W3C home > Mailing lists > Public > www-validator@w3.org > August 2005

Re: Suggestion for improved wording when reporting unrecognised attribute

From: Olivier Thereaux <ot@w3.org>
Date: Mon, 8 Aug 2005 17:55:48 +0900
To: Philip TAYLOR <P.Taylor@Rhul.Ac.Uk>
Cc: www-validator@w3.org
Message-ID: <20050808085548.GB16627@w3.mag.keio.ac.jp>

Hi Philip,

On Mon, Aug 08, 2005, Philip TAYLOR wrote:
> 
> Thanks, Olivier.  I took the opportunity of the announcement
> to look at the earlier query relating to the diagnostic
> 'there is no attribute "height"', as reported below, and
> although I have little sympathy with the original query
> I am forced to agree that the diagnostic in its short
> form /is/ a little misleading. 

Obviously, yes, I agree. As you probably know, the error messages
themselves come from the underlying parser (opensp), and this one (as
well as others), tend to be rather inane. We've been trying our best to
accompany these with useful explanations as a workaround, e.g.
http://validator.w3.org/docs/errors.html#ve-108
but the workaround reaches its limit when people want the validator to
be wrong (and their docs to be right) so much that they don't read past
the bad message...

[getting a bit technical below]

This (fixing the error messages themselves) is something I would really
like to get around to, but it is not trivial. It would involved some
text mapping/substitution, e.g for this message we know that openSP will
give us something like:
"there is no attribute %1", where %1 is... something. (other cases have
several values, such as "value of attribute %1 invalid: %2 cannot start a
name". 

Simply recognizing the "%1" value and replacing the text with what we
think is right would kind of "work", but would not be robust (what if
opensp's messages change?). 

Therefore, the solution really is to get the actual messages, in openSP,
fixed, or at least improved. Note that I'm not just saying it's "someone
else's problem", as some of the developers of one tool also contribute
to the others... 

Note that the questions raised by this problem are fairly similar to the
questions we'll have to solve to make the validator actually
localizable - something this new version gets closer to, by the way -... 

Thanks, as always, for your feedback and suggestions, Philip.
-- 
olivier
Received on Monday, 8 August 2005 08:55:49 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 25 April 2012 12:14:19 GMT