Re: Key question on SC text and testability

The SC needs to be testable from the text of the SC, but the specific technique to accomplish it may vary by technology, which is why we have techniques.

For example, in 1.1.1 non-text content needs a text alternative, with some exceptions. The techniques talk about using the alt attribute in HTML for images and using the alt property for images in PDF, and the accessibilityName property in Flash, but all of these are in service of the provision of text for the image.

It is much harder when we are talking about adapting text. If we say “the page must not lose essential information when the user changes the line spacing” we are opening the author up to a complaint that a user sets the line spacing to 50x the default and if/when the page breaks the author won’t be able to say that they tested properly to cover this situation. As a result, authors will not know how wide a set of values to test to confirm that the page is conforming. 

On the there hand, if we specify that line spacing up to 1.5x is required then there are bounds to the testing and authors can feel confident that they are conforming.

We can’t say that 1.5x or 3x is what is required in techniques, since they are non-normative.


Andrew Kirkpatrick
Group Product Manager, Accessibility

On 4/18/17, 10:12, "Laura Carlson" <> wrote:

>Hi Andrew, Josh, and all,
>A key point of contention is in the Adapting Text SC discussion is in
>regard to SC text and testability.
>So I have a question to the group:
>Must SCs be testable on the face of the SC? In other words, from only
>reading the SC text, must testability be evident to provide normative
>testability? If it can't be tested true/false from the SC text, will
>it or won't it meet the SC criteria requirements?
>Some seem to say yes:
>"you need to say by how much or it isnt testable...the criteria need
>to be in the (success) criteria.    Understanding can only make it
>clearer — not define it further..." - Gregg C Vanderheiden,  24 Jan

>"if it can't be tested true/false from the SC text, it won't meet the
>SC criteria. You can flesh things out in the understanding doc, but
>the SC needs to be a true/false statement" - Alastair Campbell, 10 Feb

>Some seem to say no:
>"...moving the testing metrics for these SC to supporting documents,
>and I support that. I believe it's in keeping with the intent, which
>is that the user be able to choose their own display formatting, and
>how that's tested is essentially details (assuming, as Gregg
>Vanderheiden says, we have figured out how it will be tested, so we're
>sure it is testable)..." - Greg Lowney, 18 April 2017

>The WCAG 2.1 Success Criteria Wiki page says, "Success Criteria
>shall... Be testable through automated or manual processes."

>It doesn't go into specifics regarding this matter.
>If we can come to consensus on this point, it may help move the
>discussion forward.
>Thank you.
>Kindest Regards,
>Laura L. Carlson

Received on Tuesday, 18 April 2017 14:25:00 UTC