Request for Comments: XML binary encoding

["Al B. Snell" <alaric@alaric-snell.com>]

1) The binary encoding itself, which replaces XML, should be a logical
superset. I would suggest that entity declarations can appear at any point
(like PIs) and either have scope from that point in the document order
onwards, or lexical tree scope in document order within the current node
and its children. The format will probably resemble a direct serialisation
of a SAX event stream. There will be atoms (strings, numbers, symbols[0],
etc)

2) The parser and serialiser will have SAX-like interfaces that do not
directly emulate SAX due to the differing underlying representation, but
are similar in spirit. There will be two filters between this and SAX -
one for reading and one for writing - which embody an "XML adaption
layer" on top of the binary encoding to encode the parts of XML that are
not directly represented (DOCTYPE declarations, for example - apart from
the entity delcarations therein)). I would imagine that the XAL would
involve placing the XML document inside a top-level XML element (in the
XAL namespace) containing attributes for the XML version number and the
<!DOCTYPE ...>, if any.

There would be no need for an encoding attribute - UTF-8 all the way, with
all character entities expanded so there are no special characters in the
strings; they can be processed literally.

3) There would also need to be a DOM replacelement. Existing DOM
implementations can be used with the DOM tree built from an XAL SAX
wrapper, but to use the advanced beyond-XML features for specialist
applications (or, dare I dream, if this thing supercedes text-based XML
;-), a DOM that offers access to the nicer aspects (such as directly
accessing numbers rather than the XAL having to convert them to strings
which the application then converts back to numbers...) would need to be
defined.

4) A catchy name is needed for the above three. netXML, netSAX, netDOM -
emphasises the smaller size makes it more suitable for many of the
networked applications of XML. XMLplus, SAXplus, DOMplus - also catchy,
but isn't there already a DOMplus?

I would ideally like to see a standard such as this reach the W3C, but
if not, I can submit it as an RFC in some month's time. Until then,
discuss...

[0]Symbols are oft-used strings that are declared once and then referred
to by a number - like my namespace IDs in an earlier proposal:

http://lists.xml.org/archives/xml-dev/200104/msg00175.html

Symbols are used for entity names, element names, attribute names, and so
on.

ABS

-- 
                               Alaric B. Snell
 http://www.alaric-snell.com/  http://RFC.net/  http://www.warhead.org.uk/
   Any sufficiently advanced technology can be emulated in software