[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

RE: Attribute Order

• To: xml-dev@l...
• Subject: RE: Attribute Order
• From: "Champion, Mike" <Mike.Champion@S...>
• Date: Fri, 7 Dec 2001 16:03:06 -0500

-----Original Message-----
From: AndrewWatt2000@a... [mailto:AndrewWatt2000@a...]
Sent: Friday, December 07, 2001 3:31 PM
To: xml-dev@l...
Subject: Re:  Attribute Order




> If there are attribute sequence *requirements* then I don't think the
document is XML, 
> although it may be similar in all other respects. By definition, in XML,
attribute order 
> is not important.  

That's a bit strong ... the data is XML if it's syntactically well-formed.
The problem is simply that application is not going to interoperate cleanly
with other XML applications if it takes advantage of syntactic structures
that aren't guaranteed to be preserved by an XML parser or the InfoSet.   In
other words, the application is fragile, but it's not illegal. You're not
breaking the rules of XML by treating attribute order as significant, you're
just relying on the behavior of some tool rather than the behavior specified
in the XML contract.

For example, an application might choose to treat attributes whose values
are in single quotes differently from those in double quotes, or might
interpret <null></null> different from <null/>.  That's fine if the
application totally controls the data format all the way down to the
character level, and is using XML for some reason that doesn't involve data
interchange (for example, simply to avoid having to write a parser for some
internal file format).  It's obviously an extremely bad idea if the XML will
ever go through some other processor that the application doesn't control. 



 

• Follow-Ups:
  • Re: Attribute Order
    • From: David Brownell <david-b@p...>

• Prev by Date: Re: Attribute Order
• Next by Date: Re: Attribute Order
• Previous by thread: Re: Attribute Order
• Next by thread: Re: Attribute Order
• Index(es):
  • Date
  • Thread