Re: Kinds of iteration

On Jan 10, 2006, at 22:21, Norman Walsh wrote:
> / Robin Berjon <robin.berjon@expway.fr> was heard to say:
> | On Jan 10, 2006, at 18:24, Richard Tobin wrote:
> |>> 1. Iteration over a sequence of elements...
> |>
> |> Only elements?  I would have thought that we would want to allow  
> text
> |> nodes, and (for completeness) comments and PIs.  Attributes are  
> a bit
> |> more difficult, but XSLT manages to treat them as first-class.
> |
> | My thoughts exactly. I hesitate to extend this further to arbitrary
> | DM sequences, but I can't think of an argument not to.
>
> Arbitrary sequences of nodes, I hope you mean. For simplicities sake,
> I'd like to say that what gets passed between processes are XML nodes
> at least, not atomic values like "5".

That's the gist of my hesitation. On the one hand I would much prefer  
the simplicity of arbitrary sequences of nodes (and, where my  
personal use cases are, that is all that I need), on the other hand I  
dislike the idea that a core technology such as XQuery would be only  
partially supported in that some of its outputs could not be passed  
along the pipeline. Given a choice I'd pick the former, but if  
there's a simple way in which the latter could be supported (e.g. by  
"downgrading" them to a sequence of text node for values that are not  
already nodes) or could allow extensions for them (e.g. for version  
2) then I'd feel on much safer ground. It's hard to see just how  
popular XQuery might get in a possibly rather short term, but if it's  
a raging success I'd hate XPM to come out to meet with frustration  
from that community.

-- 
Robin Berjon
    Senior Research Scientist
    Expway, http://expway.com/

Received on Monday, 16 January 2006 12:22:29 UTC