RE: XML 1.0 - Element order significance

[Danny Ayers <danny@panlanka.net>]

Hi,

I read the XML Rec to mean that there's no guarantee of order information in
an XML document - as has been pointed out, if the order is required by the
DTD, then no additional information is conveyed by the elements being in
order (apart from document validity, I suppose). Otherwise, no additional
information can be directly conveyed by the order of elements in an XML
document (all other things being equal) according to the XML Rec. It may be
possible for applications to read/write additional information through
order, but this is beyond the scope of the XML Rec.

Having said that, if the document has an infoset, then information in the
form of element order may be conveyed through this :

----------------------
From : XML Information Set
W3C Working Draft 20 December 2000

<in the introduction>

An XML document has an information set if it is well-formed and satisfies
the namespace constraints described below. <which says essentially that it's
namespace name is absolute>

<in 2.1. The Document Information Item>

The document information item has the following properties:

[children] An ordered list of child information items, in document order.


from http://www.w3.org/TR/xml-infoset
------------------------

note the word 'ordered'

Cheers,
Danny.

<- -----Original Message-----
<- From: John Cowan [mailto:jcowan@reutershealth.com]
<- Sent: 22 January 2001 22:26
<- To: Tim Bray
<- Cc: XML Developers List
<- Subject: Re: XML 1.0 - Element order significance
<-
<-
<- Tim Bray wrote:
<-
<- > [I]f the doc has
<- >
<- > <brothers><matt /><jim /><luke /><dave /></brothers>
<- >
<- > then if I don't see them in the order (m, j, l, d) that they
<- > appear in the document, something is wrong.  Software that
<- > rearranged the elements as part of processing would be broken;
<-
<- Absolutely.  Unfortunately, such software is not actually
<- out of compliance with XML 1.0.
<-
<- > As to whether the order is "significant" in the human/design/business
<- > sense, that is totally language-specific, and I don't agree with
<- > Henry that you can reverse-engineer it from the DTD. -Tim
<-
<- I think HST's point is sound, though his wording
<- may be unfortunate.  Let me attempt to paraphrase:
<-
<- a)
<- If the elements MUST appear in a specified
<- 	order (according to the DTD or whatever),
<- 	then the order CANNOT convey any additional
<- 	information to the application.
<-
<- b)
<- If the elements MAY appear in any order (again
<- 	according to the DTD or whatever), then the order
<- 	which actually appears MAY convey additional
<- 	information to the application.
<- --
<- There is / one art             || John Cowan <jcowan@reutershealth.com>
<- no more / no less              || http://www.reutershealth.com
<- to do / all things             || http://www.ccil.org/~cowan
<- with art- / lessness           \\ -- Piet Hein
<-