Re: name/id namespace

A rumor says that Nick Kew wrote:

> Name is overloaded, and is primarily useful in form inputs, where
> duplicates definitely are acceptable (e.g. in a bunch of checkboxes).

Right, I was specifically speaking of name attribute carried by links.

>>	 The value of this attribute must be a
>>       unique anchor name. The scope of this name is the current document.
> 
> I'd read that as meaning it must be unique in order to work properly
> as a link target, not to meet the slightly different criterion of being
> valid HTML.  The DTD - the machine-readable spec for HTML - explicitly
> doesn't require uniqueness.

What means "work properly" ? Where do you read that ? No, when the HTML
WG wrote that sentence, it meant "it is an error". I clearly remember
Chris Lilley being in front of me at the HTML WG meeting saying "Good
catch, this should be an error".

>> Please note the "must". But Validator.w3.org does not detect when two
>> named anchors have same name in the same document.
> 
> If we accept your reading of that section, then we have an inconsistency
> in the spec, which makes the HTML WG the guilty party.

No. There is no way we can specify in a DTD that two attributes share
the same value namespace. That can only come from wording. The HTML4
specification is not only made of a DTD, it contains also constraints
in the text.

>>	 I did not test the
>> case when a name attribute and an ID have the same value.
> 
> That is perfectly valid, provided there are no duplicate IDs.
> Once again, it makes sense in the context of link targets, but not
> in the context of other uses.

I disagree. I was a member of the HTML WG when we discover this issue of 
name attribute and ID sharing the same namespace. That was during an 
HTML WG face-to-face meeting in Sophia-Antipolis in spring 1997. The 
resolution at that time was clearly that a name attribute  carried by an 
anchor and an ID must NOT have same value in a document.

>> IMHO, W3C's validator should test this unicity in the name/id namespace.
> 
> I agree such a warning would have some merit in a diagnostic tool,
> but that tool isn't a validator.

If we only validate against the DTD, I agree. IMHO W3C's validator
should validate _and_ check correctness. What the point verifying
if a document is valid from a DTD point of view if the document is
deeply incorrect from a spec point of view ? Reading the spec,
<a name="foo">a</a><a name="foo">b</a> is forbidden in a document.

</Daniel>

Received on Tuesday, 20 August 2002 04:07:41 UTC