W3C home > Mailing lists > Public > www-validator@w3.org > January 2004

Re: UKUUG Website query

From: Ian Hickson <ian@hixie.ch>
Date: Wed, 21 Jan 2004 16:22:04 +0000 (UTC)
To: "Jukka K. Korpela" <jkorpela@cs.tut.fi>
Cc: www-validator@w3.org
Message-ID: <Pine.LNX.4.58.0401211602120.26544@dhalsim.dreamhost.com>

On Sun, 4 Jan 2004, Jukka K. Korpela wrote:
> On Sun, 4 Jan 2004, Ian Hickson wrote:
>> Why is conformance-to-a-subset-of-the-conformance-requirements-that-
>> happens-to-be-describable-using-a-particular-schema-language a concept
>> that needs its own term?
>
> "Schema language"? This is about DTDs, not schemas.

DTDs are a kind of schema.


> The concept had a term before HTML was invented.

It was probably a problematic term then too.


> It is a relevant concept because such conformance can be objectively and
> automatically verified by a validating parser, such as a separate
> validator.

Claiming that a particular type of validation is relevant because it lets
you do that particular type of validation is somewhat redundant! I assume
you meant "It is a relevant concept because such conformance can be
objectively and automatically verified by a ... parser, such as a separate
[program]." -- i.e. that it is useful because it allows the conformance of
the document to be checked algorithmically.

The problem is that the term "validation" is being used to mean checking
against _an arbitrary subset_ of the requirements that can be checked by
algorithmic procedures.


> This is all that validators are about, so if the concept is
> irrelevant, so are validators, and this list.

Programmatic conformance checking is an important part of the conformance
checking process. Since few people have access to knowedgable proof
readers who can check the document for them, it is the _only_ part for
many people.

That doesn't mean there is anything special about a _particular_ subset of
the things that can be checked in this way that require their own term.
And it doesn't mean that the concept of checking for conformance to any
subset of the complete conformant requirements should be given a term as
misleading as "valid".


>> It seems to me that by making such a fuss over conformance to that subset
>> you are diluting the point of conformance to the whole specification.
>
> Au contraire. I often try to explain why validation is much less important
> than many people and even organizations claim. And yes, I _am_ especially
> referring to the statement "To show your readers that you have taken the
> care to create an interoperable Web page, you may display this icon - -".
> See also http://www.cs.tut.fi/~jkorpela/html/validation.html

Unfortunately, most people do not read such documentation. They hear the
word "valid" and assume it means what it means in the rest of the world --
namely, that there are no problems.


>> The funniest thing is that using your definition, the following:
>>
>>    <a><a>test</a></a>
>>
>> ...is invalid in HTML4, but valid in XHTML1,
>
> You don't need my definition for it.

Yes you do. With a different definition, such as "validity means complying
to all the conformance requirements of the relevant specifications", such
a fragment would _not_ be valid.


> It is a simple fact - though we don't see such facts mentioned very
> often, since they reveal a little bit of the true nature of XML as much
> less powerful than SGML.

The same could be shown with SGML:

   <html lang="1"> ... </html>

...is "valid" but non-conformant.


> The word "validation" was perhaps poorly chosen, as is much of SGML
> terminology.

That is indeed what I am arguing.


> But changing that now would just lead to more confusion.

Given the world we live in now, with multiple schema languages differing
about their opinion of whether non-conformant documents are valid or not,
I doubt that much _more_ confusion could really be generated.


> But when validation has been advocated to everyone and his dog, often
> with grossly false claims about its impact, confusion and
> misunderstandings have been caused - people even think they know what
> "validation" means without having ever read its definition!

It seems you agree with me in principle. :-)

-- 
Ian Hickson                                      )\._.,--....,'``.    fL
U+1047E                                         /,   _.. \   _\  ;`._ ,.
http://index.hixie.ch/                         `._.-(,_..'--(,_..'`-.;.'
Received on Wednesday, 21 January 2004 11:25:55 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 25 April 2012 12:14:10 GMT