Hi again Jochen.
Thanks for the quick update yesterday. It works, but only after i did a change
in the autogenerated code (i use jdk 1.4.2).
In the XXXHandler class (sax parser), there is a cast missing. So where one has
in the endElement method:
...
_2 = new org.my.package.MyObject[]{pResult};
...
_3[_2.length] = pResult;
...
should be:
...
_2 = new org.my.package.MyObject[]{(org.my.package.MyObject)pResult};
...
_3[_2.length] = (org.my.package.MyObject) pResult;
...
This for the case where i have as a child of this element type a sequence of
complexTypes, annotated to be an indexed array.
Cheers,
Ricardo
PS: Is this the place to report bugs?
Jochen Wiedmann wrote:
Ricardo Brito Da Rocha wrote:
I'm actually using both 1.1 and 1.2 (though i bumped into the issue
using Axis 1.1).
Understood. The problem is with the constructor QName(String, String,
String), which is missing in the Axis implementation.
However, the same constructor is present in the Java 5 version, in the
JAXB version (it wasn't originally there, but was added at some point),
and in the JaxMe version.
Sorry to say, but there is nothing I can do. That constructor was
introduced with good reason and is, indeed, used by JaxMe. This problem
can only be fixed on the Axis side by adding the constructor.
I think this could be very useful. I'm defining an XSD version of a
schema we currently have for LDAP. I'm going with an initial Java
binding, but i can see people trying to do the same for C++. All the
JAXB annotations make it all a bit confusing to those not interested in
the Java binding.
I also do believe that it could be useful. It's simply a low priority
item, because it is a feature, which makes things handy, but doesn't add
actual power.
As for a C++ implementation: A real lot of things could be reused, if
someone had the idea to do this with JaxMe. The main problem would, of
course, be, that C++ programmers will surely prefer an implementation,
which is written in C++ itself. :-) But, believe me, reading the schema
and converting it into a suitable object model is about 50% of the job.
(Mainly, because XML schema is such a terrible beast ....)
I don't know if your paragraph meant a "go for it"... i would be glad to
do it, but it would take some time to get into the implementation
details first :)
That would, of course, be excellent. :-) Feel free to use this list for
discussion of details or questions on the specification.
Jochen
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
