Re: Closing Blueberry

[John Cowan <cowan@mercury.ccil.org>]

Tim Bray scripsit:

> - John Cowan has proposed a sensible-looking method for
>   writing the XML NAME rules by reference to Unicode 
>   metadata and thus achieving decoupling from any 
>   particular version of Unicode.  I didn't see anyone
>   raising problems with Johnn's approach, and lord
>   knows there are people here who are qualified to
>   spot 'em if they're there.  Of course to use
>   this fully, your Blueberry declaration would have to 
>   specify which version of Unicode it belonged to.
>   [hmm... <foo xml:unicode="3.1">...</foo>?]

The ideas the Core WG kicked around look like:

	<?xml version="1.0.1"?>	(generic)
	<?xml version="1.0a"?>	("a" is covertly "Unicode 3.1")
	<?xml version="1.1" unicode="3.1"?>

But there are other possibilities.

> As for (b), unless someone is willing to make case 
> for opening up deployed systems to pretty massive 
> breakage in order to simplify the lives of a small
> and shrinking piece of the software development
> world... as I said, if this were a WG and I were 
> chair I'd suggest an evident lack of consensus in 
> favor of this change. -Tim

So it seems.  But why are people who will accept (a) allergic
to (b)?  Once you've opened up, you've opened up, and they
are both character-level issues.

Repeating: the NEL proposal does *not* change the grammar
of XML: what it does is add NEL and LS (U+2028) to the
list of things that are accepted externally and mapped
to LF first thing by the parser.

-- 
John Cowan                                   cowan@ccil.org
One art/there is/no less/no more/All things/to do/with sparks/galore
	--Douglas Hofstadter