Re: towards consensus on fixup, part 1

/ ht@inf.ed.ac.uk (Henry S. Thompson) was heard to say:
| There was interim agreement on two things which I think we should
| note, before moving on to the harder stuff
|
| 1) All serialized output from conformant processors *must* be
| well-formed and namespace-well-formed *XML* [i.e. impl-defned whether
| 1.0 or 1.1];
|
| 2) We need to say something more about what conformant processors
| *must* and *should* do wrt the information content of what flows
| through.  I'm thinking in terms of something like this:

Do you think it's useful or necessary to do this by appealing to the
Infoset terminology explicitly, or is this just a shorthand for us?

We haven't been making a lot of explicit reference to infoset items in
our spec and to the extent that we can make the section we're now
talking about consistent with the rest of the spec, I think that would
be a good thing.

|  Except where necessary to faithfully implement the semantics of a
|  step, every step *must* carry the information in its input(s),
|  options and parameters corresponding to the infoset properties
|
|   [local name]
|   [namespace name]
|   [children]
|   [attributes]
|   [normalized value]
|   [parent]
|   [owner]
|   [base URI]
|
|  through unchanged to its output(s).
|
|  The information corresponding to
|
|   [prefix]
|   [in-scope namespaces]
|   [namespace attributes]
|   [attribute type]
|
|  *should* be carried over insofar as possible.

                                        Be seeing you,
                                          norm

-- 
Norman Walsh <ndw@nwalsh.com> | The world is a vast temple dedicated to
http://nwalsh.com/            | Discord.-- Voltaire