Re: Xinclude test case

At 06:41 PM 12/18/2006 -0500, Chimezie Ogbuji wrote:
[...] Without an explicit pipeline, there is no way to guarantee or mandate 
an appropriate set of XML processing operations before the XML infoset is 
created for transformation via GRDDL.  And if the author *chooses* to use 
XInclude operations within the source document then he/she had better use a 
pipleline to ensure the proper XML infoset is created for the subsequent 
GRDDL transform algorithm otherwise he/she cannot guarantee a 'faithful 
rendition'.  The fact that *some* XML processors can handle XInclude 
implicitely and other can't is evidence of this.


There is no way to make a guarantee in the general case. Within a mandated 
computer environment,
such as I imagine are in places like the Cleveland Clinic and other 
organizations where messaging
is mission critical, it would certainly be possible. Moreover, it would be 
practical to establish
policies within GRDDL-aware processors to insist that all documents be 
fully expanded and
have all XIncludes resolved before running the nominated transformation. 
That just seems like
a rational policy if you want to extract information from a document that 
might contain
transcluded information.

In other words, the spec doesn't preclude making a guarantee, it just 
pushes the assertion
of such a guarantee up to the application layer.

Received on Tuesday, 19 December 2006 05:11:52 UTC