Re: validator - how true is it?

On Tue, Feb 08, 2005 at 03:42:12PM +0200, ceo@alierra.com wrote:

>    I have asked our html-coder to make http://www.alierra.com answer the
>    requirements of Validator. The result - our page has 0 mistakes now. Well,
>    this is all fine - but I still see no difference from the users' point of
>    view you. 

You've tested with every single user agent, in every single
environment that your users have? That's impressive.

>    Making the code 100% compatible with Validator does not change
>    the look of the page in browsers. Therefore, I would recommend
>    that the Validator report has two main sections. One section will
>    highlight critical mistakes for the page - unless the webmaster
>    gets rid of them, the page won't look correct in browsers.  The
>    second section will contain recommendations which might be
>    considered. However, these recommendations won't change the look
>    of the page in browsers.

How do you propose to identify the error recovery capabilities of the
many user agents out there? How about the ones which have not been
written yet?

The purpose of having a standard is to have a common frame of
reference for authors of webpages and authors of user
agents. Validation is a means of ensuring that your code follows that
standard (to the extent that can be expressed by a DTD in the case of
markup). If some, or even many, user agents are capable of dealing
with errors then that is lucky for the author (who fails to undertake
basic quality assurance) who makes a mistake when writing their
page. Its not a good reason to deviate from the standard since other
user agents may not perform the same degree of error recovery.


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

Received on Tuesday, 8 February 2005 14:01:16 UTC