Re: [SVGMobile12] Missing namespace puts document in error

On Wednesday, October 26, 2005, 9:52:42 PM, thomas wrote:

tdkc> Hi all,

tdkc> Just so we are clear on this, it makes it impossible to have a
tdkc> conforming SVG UA that is also a validating XML processor. As
tdkc> discussed elsewhere if a DTD provides, for example a '#fixed'
tdkc> value for xmlns on 'svg' elements a conforming XML parser must
tdkc> expose it to the infoset, if it chooses to read the DTD (which a
tdkc> validation requires), the wording below requires it not to make
tdkc> use of such decl (it is not "in the document").

Good catch, Thomas. The text suggested by the WG clearly was a
conformance requirement on conforming *content*, so did not make it
impossible; but Ian's wording shifts this to the user agent, or to a
combination of user agent and content; as you show, this is
undesirable, possibly even bogus.

Ian, on further discussion in the WG, we were unable to accept your
suggested text because it makes DTD validating processors non

Instead, we plan to say

  Conforming SVG content must place SVG elements in the SVG namespace by
  XML Namespace declarations in the document.

So, content which is in some other namespace and happens to use elements
called 'svg', 'path', etc is clearly not conforming SVG content, which is
the expected result. All SVG elements (the term "SVG element" is now
defined) are covered, not just the 'svg' element. And 'svg' elements
which are not the document root are covered.

tdkc> If this is the case I think that there needs to be wording to this
tdkc> effect to make is clear to everyone the real implications of this
tdkc> statement.

It was not a desired result, but was a side-effect of Ian's wording.

Looking now at processors, for conforming content, a validating parser
willbe fine since the value in the instance will match the #FIXED
declarations in the DTD. For non-conforming content,as it happens the
correct namespace will be assigned on validation. Theis is conformant
processing, but is not something we want authors to rely upon.

Thomas, I hope this satisfies your comment?

Ian, this covers your three points from your mail of 26 Oct, so
hopefully satisfies your comment as well.

tdkc> wrote on 10/26/2005 01:31:21 PM:

>> On Wed, 26 Oct 2005, Chris Lilley wrote:

>> > IH> User agents must only consider elements explicitly placed in the
>> > IH>    SVG namespace by XML Namespace declarations in the document  (e.g.,
>> > IH>    <svg xmlns="">) as being SVG  elements.

>> > I think your suggested wording is good. I assume you would be  satisfied
>> > if we used it?

>> Yup.

 Chris Lilley          
 Chair, W3C SVG Working Group
 W3C Graphics Activity Lead
 Co-Chair, W3C Hypertext CG

Received on Thursday, 3 November 2005 15:47:51 UTC