xml-dev - RE: Canonical XML, signatures, and element ordering

[Michael Brennan <Michael_Brennan@Allegis.com>]

XML Signature was probably the wrong thing to cite here. (What can I say.
I'm a novice with much of this stuff, and its hard to keep up and keep it
all straight.) I think DOMHASH is what I really meant to cite. Just looking
for a reliable and efficient way to compare XML documents within a
controlled environment. The documents are not being sent over public
networks; message integrity is not really the concern. Also, there is no
need to translate between schemas in this instance.

The XML documents are dynamically generated by software, and we're exploring
automated ways to do some regression testing to make sure when the software
is upgraded, we are still getting the expected XML responses to known
requests. The XML messages will often represent a set of database records in
response to a query. If the query didn't specify a sort order, then the
ordering of elements that represent records will not be significant.

Currently, this is a very manual process. We'd just like to be able to
automate it more.

Thanks.

> -----Original Message-----
> From: Michael Brennan [mailto:Michael_Brennan@Allegis.com]
> Sent: Thursday, July 20, 2000 2:39 PM
> To: xml-dev@lists.xml.org
> Subject: Canonical XML, signatures, and element ordering
>
>
> I'd like to be able to compare two XML files to determine if they have
> equivalent content. The catch is, in some instances (but not
> all) the order
> of elements under a particular node is not relevant. It's my
> understanding
> that XML Signature will meet my needs just fine for instances
> where order is
> relevant, but not otherwise. Is that correct?
>
> Are there any suggestions on a good way to do this? The only
> thing that
> comes to my mind is to convert each XML document to its
> canonical form, then
> load it into a DOM tree and procedurally sort the elements under the
> relevant node. Would this approach work or are there "gotchas" I'm not
> considering? Is there a better way?
>
> Thanks for any tips on this.
>