Lists Home |
Date Index |
In article <200505052313.j45NDiK07174@flightlab.com> you write:
>Not so weird. As far as I can tell, other than well-formedness
>and validity errors the XML Rec places no constraints at all on
>what data a processor must report.
Not quite. It does mention a few things that must be reported. For example:
All non-markup characters (2.10)
For validating parsers, which characters are element-content-whitespace
Defaulted attributes (3.3.2)
Names of unparsed entities and their notations (4)
Entities not included by non-validating parsers (4.4.3)
It's fairly clear from this that the authors of the XML spec did not
attempt to exhaustively list what must be reported, but instead noted
the fact in cases where there might be doubt.
>That's what the Infoset Rec is for.
There is no requirement that any system report everything in the
Infoset, but it is useful as a guide to what is meant to be
significant. For example, the fact that element order is significant
but attribute order not is implied by the Infoset making the
[children] property an ordered list and the [attributes] property an