Re: validator - how true is it?

(NB: Since this thread is about the Markup Validation Service and not
the CSS Validation Service, I've CCed this to the www-validator@w3.org
list and set the Reply-To header to the same.)

On Fri, Jan 28, 2005 at 03:40:39PM +0200, ceo@alierra.com wrote:

>     1. On the very front page of your respected http://validator.w3.org you
>        should mention that the program is not bug-free. Otherwise, the
>        absence of this message makes people believe that validator is a
>        reliable program.

No software is bug free, but the Markup Validation service is a
reliable program. It rarely gets things wrong - certainly not (IMO)
sufficiently often enough to display a "Don't trust me!" message.

>     2. Have http://www.msn.com, http://www.google.com, or http://www.ebay.com
>        typed into the address area. How will you comment their error report?
>        I will doubt that MSN, Google, or Ebay corporations hire the worst
>        html-coders.

Browsers have hefty error correction routines in their markup
parsers. This leads to websites authors writing sloppy code. Large
companies are not commonly exceptions.

>     3. I had my own site http://www.alierra.com "validatored". I particularly
>        liked the following mistake:
> 
>    Line 11, column 6: end tag for element "HEAD" which is not open
> 
>    </HEAD>
> 
>    However, line 2 has the following tag <HEAD>.   
> 
>    How will you comment this?

Since you have no Doctype (hint: Fix the first error first, it can
have consequences on later errors), the validator assumes you are
using HTML 4.01 Transitional.

In HTML 4.01 Transitional the end tag for <head> is optional, as is
the start tag for <body>.

In your <head> section you have <meta http-equiv="Content-Type"
content="text/html; charset=iso-8859-1" />, using XHTML style
self-closing tags in HTML is a mistake. Under HTML rules <foo /> means
the same as <foo>> or <foo>&gt;.

Since character data (such as a greater than sign) is now allowed in
the <head> section, but it is allowed in the <body> section this
implies that you close the <head> section and open the <body> section
immediately before the > sign.

So your code reads the same as:

<head>
<title>Alierra Custom Website Design Company - Professional Design and Consulting Services</title>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1" >
</head>
<body>
&gt;
<link href="img/styles.css" type="text/css" rel=stylesheet>

So this is your mistake, not the validators. If you had enabled the
"Show Parse Tree" option of the Validator, it would have shown you
this.

>     4. I believe every site, which is a more or less complicated, will have
>        at least 30 mistakes within the Validator.

Well then, lets pick some sites. A few W3C sites, a few by people I
know, and a bug bunch picked out of the list of Blogs I fetch the RSS
feeds from:

http://validator.w3.org/check?uri=http%3A%2F%2Fdorward.me.uk%2F
http://validator.w3.org/check?uri=http%3A%2F%2Fw3.org%2F
http://validator.w3.org/check?uri=http%3A%2F%2Fgreytower.net%2F
http://validator.w3.org/check?uri=http%3A%2F%2Foffog.org%2F
http://validator.w3.org/check?uri=http%3A%2F%2Fvalidator.w3.org%2F
http://validator.w3.org/check?uri=http%3A%2F%2Fwebstandards.org%2Fbuzz%2F
http://validator.w3.org/check?uri=http%3A%2F%2Fwww.theregister.co.uk%2F&charset=%28detect+automatically%29&doctype=%28detect+automatically%29
http://validator.w3.org/check?uri=http%3A%2F%2Fsimon.incutio.com%2F
http://validator.w3.org/check?uri=http%3A%2F%2Fphotomatt.net%2F
http://validator.w3.org/check?uri=http%3A%2F%2Fphotostack.org%2F
http://validator.w3.org/check?uri=http%3A%2F%2Fweblog.delacour.net%2F
http://validator.w3.org/check?uri=http%3A%2F%2Fln.hixie.ch%2F
http://validator.w3.org/check?uri=http%3A%2F%2Fwww.benhammersley.com%2F
http://validator.w3.org/check?uri=http%3A%2F%2Fwww.benmeadowcroft.com%2F
and so on.

>    I was just merely saying that inexperienced users firmly believe in
>    Validator and they require their sites to be in conformity with its rules.
>    Whereas, the users do not understand that the rules are not perfect.

It is true that the rules are not perfect, however there is very
rarely a good reason to break them[1], and the Markup Validation
Service rarely makes a mistake when checking if a document follows
them.

[1] I'm being liberal when I say that; I've never found a good reason
to break them.

-- 
David Dorward                                      http://dorward.me.uk

Received on Friday, 28 January 2005 14:05:22 UTC