- From: David Carlisle <davidc@nag.co.uk>
- Date: Sun, 19 Aug 2012 11:30:35 +0100
- To: John Cowan <cowan@mercury.ccil.org>
- CC: "public-microxml@w3.org" <public-microxml@w3.org>
On 18/08/2012 07:02, John Cowan wrote: > David Carlisle scripsit: > >> This last statement is only true if the micro-xml data model is >> going to include properties for whitespace control and the other >> things. > >> If micro xml users need not invent new conventions for whitespace >> control and can go xml:space="preserve" then the micro xml data >> model had better have properties on text nodes to say that that are >> or are not ignorable white space. > > I still can't understand why you think this. > > There has never been any representation of language in any XML data > model, yet xml:lang is probably the most widely used of the four > attributes. > XML itself takes mostly a syntax-only view, which allowed different data models to build up. But it also necessitated a host of extra specifications forming the whole xml stack. If xml:id is going to be allowed then to avoid any confusion about what it is supposed to mean the micro-xml spec should say something that means that it should make fragment identifiers and the id() xpath function do the right thing. I can't see how it could do that without saying something equivalent to saying that the data model supports attributes to denote at least ID attribute typing. Similarly for white space and xml:space. It is possible of course for the data model not to help here and the attribute data just to be recorded as attributes. But then you are paying the price of pre-defining fixed attribute names with no benefit. If xml:id isn't going to infer any automatic ID property and basically it works in the same way as id (or any other attribute) then many language designers (and I'm sue most users) would prefer just to use id= rather than xml:id=. David
Received on Sunday, 19 August 2012 10:31:09 UTC