John Cowan wrote:
It allows the designers of an XML-based protocol not to get involved with the minutiae of "single or double quotes?" "CDATA or <?" etc. etc.
Yes. I've always seen the main benefit of the infoset as providing authors of specifications an easier way to talk about the contents of XML instances, exactly as John mentions. It makes all sorts of sense to use the infoset in this way.
It makes little sense, in the protocol context, to talk about actually interchanging anything but well-formed XML instances. From the point of view of the implementor, the XML processor can be relied on to make the issues of quotes and CDATA and so on go away. -Tim
Simon St.Laurent "Every day in every way I'm getting better and better." - Emile Coue