RE: break or exit in xsl?

Hello Mr. Kay,
  I agree with you! 

But I would like to say something about this statement
you said..
"So long as the order of the output sequence
corresponds in this way to the order of the input
sequence, the nodes can actually be processed in any
order or in parallel."

..How practical is it(and is it possible to implement
easily) to process input nodes in any random order or
in parallel? I guess ,ofcouse it can be done.. But
processing input nodes in some random order or
parallel(rather than using some popular or natural
algorithms of tree traversal..) , and then making
effort to generate output matching the document order
could be inefficient! 

Regards,
Mukul

--- Michael Kay <mike@xxxxxxxxxxxx> wrote:
> 
> This statement is actually rather confusing,
> especially when quoted in this
> particular context. It gives the impression that the
> specification requires
> the processor to process the first node first, then
> the second node, and so
> on. In fact this is not the intent. What it is
> really describing is a
> mapping of an input sequence to an output sequence:
> the nodes obtained by
> processing the first node in the input sequence come
> first in the output
> sequence, followed by the nodes obtained by
> processing the second node in
> the output sequence, and so on. So long as the order
> of the output sequence
> corresponds in this way to the order of the input
> sequence, the nodes can
> actually be processed in any order or in parallel.
> For example, you could
> start by processing the last input node to generate
> the final output nodes,
> then the second-last input node, and so on.
> 
> Michael Kay
> http://www.saxonica.com/



	
		
__________________________________ 
Celebrate Yahoo!'s 10th Birthday! 
Yahoo! Netrospective: 100 Moments of the Web 
http://birthday.yahoo.com/netrospective/