[
Lists Home |
Date Index |
Thread Index
]
> From: Bullard, Claude L (Len) [mailto:clbullar@ingr.com]
<snip/>
> 1. We don't need a lot of implementations of XML Schema.
Although it is certainly good to have some healthy competition, and for
developers to have some choice.
> 2. Where we need them, we need them to be compatible, otherwise,
> many will do as they have done before and pick one
> product, and it
> will be MSXML for a large number of users.
>
> 3. If the XML.COM article is right, and I think it is, Web services
> need XML Schema to work. They only need compatibility across
> implementations.
Absolutely. And we don't need another language-neutral standard API. There's
no reason why more vendors can't innovate and provide added value to
developers beyond the standards, as Microsoft has done with MSXML4 and Sun
did with MSV. In the Java world, the popularity of alternatives to the DOM
is a clear indication that developers value standards compliance for
interoperability, but don't want to be constrained in their implementations
by lowest common denominator standards.
> This is a tool vendor survival issue. No one thrives on being
> incompatible. The users can push as hard as they want but
> the fastest fix will come if the Altovas of the world test
> their implementations in concert with the MSXMLs of the world
> and get it right fast. I should think this would become an
> immediate concern of the WSIO.
I agree. All of the web service toolkits rely upon XML Schema. If
implementations don't agree, web services won't work.
I was able to throw together a simple proof-of-concept demo for managers,
here, showing how we can take existing integration interfaces that predate
SOAP, and easily support data-binding toolkits that provide
language-specific APIs leveraging existing skills of developers that don't
know XML. I used XSLT to adapt an existing XML-based interface to SOAP, and
wrote a WSDL file and XML Schema, ran that through Microsoft's .NET proxy
generator, and created C# classes that encapsulate the XML-based interfaces
(and there is no RPC involved here, BTW). It was a breeze. But of course, if
vendors' implementations for XML Schema don't agree with each other, this
all falls apart. If different vendors implementations *do* agree with each
other, then we can do the same for Java or any other language, and provide
APIs adapted to the existing skill sets and preferences of the developers
rather than demanding they adapt to one specific approach to implementation.
Developers are in control of their own implementations (which is the way it
should be), while interoperability is still achieved because internally
their implementations are interpreting and supporting the schemas in the
same way.
|
