HTML4 Implementation Requirements

The HTML 4.0 specification states that:

#    A conforming user agent for HTML 4.0 is one that observes the
#    mandatory conditions ("must") set forth in this specification.

And defines must like this:

#    The key words "MUST", "MUST NOT", "REQUIRED", [...] in this
#    document are to be interpreted as described in [RFC2119].

But does not state at *any* point (I have done a very careful full
text search) that supporting *any* element is mandatory, with the
exception of TITLE, Q, and possibly SCRIPT and STYLE (depending on the
interpretation of certain statements).

The specification also defines the word "obsolete" like this:

#    Obsolete
#     An obsolete element or attribute is one for which there is no
#     guarantee of support by a user agent. 

Since guarantee of support in a [conforming] user agent is limited to
all those elements that are mandatory ("must"), that would make all
elements other than TITLE, Q, SCRIPT and STYLE obsolete. However,
since the specification goes on to say that:

#     Obsolete elements are no longer defined in the specification,
#     but are listed for historical purposes in the changes section of
#     the reference manual.

...I would assume that logically, all elements but those listed in the
changes section (i.e., all but LISTING, PLAINTEXT, and XMP) are NOT
obsolete, and are therefore guaranteed to be supported by a
[conforming] user agent.

 * * *  IS THIS CORRECT?  * * * 

Furthermore, does this extend to all attributes?

To give a concrete example: Is support for the "longdesc" attribute
mandatory? (The spec does not specifically say it is optional.)


NOTE. I am being *incredibly* pedantic here. My personal view is that
UAs should support *everything* given by the HTML4 specification.
However, in my efforts to make an HTML 4.0 Test Suite, it has been
said that I should limit myself to testing the mandatory parts of the
specification only.

That is why I am trying to obtain clarification regarding what is, and
what is not, mandatory.

-- 
Ian Hickson

Received on Friday, 5 February 1999 13:22:24 UTC