On 11 Nov 2004, at 14:43, Bill de h�ra wrote:
Anyway, this won't cater for the situation where the publisher wants to insist you do not process the document without be able to process a specified namespace. Tim's proposal does.
I was speaking particularly about PaceExtendingAtom [1], and was not commenting about
PaceMustUnderstandElement [2] where Tim Bray describes what you are mentioning. My proposal does not, on first glance, contradict PaceMustUnderstandElement.
The problem I have with Tim's Proposal on PaceExtendingAtom is that it limits where new
attributes can appear. I want them to be able to appear anywhere: in the feed, in the entries, etc... I don't see why the specification should try to limit this, other than
state that since this is an Atom document, any document containing externally namespaced elements (without a MustUnderstand element) should be a monotonic extension of the stripped document.
I am sure that with a little effort we can work out how to put this in one clear crisp sentence.
Then you should do that :) But even if you did, I doubt that you will be able to constrain things in the wild to be suitably monotonic.
Thanks for "monotonic". I see you understand my proposal much better than you will admit. :-)
So perhaps I should rephrase:
You can extend Atom with foreign namespaced elements and attributes in any position, as long as the following conditions are met:
1- if all those foreign name spaced elements are removed from DocOrig, you end up with a valid Atom document (call it DocMin)
2- DocMin should be a well formed Atom Document as stated in this spec
3- unless a mustunderstand element is present, a parser parsing DocMin is correctly
parsing DocOrig
Henry
[1] http://www.intertwingly.net/wiki/pie/PaceExtendingAtom [2] http://www.intertwingly.net/wiki/pie/PaceMustUnderstandElement
