Xerces-J 1.4.4 bugs

[David Sheldon <dw...@decisionsoft.com>]

Firstly I will point out that I am aware that Xerces-J 2 is the
supported version, and that 1.4.4 is now in hibernation. However, we
along with, I am sure, many other people are still using 1.4.4, not
least because the error messages that it gives for schema errors, tend
to be more readable to the XML novice than those generated by Xerces-J
2.

Well, having said that, there are a couple of problems. The first is an
error message bug. The toString method of XMLContentSpec, will
erroniously report that the content expected is "##any:uri=......" when
it is actually just an element. This is fixed by the attacked CSpec.diff
file. 

The second is that in some cases, where there is an ANY in a top-level
element. If, in an instance, it contains elements in a different
namespace, then it would sometimes cause a null-pointer exception in
SchemaGrammar.getElementComplexTypeInfo, as if you try accessing a
wildly out of range element (such as when you are in the process of
switching grammars), the implementation of the two dimensional array
objects when you access it. I have patched just the
getElementComplexTypeInfo function, and that seems to solve it, however
there are several other functions that might have a similar problem. 


Basically, I would like other people to be able to gain from the time I
have spent debugging these problems. Is there a way to submit them to
CVS in some useful manner? or should I bug report them and attach the
patches so that people searching bugzilla might find them? The final
option is that I could put them on my webspace so that google might find
them. Do you guys have any suggestions as to what I do with them? The
problems appear to be fixed in both Xerces-J2 and in Xerces-C++.

David
-- 
David Sheldon, Client Services        DecisionSoft Ltd.
Telephone: +44-1865-203192            http://www.decisionsoft.com