Re: The Markup Validator belongs to this community... (was Re: validation in Opera)

On Mon, 17 Mar 2003, Olivier Thereaux wrote:

> Yet it has to juggle between the need to be:
> - "perfectly accurate", because many think of it as "the normative 
> reference" (and not the spec itself...).
> - user-friendly, because it's the most famous validator for markup 
> (even though there are other similar services, some as good, even 
> better, that W3C's).

Indeed.

> As for this particular issue, my *personal opinion* is somewhat in 
> between : I would like to make the validator be more user-friendly, yet 
> I appreciate the validator's strictness, I am not a big fan of 
> "fallback" solutions... And I honestly think that the doctype error 
> message is one of the best we have... I would rather see this community 
> submit better documentation and improvements to the UI than adding 
> fallback behaviour.

The issue is about it being treated as a *fatal* error.  Changing that
is perfectly compatible with keeping the error message.  We don't
(nor should we) treat parse errors as fatal, despite the risk of
a cascade of 'spurious' errors from a single typo (like spurious
XHTML syntax in the HEAD section of an HTML doc, for instance).

> Here are a few ideas, feel free to take them home (and back):
> - a "what does this mean" page linked from "this page is valid foo" 
> results
> - a "what does this mean" page linked from "this page is not valid foo" 
> results
> - a better, newbie-friendly explanation for the "no charset specified" 
> page
> - a fluffed-up FAQ

We've contemplated this before, and come up against a shortage of
round tuits.

> - alternate, newbie oriented explanations for the error messages
> (rewriting the error messages themselves is something that is being 
> worked on in the code, but we have to free ourselves from the tight 
> dependency to openSP before we can do this properly).

I've recently had another hack at the OpenSP error messages, and I like
to think it's the best yet:-)  They're running at Page Valet, and can
easily be "dropped in" to Validator.

-- 
Nick Kew

Received on Thursday, 20 March 2003 17:43:31 UTC