Thanks - once I realized that the java:whatever hack was castor-specific, I figured it was probably a lost cause.

I suspect that the right thing to do is to backpatch old code to deal with my new data.

Cheers,
    -mik

Stephen Bash wrote:
Michael-

Looking through the Marshaller code I don't see anyway to force xsi:type to be output.  There is probably some combination of tricks you can use to make all the "drop xsi:type" conditions false, but it didn't seem obvious to me how you would do that.  Looks like most of the time is spent seeing if the xsi:type can be avoided rather than if it should be put in.  Sorry.

Stephen


On 3/24/06, Michael Thome <[EMAIL PROTECTED]> wrote:
In the name of backwards compatibility, I'm looking for a way to force my (now hand-written) Descriptor to marshal with an attribute like
    (a) xsi:type="java:com.bbn.MyClass"
instead of
    (b) xsi:type="prefix:my-class"
or
    (c) no xsi:type attribute at all

(a) is what my old xml data model marshalled as (introspective simple class), (b) is what I get if I don't set up a mapping or pre resolve my class, (c) is what I get if I do pre-resolve.

The problem is that old code (using plain old introspection) is baffled by (b) and (c), so "unmarshals" the forms as AnyNode structure, whereas both old and new code are happy to use the java:classname variations.

Thanks,
    -Michael

--
Michael Thome
BBN Technologies 10 Moulton St, Cambridge MA 02138 USA
phone: +1 617 873 1853
------------------------------------------------- If you wish to unsubscribe from this list, please send an empty message to the following address: [EMAIL PROTECTED] -------------------------------------------------


--
Michael Thome
BBN Technologies 10 Moulton St, Cambridge MA 02138 USA
phone: +1 617 873 1853
------------------------------------------------- If you wish to unsubscribe from this list, please send an empty message to the following address: [EMAIL PROTECTED] -------------------------------------------------

Reply via email to