W3C home > Mailing lists > Public > www-validator@w3.org > December 2001

Re: "valid [X]HTML x.x!" icons are Evil

From: James Ralston <qralston+ml.www-validator@andrew.cmu.edu>
Date: Thu, 20 Dec 2001 13:25:23 -0500 (EST)
To: <webmaster@domovina.net>
cc: <www-validator@w3.org>, Martin Duerst <duerst@w3.org>
Message-ID: <Pine.LNX.4.33.0112201306190.12982-100000@pcmy.sei.cmu.edu>
On Mon, 17 Dec 2001, Frank Tiggelaar wrote:

> I think that is unfair: IMHO a standard is a definition/set of specs
> you measure your product by, and when both the standard and the
> measure stem from the same house, one should be safe.

> I feel W3C must not withdraw the 'OK' from pages which its software
> did validate, whether the author(s) heeded the encouragement "To
> show your readers that you have taken the care to create etc." or
> not.

Yes, it's not fair: you thought your pages were valid, but correcting
a bug in the spec caused them to be invalid.  The same would be true
if the validator software contained a bug that caused it to
erroneously label invalid pages as valid, and then the bug were
corrected.  Unfortunately, life's not fair.

> In a rapidly changing web-world we are trying to make sure these
> pages can still be read in ten years time, and we thought W3C
> validation was our best bet.

It *is* your best bet, exactly for the reasons you listed.  The
problem is that you didn't know that the validator's evangelism was
overconfident, and should be ignored.

> There are a number of options to correct glitches in the validator
> software (or faulty implementation of standards in same): adding a
> letter relating to the validator software version (4.0a, 4.0b) seems
> a fairly simple one.

That's completely bogus.  Either a web page follows the spec it's
using, or it doesn't.  If you want to add new versions, it should be
of the relevant DTDs, not the validator software itself.

And in the case of correcting a bug with the validator, adding a new
version of the validator is pointless: the old software is wrong, and
must die.

> But once validated [by a specific version of the validator] pages
> must remain 'valid' for the rest of their lives [by that particular
> standard/validator combination] or else W3C should altogether cease
> to offer validation software.

What purpose does (in the case of correcting a bug with the validator
software) returning "valid" for invalid pages serve?  The entire
*point* of the validator service is to tell you if your pages are
valid or not.  If the validator [now] knows that a particular page is
invalid, it *must* declare it as so, *even if* it previously
(erroneously) declared the page to be valid.

> I discussed the W3C issue with a senior consultant in our office (a
> UK Certified ISO-9000 Auditor) and he said that what W3C does is
> 'unbelievable'.  Of course ISO-9000 sets standards for manufacturing
> companies, but also under ISO-9004 (for the service industry), it is
> simply impossible to withdraw validation on the grounds of a
> certifying authority's faults.  After looking at the validator
> pages, my colleague is of the opinion that the validator software is
> presented as a 'certification authority' by itself - there are no
> disclaimers or conditions whatsoever on the page which returns the
> validation results - OK is simply OK.  Or is it?

I don't wish to put words in their mouths, but I don't believe the W3C
ever intended the validator to be a "certifying authority".  It was
designed simply to be a tool (an imperfect one) to let web authors
know if their pages are valid (to the best of the validator's ability
to accurately determine that).

-- 
James Ralston, Information Technology
Software Engineering Institute
Carnegie Mellon University, Pittsburgh, PA, USA
Received on Thursday, 20 December 2001 13:25:45 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 25 April 2012 12:14:00 GMT