Re: % height on <html> (was Re: Table height/width properties)

From: Bjoern Hoehrmann <derhoermi@gmx.net>
Subject: Re: % height on <html> (was Re: Table height/width properties)
Date: Tue, Jul 3, 2001, 4:44 PM

> * Tantek Celik wrote:
>>> No, the document root is always the parent of the document ("root")
>>> element, don't mix them up.
>>
>>The document tree consists of elements, therefore the root of the document
>>tree is the root element.
>
> No, the document tree consists of nodes and the top-most node is the
> root node. Consider e.g. the following document
>
>   <?xml version='1.0'?>
>   <?php print("Hello World"); ?>
>   <elem />
>
> The root node has two children, the processing instruction node and the
> element node.

Ah, I see what you mean by "document root".

There is no element that corresponds to the "document root" - it is an
abstraction that is the parent of all elements in the document which do not
have an element as a parent.

> The CSS Level 2 specification talks exactly about this
> root node and not the document element, that is defined as beeing a
> children of the root node.

I see your point, in that CSS2 could easily be interpreted in that way.

However, I think the way it is written in CSS2 accidentally coincides with
this abstraction of "document root" - I don't think the authors intended
that interpretation.

> See the XML Infoset, the DOM or the XPath and
> XQuery Data Models on this issue, they all say the same thing.

Right, and all of those post-date CSS2.

> With this model in mind, the CSS Level 2 recommendation makes great
> sense,

Yes, it can be made to make sense with that model in mind - that is true.

> however, I let it to the authors to say, what their intention
> was.

Right - I think that is the real issue, the intention of the authors was
unclear, and though can be made unambiguous/consistent with the XML Infoset
model of document root, I conjecture that it is unlikely that that was the
intent.

We've logged this as an issue that the CSS working group needs to discuss
and explicitly errata one way or the other.


Tantek

Received on Wednesday, 4 July 2001 16:46:15 UTC