[Bug 68] provide a warning when an XHTML page lacks an xmlns declaration.

http://www.w3.org/Bugs/Public/show_bug.cgi?id=68


iamback4now@yahoo.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |iamback4now@yahoo.com




------- Comment #10 from iamback4now@yahoo.com  2007-08-19 16:15 -------
(In reply to comment #9)
> Done in HEAD (soon to be 0.8.0), see:
> http://qa-dev.w3.org/wmvs/HEAD/check?uri=http://qa-dev.w3.org/wmvs/HEAD/dev/tests/xhtml1-strict-missing-xmlns.html;ss
> 
> The problem is that DTDs can either set an attribute to FIXED or REQUIRED but
> not both...

How is this a problem?

There seems to be a lot of confusion about - or different interpretations of -
the meaning of FIXED.

I understand it like this (and I'm not alone in this):
FIXED means 1) the attribute can have only one value, and 2) the document
already *has* that value by means of the FIXED value in the DTD it refers to.
This implies that if a value is defined as FIXED and a document refers to a DTD
that defines it like that, declaring the attribute isn't mandatory: that's
(almost) the /opposite/ of REQUIRED which mandates that the attribute be
declared. At the same time, it's /allowed/ to be declared, as long as it has
the value as shown in the FIXED clause, and not anything different. (Which is
the reason FIXED and REQUIRED cannot appear at the same time. :))

That implies:
1. there is /no need/ for a warning at all when the DTD for the declared XHTML
version states this as being FIXED (rather than REQUIRED) and the document does
not declare the attribute;
2. there certainly /should not/ be an error message, as the validator gives
now, when the attribute is not declared: when not declared, the document
already /has/ the correct namespace; and
3. there /must/ be an error message when the attribute /is/ declared but has a
value /different/ than the one indicated as FIXED in the DTD

IMO the resolution of this issue has actually created a *bug* in the validator,
by issuing an error message for what is not an error at all.

Since it is apparently not possible (for me, at least) to reopen this issue (or
any issue?), I'll file a new bug.

(Other related issues: 800, 3385, 4475, 4495)

Received on Sunday, 19 August 2007 16:15:57 UTC