W3C home > Mailing lists > Public > www-validator@w3.org > July 2012

Re: Suggestions for your Validator from a HTML Idiot

From: Jukka K. Korpela <jkorpela@cs.tut.fi>
Date: Mon, 30 Jul 2012 08:20:01 +0300
Message-ID: <50161981.9010406@cs.tut.fi>
To: Albert Dodd <rockiedodd871@gmail.com>
CC: www-validator@w3.org
2012-07-26 15:50, Albert Dodd wrote:

> But when it comes to
> your validator I am just completely clueless of where to start and what
> to fix.

That's understandable, because the validator is, or at least was, a tool 
for checking syntactic conformance according to formally defined rules. 
Although many of the validator's messages are intuitively understandable 
(say, 'element "TABEL" undefined', when you meant <table> but mistyped 
it as <tabel>), they generally require understanding of both the 
formalized description language (certain features of SGML or XML) _and_ 
the specific rules in HTML specifications, in DTDs. Part of this is 
explained at

But there's a more practical viewpoint. You can just decide that the 
messages are mostly gibberish, but they indicate that there is 
*something* wrong. So take them just as incentives to look at your 
favorite HTML textbook and reference and find out there what's wrong.

> Then I went to the Tidy version and copied it.

HTML-Tidy is mostly just confusing, and often plain wrong (e.g., in 
dropping out markup).

> First of all your programmers need to get it setup where the main output
> page has 3 dividers.

The validator is based on old code. It's based on a SGML validator, with 
some enhancements and modifications, but with little development work 
for many years. It is unlikely that this will change. It could be tuned 
up a little, but any major improvements would require a rewrite.

This does not really apply to the "HTML5 functionality" of the 
validator, which is very different from old validation. It is based on 
(or just replicates) code at http://validator.nu and constitutes an ad 
hoc analyzer specifically designed for HTML5. It has its share of 
problems (rather big I would say), but it's definitely possible that it 
will be actively maintained and developed. As of now, it is 
(appropriately) labeled as experimental, and it isn't even clear exactly 
what it checks against (which HTML5 draft).

Received on Monday, 30 July 2012 05:20:33 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 March 2016 14:18:06 UTC