RE: Is inclusion stable?

The WG does not want to preclude streaming implementations.  To that end
XInclude places no requirements on whether URIs fetched twice return the
same representation or infoset.  We'll add a statement clarifying this
to the spec.  Resources and/or their representations can change over
time.  Two conformant XInclude processors might not generate the same
results if they are run at different times.  A single XInclude processor
might not generate the same results on two different runs.  In the face
of this, trying to ensure full compatibility between two XInclude
processors run at the "same time" doesn't seem to buy us much.

So, if you want to cache it, go ahead.  If you don't, that's fine too.

> -----Original Message-----
> From: www-xml-xinclude-comments-request@w3.org
[mailto:www-xml-xinclude-
> comments-request@w3.org] On Behalf Of Elliotte Rusty Harold
> Sent: Monday, January 26, 2004 4:27 PM
> To: www-xml-xinclude-comments@w3.org
> Subject: Is inclusion stable?
> 
> 
> I was wondering if there should be a requirement that XInclude be
> stable? In other words, if the same URI is fetched twice in the
> process of resolving includes, must/should the representation
> returned be identical both times? For example,
> 
> <root xmlns:xi="http://www.w3.org/2003/XInclude">
>    <xi:include href="http://www.cafeconleche.org/"/>
>    <xi:include href="http://www.cafeconleche.org/"/>
> </root>
> 
> What if the resource changes between two retrievals? e.g. it has a
> hit counter of some kind in the document text? or a time stamp? Or
> perhaps just a frequently changing stock quote?
> 
> Are implementations allowed to cache the result of the first
> retrieval? Are they prohibited from doing so? Is it implementation
> dependent?
> 
> Then the same question, but the URIs have different fragment IDs; e.g.
> 
> <root xmlns:xi="http://www.w3.org/2003/XInclude">
>    <xi:include href="http://www.cafeconleche.org/#1/3"/>
>    <xi:include href="http://www.cafeconleche.org/#/1/2/4"/>
> </root>
> 
> Suppose caching is not required, or even prohibited. Is there any
> processing order? i.e. is an implementation required to process
> XIncludes in document order? or could it move in reverse order even
> if this would change the result infoset?
> 
> What if the same resource is loaded with different accept attributes
> and resulting headers?
> 
> I'm not sure what the answers are or should be but some guidance from
> the spec would be helpful on these points. I prefer this not to be
> implementation dependent, but if must be, then it should be noted as
> an implementation dependent issue.
> --
> 
>    Elliotte Rusty Harold
>    elharo@metalab.unc.edu
>    Effective XML (Addison-Wesley, 2003)
>    http://www.cafeconleche.org/books/effectivexml
> 
>
http://www.amazon.com/exec/obidos/ISBN%3D0321150406/ref%3Dnosim/cafeaula
it
> A
>

Received on Wednesday, 18 February 2004 12:18:09 UTC