Re: EARL Identifing The Error

Hi Chris (and c/c the list),

    thanks for your posting. It is really interesting. Consider my reply as 
a first-sight comment. I will have a proper look at the code in the next 
few days.

    I like the idea, although I believe we should come with some different 
levels of accuracy. For instance, I would not exclude at all the 
possibility to check against an 'id', if provided. If not, we could go 
deeper with other heuristics, and so on. Do you get what I mean?

    I was thinking about the possible scenarios regarding the location of 
accessibility problems in a persistent way, which I believe we should 
consider in an "pervasive" way, from the very beginning. Of course, I 
guess, we are dealing with HTML documents at the moment,  but should we 
also consider other formats and types of resources?

    Anyway, an HTML document, at a different time, may be:

1 - completely identical
2 - different:
         a - only the structure changes (e.g. users can select a sort of 
template for the document, which changes, for layout purposes, the order of 
the elements), but content stays the same
         b - the structure stays the same, content changes
         c - both structure and content changes, due to content-negotiation 
mechanisms
         d - due to editor's choice and decision, document has moved to 
another 'version' or release which is completely independent from the 
previous one

Unless we are dealing with static pages, in my humble opinion, the most 
common cases are 2b and 2c. Any thoughts?

I also believe our main concern is strictly related to the document 
structure rather than its content (although some parts are crucial such as, 
for instance, the URL of an 'src' attribute or the ALT attribute, and so on).

I will keep on studying possible "solutions" and reading the other 
resources that have been posted on the list. In the meanwhile, I'd love to 
hear what you think about what I said.

Thank you,
-Gabriele

At 08.57 09/03/2005, you wrote:

>Here's a link to our Java source code shows how we identify the thing 
>that's causing the accessibility problem:
>
>http://checker.atrc.utoronto.ca/Identifier.java
>
>I know that source code is not the ideal method of documentation but it's 
>all we've got for now. If there's interest, I can produce a more readable 
>document later.
>
>If you look at the statements that start with "if (nameElement.equals 
>()){" you can see the element types we deal with.
>
>The fuzzy identifiers are stored as the element name and a list of 
>attributes stored in hashtableAttributes. For example to identify an image 
>we use the name of the element (img) and an attribute value (src attribute 
>value). This means that all images with the same src attribute value are 
>treated the same.
>
>The system is not perfect but it's been working so far and we hope to 
>improve it.
>
>Cheers,
>Chris
>

--
Gabriele Bartolini: Web Programmer, ht://Dig & IWA/HWG Member, ht://Check 
and ht://Miner maintainer
Current Location: Prato, Toscana, Italia
me@gabrielebartolini.it | www.gabrielebartolini.it | ICQ#129221447
 > "Lasciate ogne speranza, voi ch'intrate", Dante Alighieri, Divina 
Commedia, Inferno

Received on Wednesday, 9 March 2005 12:39:29 UTC