W3C home > Mailing lists > Public > public-wai-ert@w3.org > March 2005

Re: Locating In EARL Example

From: Charles McCathieNevile <charles@sidar.org>
Date: Fri, 01 Apr 2005 00:39:57 +1000
To: "Chris Ridpath" <chris.ridpath@utoronto.ca>, public-wai-ert@w3.org
Message-ID: <op.soia0vkjw5l938@researchsft>

On Thu, 31 Mar 2005 00:18:33 +1000, Chris Ridpath  
<chris.ridpath@utoronto.ca> wrote:

> Here is a link to an HTML file that shows how we may identify an  
> accessibility error using EARL:
> http://checker.atrc.utoronto.ca/test1-earl.html
> Note that the EARL is not yet correct! We still need to work on getting  
> it right and this EARL will not validate. However this is a concrete  
> example and I hope that we can improve the EARL to make it right.

The attributes that you have added are not valid. They probably should be  
modelled as properties - i.e. they could equally be expressed as

<earl:subject r:resource="#subject">

I think it is probably better to think about how this is modelled. One  
possibility is to think of a subject (a page) and to have a context  
property (this is what annotea does) that *attempts* to provide a more  
accurate pointer. This is going to have a number of possible properties,  
such as an Xpath or fuzzy pointer, perhaps a contextualising code  
fragment, a line number in a given serialisation (this is realy weak. Too  
many tools change th line numbering, so unless you exect to add contextual  
stuff and use the line number as a hint for where to start searching it is  
likely to be extermely fragile indeed).


<earl:subject r:resource="#subject"/>
   <earl:near r:parseType="Literal">
   blah blah look at <img src="rex.png"/> if you like that kind of thing
   <earl:pointer ... >

You should also be using parseType="Literal" in your message.

> The example uses the accessibility test that is part of the WAI test  
> suite:
> http://www.w3.org/WAI/GL/WCAG20/tests/test1.html
> I can generate these EARL example files for all the other tests in the  
> suite but would like to get comments first.
> Is this format usable by most people? What other information would be  
> helpful?

Well, it depends on the error type. I think Shadi's idea of trying to  
figure out what kind of errors there are, and describe the information  
that will help locate each of them is a good one. For example id  
attributes are handy in an xpath as they are much more robust than most  
other kind of xpath. But where the error is an XML validation problem that  
two things have the same id (common enough in HTML-like tagsoup found in  
the wild) it isn't really the answer...

I am not sure if we can make an exhaustive list of error types and the  
most useful ways of describing them. Perhaps we can list the ways of  
describing the location(s) of an error and assume that people can describe  
their new type of error as being appropriate for one or other type of  
locator information?



Charles McCathieNevile                      Fundacion Sidar
charles@sidar.org   +61 409 134 136    http://www.sidar.org
Received on Thursday, 31 March 2005 14:40:10 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:55:52 UTC