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

Re: [VE][108] Error Message Feedback: suggest style="list-type: upper-alpha" in place of type="A"

From: Dan Connolly <connolly@w3.org>
Date: Wed, 26 Oct 2005 00:41:59 -0500
To: "Jukka K. Korpela" <jkorpela@cs.tut.fi>
Cc: www-validator@w3.org
Message-Id: <1130305319.27261.491.camel@dirk>

On Tue, 2005-10-25 at 23:08 +0300, Jukka K. Korpela wrote:
> On Tue, 25 Oct 2005, Dan Connolly wrote:
> > Error [108]: "there is no attribute X"
> We have discussed this particular error message to some extent.
> It has often confused novices, since it seems to complain about the
> absence of an attribute when there definitely _is_ attribute "X".
> I think there have been proposals to improve the wording without
> changing the logic of the validator. Would some new suggestions and
> arguments be needed to have this fixed? (Admittedly, the explanation
> makes the situation clearer than the error message. But an explanation
> should explain, not correct, and the explanations are in tiny text,
> legalese-size.)
> (Out of my hat, even "attribute 'X' not allowed here" would be better.)
> > in particular, "there is no attribute type" on ol.
> On the practical side, the problem is that the document declares
> a Strict DTD but uses Transitional features. In fact, this is the
> first thing that the error message explanation suggests.

Yes, but that doesn't help the user fix it.

> > But it could usefully be much more specific, suggesting I change
> >  <ol type="A">
> > to
> >  <ol style="list-type: upper-alpha">
> >
> > and cite the relevant spec or some tutorial(s).
> (As an aside, embedded style sheets, in style="..." attributes, should
> not be _recommended_. We all use them in testing and to explain things
> concisely, but they aren't recommendable.)
> I'm afraid much of the practical problems with the validator are already 
> caused by its trying to be too helpful and making guesses.
> It's a validator, not a quality checker, still less a quality improver.

It's one of the main ways that a reasonably large and motivated audience
is exposed to W3C specs. I think we're not making the best use
of the opportunity we have here. That's why I worked on the "tip of
the day" feature, and why I took time to suggest this improvement.

I can see that turning all the diagnostics into good how-to hints
is a daunting task.

Perhaps it's more feasible to expand the tip-of-the-day series,
slowly, to cover issues like this one, and to make the selection
of tips sensitive to the content of the document being checked,
if not the errors.

> We could surely use a good overall WWW page checker, but such a thing
> cannot be created just by adding features to a validator.

Oh? It's not at all clear to me why it cannot.

> A validator is supposed to check the validity of markup. It should be 
> quite sufficient to list down the errors. An extra hint about a different
> DTD would probably help more than confuse. But I'd stop there.

That's a coherent position to take, but I don't think it's maximally

>  It's no 
> concern of a validator whether someone uses Transitional or Strict DTD.

It is to this validator user.

> Besides, CSS is a moving target, and it would be risky to bind a markup 
> validator intimately with a style sheet language. In at least one CSS 2.1 
> draft, upper-alpha had been dropped away from the list of allowed
> values for list-style-type.

That risk seems manageable.

> If you started suggesting something for <ol type="A">, what would you do 
> with <ol start="0">? There is no CSS counterpart to it.

Good question. I'd have to study it more. I'm not very practiced
with CSS, actually.

> > I searched google for "css list style" and it nominated a W3C schools page
> > http://www.w3schools.com/css/pr_list-style-type.asp
> The w3schools site has often been characterized as unreliable, to put it 
> mildly.

Well, it's what google nominates, so it's getting a lot of attention.
If it's not right, we do well to get it fixed or provide something
that's going to have more google-juice than w3cschools.

>  It is often confused with the W3C, and maybe that was the 
> intention when the domain name was chosen.

Indeed, I found it annoying that there was no link (that I could
find) from their materials to the official spec.

>  In any case, I would strongly 
> advise against referring to w3schools in any official or semi-official 
> way.

I can go along with that.

Are there any relevant training materials that you _would_ recommend?

Dan Connolly, W3C http://www.w3.org/People/Connolly/
D3C2 887B 0F92 6005 C541  0875 0F91 96DE 6E52 C29E
Received on Wednesday, 26 October 2005 05:42:16 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 March 2016 14:17:47 UTC