Re: Looking for clarity on Layout tables

On 20/02/2021 00:57, Marjanska, Martyna (SSC/SPC) wrote:
>
> Do we MUST use the WAI-Aria role=presentation on a Layout Table?
> 
> Is the WAI-Aria mandatory or only advisory, best practice or else?
> 
> I am looking for a concrete direction, a reference in WCAG, W3C, that 
> says clearly that layout tables must use role=presentation.
> 
> I found many references but none says role=presentation is required.

Opinions are split on this. Many will argue that using a table (and its 
intrinsic structure/relationships) clearly fails WCAG 2.1 SC 1.3.1

https://www.w3.org/TR/WCAG/#info-and-relationships

unless the table semantics are suppressed using role="presentation" or 
role="none". Whether not doing the latter is a big problem, or just a 
technicality, will depend on how complex the layout table is - only two 
columns and one row? yes, it's verbose in assistive technologies, but 
not to the point that it makes a site completely unusable; a 
super-complex table layout with a large number of rows/columns? probably 
more of an issue.

(of course this assumes that other success criteria are met ... 
meaningful reading order, correct focus order, reflow)

Others will point to 
https://www.w3.org/WAI/WCAG21/Techniques/failures/F46 which fails due to 
use of specific sub-sets of table markup that is reserved for data 
tables when the table is not a data table. That failure technique even 
includes the sentence:

"Although WCAG 2 does not prohibit the use of layout tables..."

(But noting that techniques are not normative, and only meant to be 
informative, so they can sometimes contain...less than ideal advice)

It's perhaps also worth keeping in mind that the hesitance with outright 
making layout tables fail 1.3.1 was very much a decision "of its time". 
When WCAG 2.0 (which introduced 1.3.1) first came out, in 2008, there 
was still a large number of sites using layout tables, so outright 
making them non-conforming was felt to be a bit too harsh. Now, in 2021, 
it would likely be that layout tables would be explicitly called out as 
being non-conformant, as there are far better and well-supported ways to 
control layout, compared to the arcane practice of layout tables.

So, based on a modern understanding of what 1.3.1 actually asks, most 
auditors will likely interpret this as meaning that layout tables in 
general are a (mild, in most cases) failure of WCAG. And that if markup 
can't be easily changed, a "quick fix" is to neutralise the table-like 
nature of the markup using role="presentation".

But if you take the informative sentence there in that one failure 
technique from 2008  as gospel, then you could argue that no, there is 
no hard requirement for not using layout tables. And using 
role="presentation" is more of a strongly suggested best practice. (I'd 
also note that positive as well as failure techniques referenced from 
WCAG are not exhaustive/comprehensive...they don't cover every possible 
scenario to give examples of every situation that is or isn't a fail).

Sorry, a bit of a non-answer. For my own part, I tend to fail layout 
tables that still expose their table-ness.

P
-- 
Patrick H. Lauke

https://www.splintered.co.uk/ | https://github.com/patrickhlauke
https://flickr.com/photos/redux/ | https://www.deviantart.com/redux
twitter: @patrick_h_lauke | skype: patrick_h_lauke

Received on Saturday, 20 February 2021 02:06:42 UTC