Daniel Veillard wrote:
Had originally started down that path but really didn't think slowing down writing for the additional needed checks were worth it at this point. Can always revisit later.For performance reasons, unlike .NET, this change only tracks the namespace on the starting element tag rather than within the entire scope of the element the namespace was declared on.Fine, the problem is only serious when it's on the same element. Propagation of the set of namespace in scope would allow the check in the subtree in a relatively cheap way though (a bit similar to what SAX2 parser does).
Thanks for pointing that out. Its the mix of tab/space indenting. Will fix that.Just double check indentation before commiting it was looking a bit weird in the patch, but that could be an effect of the patch.
Rob