Serialization of XDM

[David Carlisle]

> is it important that
>     $a[1]/foo/bar  is  $a[2]

well it's inportant if the next process in the pipeline needs to do any
kind of query processing, which is presumably the point of wanting to
pass an xdm instance in the first place.

if the sequence appears to the next process as $input then something as
simple as

count($input//bar)

is going to give different answers depending on whether identity is
preserved.

But in practice it's probably only important to know what the answer is
(whether identity is presrved or not) if you pass xml between processes
by serialising and reparsing then typically you end up with new
identities as far as xdm is concerned but this is rarely a problem in
practice.

so if the specification of the serialisation were to specify that you
were serialising a _copy_ of any XDM instance, so in particular every
item of the sequence always gets a new identity I think that would be OK
for the vast majority of cases.

David

________________________________________________________________________
The Numerical Algorithms Group Ltd is a company registered in England
and Wales with company number 1249803. The registered office is:
Wilkinson House, Jordan Hill Road, Oxford OX2 8DR, United Kingdom.

This e-mail has been scanned for all viruses by Star. The service is
powered by MessageLabs. 
________________________________________________________________________