W3C home > Mailing lists > Public > public-html@w3.org > April 2008

Re: [html4all] several messages about alt

From: Henri Sivonen <hsivonen@iki.fi>
Date: Sun, 13 Apr 2008 21:57:47 +0300
Cc: HTML4All <list@html4all.org>, Ben Boyle <benjamins.boyle@gmail.com>, W3C WAI-XTECH <wai-xtech@w3.org>, HTML WG <public-html@w3.org>
Message-Id: <AB891F04-795E-4D4D-8754-A94BE60DA756@iki.fi>
To: Leif Halvard Silli <lhs@malform.no>

On Apr 13, 2008, at 18:33, Leif Halvard Silli wrote:
>
> Henri Sivonen 08-04-13 14.05:
>> On Apr 13, 2008, at 14:15, Ben Boyle wrote:
>>  > Where does "conformance" fit in on the scale?
>>
>> It depends on what kind of conformance is meant. First, there's  
>> the  question of conformance to what? To HTML5? To WCAG? Second,  
>> there's  the issue that overall conformance criteria may have parts  
>> that are  not machine checkable.
>>
>
> And as you say below, HTML 5 has such criterias.
>
>> In theory, HTML5 conformance and HTML5 validity are the same thing.  
>> In  practice, though, people tend to think that validity is what a   
>> validator checks, which is machine-checkable conformance criteria.
>>
>
> Your product is called a conformance checker. So, I take it that you  
> are in doubt about whether it should actually be a conformance  
> checker, or if it should be a validity checker. And also it seems  
> that even if it will be a validity checker, you still consider  
> calling it a conformance checker. (Conformance to HTML 5, that is.)

Conformance checker software is understood (or should be understood)  
to only check for machine-checkable conformance criteria.

Quoting the spec draft:
"Conformance checkers are exempt from detecting errors that require  
interpretation of the author's intent"

The spec defines "HTML5 validator" to mean a conformance checker. The  
spec doesn't define 'validity' or 'valid', but it is assumed that a  
validator checks if a document is valid.

>> Examples of non-machine-checkable conformance criteria:
>>  * "The img must not be used as a layout tool." (HTML 5)
>>  * "Authors must only use elements, attributes, and attribute  
>> values  for their appropriate semantic purposes." (HTML 5)
>>  * "Content MUST NOT use a code point for any purpose other than  
>> that  defined by its coded character set." (Charmod)
>>  * "Images of text are only used for pure decoration or where a   
>> particular presentation of text is essential to the information  
>> being  conveyed." (WCAG 2.0
>
> If we formalise that the first step of validation/conformance  
> checking, namely the checking of whether images have the correct alt  
> text and are used in the right way, if tables have summary, and so  
> on and so forth, as a step that must be done by the author/ 
> webmaster, then your product could be allowed to check only the more  
> formal points -

An automated tool becomes less automated if it starts giving more and  
more messages of the nature "Please check yourself if you are  
violating rule foo here." If you take it to absurdity, the tool should  
ask the user to verify the semantic correctness of the use of each  
element and attribute.

> Henri Sivonen 08-04-13 10.55:
>> The syntax rules need to be lax enough for all kinds of sites to  
>> be  able to comply. If all sites can't be accessible, too, then   
>> accessibility and syntactic correctness are different evaluation  
>> axes.
>>
>
> Those are different evaluation axes. But experience has shown us  
> that the only validation that authors care about is the general CSS  
> and HTML stamps. Therefore, we must (continue to) incorporate social  
> consciousness into the general stamping tools.

The validator I develop is not a stamping tool. It is a tool that  
helps authors detect mistakes that they didn't intend to make, so that  
they don't need to spend time wondering about the effects about their  
unintentional doings. For example, the validator I develop helps  
author detect that the alt attribute was typoed as 'atl', which is  
useful, because atl wouldn't work.

> And also - and this goes against what you said in your rathole  
> letter: people tend to think that the HTML stamp actually also  
> incorporate some general accesibility checking, as it does check if  
> there is an alt and a summary etc.

I have no interest in spreading claims that you get accessibility by  
syntactic correctness.

> The W3 HTML checker has always done a small bit of accessiblity  
> checking , and that is part of why people want to check their pages  
> in that validator. To offer a checker as a same kind of prestiged  
> checker as the current W3 tool, without incorporating some basic  
> accessibility checking, would be a bit like stealing goodwill from a  
> wholly different kind of tool.

I'm pretty sure I haven't advertised Validator.nu in a way that stole  
goodwill deceptively.

Please let me know if you find bogus claims in Validator.nu  
documentation, UI or advertising. Unfortunately, I can't fully stop  
people from transferring bogus impressions created by others onto  
their preconceptions about Validator.nu.

Aside: I'm not a big fan of "prestige", "gravitas" and words of that  
nature that keep popping up in accessibility debates where appeals to  
authority are made.

> It was Anne, who in his blog once said those very wise words (quoted  
> from my mind) that "people are making XHTML pages thinking they are  
> making more accessible and semantic pages - though in reality both  
> <FONT> etc are permitted in current XHTML". He is right, but this  
> also shows that people expect to be measured against a accessibility  
> and semantisism standard when they run their pages through the  
> validator mill.

A validator cannot check that a page is semantically correct. It can't  
properly check for accessibility, either.

We should dispel misconceptions about what validators do instead of  
catering to the misconceptions.

> Btw, it strikes me that validation tools have two kinds of purposes:  
> One is acting as a formal stamp. The other is that they are  
> developement tools.

I'm not interested in developing a formal stamp. I am interested in  
developing a development tool.

> For the second, the "unready" stamp does not conflate those two  
> things. It helps keeping them apart. And allows an HTML 5  
> conformance checker to give a definitive 'Yes', when the author has  
> done his part. And it raises the consciousness about the fact that  
> writing HTML documents is a process.

Software can't give a definitive yes.

I have made a specific effort in the UI and documentation of  
Validator.nu and any statements that might be considered advertising  
never to suggest that passing validation meant conforming to *all*  
conformance criteria there is.

> Only the author can know if the document is ready - only he is in  
> charge of how the text needs to be "alt-ed".

Right.

> WCAG is a higher degree of accessibility checking. Given all the  
> things you say that a normal page maker never will do, how come you  
> think he/she will do WCAG cheking?

I think the population of authors who will check their pages for WCAG  
conformance will be smaller than the population of authors that will  
check their pages for machine-checkable HTML5 syntactic conformance  
criteria. The reason why I think so is that checking for WCAG  
conformance requires human labor while checking for machine-checkable  
syntactic conformance criteria can be, by definition, automated.

> Will you offer a direct link in your tool for running the page in a  
> WCAG checker, for instance? (That could be nice, regardless.)

You really need a human to check for WCAG 2.0 conformance. I do intend  
to offer links to JSlint and the CSS validator, though.

-- 
Henri Sivonen
hsivonen@iki.fi
http://hsivonen.iki.fi/
Received on Sunday, 13 April 2008 18:58:31 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 9 May 2012 00:16:14 GMT