerome,
>> I guess that the issue is more at the architectural level rather than
>> at the API level. Surely you can model resources using POJOs, but it
>> will leak Java semantics. This would cause interoperability problems
>> to others on other platforms. I know I am very XML biased, but isn't
>> the idea of XML is to have a platform/language independent data format?
> But we're not concerned with language independence, are we? After all
> this is a Java API.
> I have some understanding for your concerns, given that I positively
> hate the approach taken by most Web services toolkits, including .NET
> (defining service contracts in the programming language, i.e. code-first
> vs. contract-first). But we don't *have* a contract -- I have no
> objections to WADL as long as it's a by-product and not a pre-requisite.

I agree. In my experience language artifact leakage is something that
occurs from using XML schema binding tools (which also leak their own
artifacts) when using the code first model with those tools. The best
practices for the use of XML in conjunction with XML schema data binding
tools (using code first or schema first) seems orthogonal to this API.
The XML Schema Patterns for Databinding [2] is providing useful work in
this area.

Some other points below that people may find interesting.

See Bill de hÓra blog [2] for interesting observations about XML in the
wild (and also Sean McGrath's article linked from it).

Also Mark Baker made some very interesting observations at the W3C Web
of Services workshop [3]:

"The point I was trying to make in my presentation, about SOA/WS not
separating interface and implementation, must have gone totally over
peoples’ heads because nobody commented on it, either to agree or disagree"

" - WSDL not part of runtime architecture of Web services systems

     - If you delete *.WSDL what breaks?



| ? + ? = To question
    Paul Sandoz