xml-dev - RE: XInclude [was: XLink transformations]

["Bullard, Claude L (Len)" <clbullar@ingr.com>]

No, just trying to keep the models in my mind 
sorted out as I have to teach an XML course 
here and want to be sure that I am saying 
the right thing.  I guess one can 
do as was done in the semantic DTD which 
was to provide an attribute with the GI of 
the original set ensuring it is preserved 
in the merge.   The XInclude looks fairly 
straightforward to use; basically, merge 
at the point where marked.  Should be useful 
as it is essentially an append from view 
in a relational set plus the notion of 
order given that no order is guaranteed 
for a relational recordset.  

Some XML concepts and techniques have to be 
completely understood to offset the tendancy 
to cast them in terms of one's technical 
background.  IMO, it is essential to 
understand the relationships among 
location and named value as they might infer 
identity where uniqueness, scope and 
persistence characteristics of identity.  
As we discussed earlier, understanding the 
conditions of an operation as an operation 
affects identity is essential.  An XInclude 
is declared but operational.

It is hard to get technical folks to understand 
that XML itself doesn't DO anything. :-)

Len Bullard
Intergraph Public Safety
clbullar@ingr.com
http://fly.hiwaay.net/~cbullard/lensongs.ram

Ekam sat.h, Vipraah bahudhaa vadanti.
Daamyata. Datta. Dayadhvam.h

From: Didier PH Martin [mailto:martind@netfolder.com]

It is my understanding that the merge is occuring in a transient (or could
be permanent) infoset. It could be nearly impossible to undo the operation
if the xinclude processor does not keep track of the inclusion operations.
However, an xinclude may keep the necessary information to undo the
operation.

Is there any case you have in mind where this could be necessary?