Re: XML5

[Henri Sivonen <hsivonen@iki.fi>]

On Dec 16, 2010, at 09:29, Liam R E Quin wrote:

> The XML specification forbids a parser from saying something was XML
> if it wasn't. But it doesn't require a "yellow screen of death" and
> never has.

Indeed, the WebKit implementation where the document is cut short but the part before the error is not removed from the DOM (or from the view) is also conforming.

> Error correction is perfectly OK as long as the API
> provides a way for the application to determine that the input was
> not well-formed XML.

I think it's clear that that's not what the spec intends to say, although the letter of the spec can be met if you creatively interpret "MUST NOT continue to pass character data and information about the document's logical structure to the application in the normal way" to be met if you switch from the normal SAX ContentHandler interface over to an identical set of callbacks under, say, an interface called BrokenContentHandler.

But since James Clark is on this list and is on the record (http://lists.w3.org/Archives/Public/w3c-sgml-wg/1997May/0079.html) voting against Draconian error handling but still implementing it, I guess he'd be better positioned than me to recount what the correctness of the "perfectly OK" characterization is relative to what the XML spec tries to say.

> So, you don't need to call it a new version of XML, and you don't need
> a new version of XML.

I'd be fine with describing XML5 as a new parsing spec for XML 1.0 and not a new version of the language itself.

-- 
Henri Sivonen
hsivonen@iki.fi
http://hsivonen.iki.fi/