Dependent on your goals, perhaps variations on the theme of a sorted list of all data paths might be useful? For example: considering the paths of all simple elements and attributes, including indexes, and representing text content as an appended last "step" - à la /a[1]/b[2]/c[2]#my text value.

I use this in test cases, comparing expected and found instances of complex structures; with any difference expressed in terms of (a) paths only in A, (b) paths only in B, (c) paths in both but with different text value.

Graydon Saunders <graydonish@gmail.com> schrieb am 14:38 Mittwoch, 19.Oktober 2016:

Hi Christian --

I was after a better, or at least less convoluted, way to add the UID attribute nodes. Your code snippet works in place, and faster; thank you!

If anyone happens to have a good general case structure-but-not-position document compare algorithm lying around, I'd be delighted to hear about it.

(It can't be just XSD because there's a pile of WSDL documents, too. And who knows what in the future.)

Thanks!

Graydon

On Tue, Oct 18, 2016 at 3:28 PM, Christian Grün <christian.gruen@gmail.com> wrote:

> Is there a better way to do this?

Your query could probably be shortened as follows:

let $new := db:open($newName) update {
.//* ! (insert node attribute UID { generate-id() } into .)
}
...

It does the same as your original query. But maybe you were asking for
a better solution to compare documents against each other?