Re: Web Accessibility Tutorials: Images and Tables

Hi Eric,

thanks for the additional pointers. 


Olaf


On 17 Sep 2014, at 09:46, Eric Eggert <ee@w3.org> wrote:

> On 16 Sep 2014, at 19:37, Olaf Drümmer wrote:
> 
> Hi Eric,
> 
> the context for this discussion is
> http://www.w3.org/WAI/tutorials/tables/
> though it seems somebody change the two instances of
> alt
> into
> alt=""
> already.
> 
> So it seems that the issue is resolved. Thank you for your feedback.
> 
> Regarding the syntax discussion:
> I am interested in learning where you have found an indication that using alt without a value is just fine. I have provided sources that make it very clear for HTML4 that it is incorrect (alt is not an attribute of type boolean), and that one can argue it is not correct for HTML5 (and I admitted the wording is not very crisp, but wiithout wording elsewhere that seems to allow such use of alt I would insist my interpretation is correct).
> 
> In HTML5 it says
> http://dev.w3.org/html5/html-author/#empty-attr
> 
> “An empty attribute is one where the value has been omitted. This is a syntactic shorthand for specifying the attribute with an empty value, and is commonly used for boolean attributes.”
> 
> (Emphasis mine. )
> 
> In HTML4 that was not meant to be the case, yet browsers did it like that all the time, so I guess they are just paving cowpaths there.
> 
> Best, Eric
> 
> Olaf
> 
> On 16 Sep 2014, at 16:54, Eric Eggert ee@w3.org wrote:
> 
> hi Olaf,
> 
> On 16 Sep 2014, at 16:33, Olaf Drümmer wrote:
> 
> Hi Eric,
> 
> after some research I found this (3.2.3.1 Empty Attribute Syntax )
> http://dev.w3.org/html5/html-author/#empty-attr
> 
> Also see:
> http://www.w3.org/TR/html/infrastructure.html#boolean-attributes
> though wording is much less straightforward here.
> 
> So except for boolean attributes an attribute without a value is illegal. And even for boolean attributes it is illegal for XHTML. This implies that the synatx used on the mage tutorial overview page is simply incorrect correct and requires fixing.
> 
> The empty attribute syntax is not only valid for boolean attributes, just “commonly used” for them. The spec is quite unclear for developers here. Basically the behavior is like this:
> 
> If an attribute is used without a value, assume its value is an empty string (attr -> attr="").
> If a boolean attribute’s value is an empty string, set it to true (bool="" -> bool="bool").
> 
> That means an attribute like alt would transform to alt="";
> and an attribute like disabled would transform to disabled="" and then to disabled="disabled".
> 
> That said, I’m a bit lost on where we have used (or should have used??) an alt without an empty value. We nowhere advices to use the alt without the empty string attached. It is always
> 
> alt="" (which is fine in XHTML as well as HTML)
> 
> and not
> 
> alt (which is syntactically correct in HTML, but not used for the reasons outlined in my previous email)
> 
> alone.
> 
> Can you point me to the section of the page you have an issue with? And to clarify, it could probably be helpful how you’d phase the sentence affected.
> 
> BTW - Personally I think, when writing a decent spec it should not have an impact that some agents can handle illegal syntax - the spec should simply get it right.
> 
> I think this is a different discussion ;-)
> 
> Best, Eric
> 
> Olaf
> 
> On 15 Sep 2014, at 18:38, Eric Eggert ee@w3.org wrote:
> 
> Hi Olaf,
> 
> [3]
> The last bit goes back to my limited familiarity with details in HTML:
> - is
> alt
> really equivalent to
> alt=""
> ?
> - if so, why do the examples in W3C WAI related content I have run into (and definitely those in the image tutorial) use
> alt=""
> ?
> 
> In modern HTML5 browsers, alt will be interpreted as alt="", but I don’t consider using this as best practice, especially when educating website authors:
> 
> Having the ="" in there show that the empty alt is a deliberate decision and not something forgotten.
> 
> Older browsers or assistive technology might treat alt without a value as missing and might announce the filename instead of nothing.
> 
> Using two syntaxes like this with minor differences and little to no gain would only confuse some people and would be hard to explain, imho.
> 
> Best, Eric
> 
> --
> 
> Eric Eggert, Web Accessibility Specialist
> WAI-ACT Project
> 
> I’m yatil on IRC.
> 
> --
> 
> Eric Eggert, Web Accessibility Specialist
> WAI-ACT Project
> 
> I’m yatil on IRC.
> 
> --
> 
> Eric Eggert, Web Accessibility Specialist
> WAI-ACT Project
> 
> I’m yatil on IRC.
>