> > Of course it doesn't. The biggest single failing in the XML
> spec, sadly
> > unremedied in subsequent versions, is that it doesn't
> define the information
> > content of an XML document. Without that "lossless" cannot
> be defined, and
> > it can't be defined, then it can't be achieved.
>
> Certainly it could be achieved. You could define a 1-1
> mapping from the
> binary format to the actual bytes or characters of an XML
> document. This
> would preserve a lot of things most people don't care about like
> attribute order, numeric character references, and white space in the
> prolog, but it would preserve everything that anyone cares about.
Except the base URI and the media type, perhaps?
Michael Kay
