RE: [xml-dev] "Introducing MicroXML, Part 1: Explore the basicprinciples

[David Lee <dlee@calldei.com>]

> I forgot to mention one other point.
> 
> I think there are many situations where it would be useful to have a
> single byte stream contain a sequence of MicroXML documents, eg log
> files or XMPP-like applications.  If you allow PIs both before and
> after the root element, you can't unambiguously parse such a byte
> stream because you don't know whether a PI between two root elements
> goes with the preceding or following element.
> 
> James
> 

I started a few years ago working on some kind of spec/proposal to allow sequences of XDM.
I think suporting sequences of MicroXML would fit well with this.
But I came to the conclusion that to support streams of XDM (or XML ) that there needs to be
a separator or start or end character (or all 3) which is invalid in XML and any text serialized XDM.
This allows layering of processing so that the code that recognizes the units don't have to be include full XML parsers.
This allows you to write things like cat, split, head ,  NofM extract etc with very simple code that doesnt need to be XML aware.

http://xml.calldei.com/XDMSerialize

I'm still debating on the details, in particular if there really needs to be a "KIND" enumerator or if "duck typing" can be used.
I plan on implementing something like this in xmlsh and would love to add MicroXML as an option.
In fact if MicroXML is to be fully supported in XML then ideally it should be parsable into XDM.   Then I can still use Saxon Xdm as the 
data model and MicroXML as simply an alternative text format.
I recall that a goal of MicroXML is that it be readable by a full XML parser.  But I have not followed the discussions enough to determine
if it can be parsed into XDM ... and how similar the result would be if the same text stream were parsed as XML into XDM.




----------------------------------------
David A. Lee
dlee@calldei.com
http://www.xmlsh.org