Re: ACTION-2039 proposal for revisions to aria-autocomplete description ready for review

I have made the following revisions to the proposal for new
aria-autocomplete text to address feedback provided in the June 2, 2016 ARIA
caucus. The revised proposal is now ready for review in the
action2039-autocomplete branch at:

http://rawgit.com/w3c/aria/action2039-autocomplete/aria/aria.html#aria-autoc
omplete

 

Revisions:

 

1. For inline autocomplete, revised language to removed the word "state"
when refering to selected text to address concerns that authors may think
that aria-selected needs to be applied to text input.

 

2. Added an author MUST statement requiring aria-controls to be used if
aria-autocomplete is set to list or both.

 

3. Changed the tense used in the values table from future to present.

 

4. Fixed some typos.

 

I also made additional improvements to address issues I discovered while
completing the above revisions.

 

Issue: if the user provides input for which the application is unable to
make a prediction of the user's intent, there will not be any suggestions
for how to complete the input. The property definition and value definitions
did not effectively communicate this conditional nature of the appearance of
predictions.

 

Resolution:

 

1. In the first paragraph that provides the property definition, changed
"triggers" to "could trigger" and added an if statement to the second half
of the definition.

2. In the value definitions table, incorporated the word "may" to reflect
the conditional nature of autocomplete behavior.

 

Issue: The paragraph describing limitations on the appropriate use of
aria-autocomplete to predictive behaviors implied authoring requirements
without a normative authoring statement.

 

Resolution: Added an author SHOULD statement describing when to set
aria-autocomplete to none even if input proposals are provided.

 

Issue: the prose for normative conditions was dense, making it difficult to
parse out all applicable normative requirements for some situations.

 

Resolution: Simplified the presentation of normative requirements by using
ordered lists where multiple requirements apply to the same set of
conditions.

 

Matt King