Michael Kay <mike@xxxxxxxxxxxx>, 2006-09-19 18:06 +0100:

> If that's the only difference in the serialized output, I'm amazed. Many
> aspects of serialization aren't defined in the specs to that level of detail
> - for example, the order of attributes is unpredictable. The general rule is
> that if the XML parser treats two documents in the same way, then they are
> equivalent. The correct way to compare results is to put them both through a
> canonicalizer that removes these differences.

The one concern I had was just to ensure that the serialized
output is newline-terminated. For HTML output, running the output
through a canonicalizer isn't an option.

  --Mike