axis-java-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Brenda Bell <bb...@juicesoftware.com>
Subject RE: Support for xsd:anyURI - Proposal
Date Fri, 26 Jul 2002 23:04:32 GMT
>From a user's perspective, this has been an interesting thread so I thought
I'd throw in my two-cents.  Be it the right or wrong, temporary or final, I
for one would like to see any type of support for anyURI as long as it's
reasonable... which is why I've plugged in a rudimentary implementation of
my own to get me over my current crisis.

I agree that it's important to do the right thing, but I'm starting to see
WSDL's that use this type and I expect to see it occur more often.  I don't
particularly think that time is on your side on this one :)

-----Original Message-----
From: Glen Daniels [mailto:gdaniels@macromedia.com]
Sent: Friday, July 26, 2002 6:45 PM
To: 'axis-dev@xml.apache.org'
Subject: RE: Support for xsd:anyURI - Proposal


> Glen Daniels wrote:
>  > Typing is critically important when passing data around.  
> If you have
>  > an AxisURI class, you can unambigiously know how to 
> serialize that on
>  > the wire in exchange for a little added complexity setting/getting
>  > values.  I think this is the right way to go - otherwise you risk
>  > major interoperability problems unless you are VERY 
> careful with your
>  > metadata.
> 
> The very real risk exists that every component that one deploys in a 
> system defined their own mapping of XML concepts to their own 
> component 
> specific type.  I would suggest that if we don't find an 

That is absolutely true, and we DO want to allow people to do their own
thing to some extent - if they want to explicitly remap xsd:anyURI for
their service, they should go for it!  All I'm asking is that we're as
careful about our default typing as we'd hope they would be with any
custom stuff.

> existing Java 
> construct to map to, then we should work with the JAX RPC 
> experts group 
> to define one.

+1, although I see nothing wrong with defining our own until such time
as JAX-RPC revs (which can also serve as input to the JAX-RPC process).

--Glen

Mime
View raw message