Re: Balancing the myth-busting.

Hi Tina,

On 09/08/05, Tina Holmboe <> wrote:
>   the W3C validator has one specific job, and one job
>   only: it checks the syntax of a markup language against a DTD.
>   That is *all* it does - and it is wrong to say that it doesn't catch
>   an error, when the markup doesn't contain the erroneous attribute.
>   Testing whether markup is correct after it has been manipulated by
>   something-or-other on the client side is way out of scope for the
>   validator.

I disagree. For the markup validator to remain useful, I personally
think it should do more than just confirm whether or not the markup
conforms to a DTD. The validator creates a document tree in order to
parse the nodes and ensure they validate against the appropriate DTD.
If they already have a document tree, it would be relatively easy for
them to apply any changes to the DOM through scripting when a document
is loaded to their document tree.

Jeffrey Zeldman recently praised Bobby van der Sluis [1] on a new way
to embed Flash while supporting web standards [2]. I don't wish to
undermine Bobby's work, as it does more than purely embed a Flash
movie, but it doesn't support standards; it just appears to support
standards using the W3C's markup validator, but the technique inserts
an embed element into the DOM if required. If required is better than
always, and getting Flash to work cross-platform is an important
issue. I'd prefer a standards compliant method, but despite what it
says on the tin, this isn't a standards compliant method. The only
reason anyone may think it is, is because the W3C's markup validator
only checks the markup.

If validation is important, then surely it's important that the DOM
remains valid, particularly as developers are encouraged to separate
structure, presentation, and behaviour. I'm not taking an
authoritative stance on this issue; it's just that I believe the
markup validator needs updating to remain useful. Taking its literal
meaning, the markup validator clearly does a good enough job in its
current state. Reflecting changes made to the DOM when a document is
loaded isn't unachievable, and would make a useful too that much more

>   But the W3C syntax checker - the technical term for which is
>   "validator" - isn't an accessibility tool. It checks grammar of
>   supplied markup, and that's *all* it does.
>   I suggest you remove that test from your article, as it really is not
>   related to the other tools.

I do appreciate that I'm in the minority in believing that validity is
important for accessibility. For accessibility not to be considered a
bolt on, it needs to be considered at all stages of the development
lifecycle. In my experience, developers tend to take an iterative
approach when building websites, and validate the markup at regular
intervals to ensure that errors aren't being compounded through the
development lifecycle. As I believe that validity is important for
accessibility, I believe the results from the markup validator belong
in the results. Obviously, anyone who doesn't believe that validity is
important for accessibility can just discard the markup validator


Best regards,


Supplement your vitamins

Received on Tuesday, 9 August 2005 18:11:21 UTC