Skip to content

Background

Matt King edited this page Feb 12, 2023 · 1 revision

Background, Objectives, and Workstreams

1. Brief history

2007 time frame - To get ARIA 1.0 off the ground:

  • Promise to browser vendors that supporting ARIA attributes carries no requirements to change rendering or interaction behaviors
  • Promise to assistive tech developers ARIA will not specify how assistive technologies will behave
  • Thus, ARIA does not include normative MUST statements for assistive technologies.

2014 - ARIA 1.0 ships:

  • Major accomplishment! Difficult to overstate importance.
  • However, had one big gap: shipped without authoring guidance.
  • Confusion among both web authors and assistive tech devs increased over time.
  • Web authors based their code on how some A/T behaved, without realizing some of that behavior was not consistent with spec.
  • A/T often adjusted their code based on how some popular web sites behaved, without realizing that some ARIA usage was not consistent with spec.
  • This setup a pernitious feedback loop, incorrect A/T behavior led to incorrect web author implementation, which led to more incorrect A/T behavior, etc.

Problem

ARIA has not been consistently interpreted across the industry:

  • Web sites use it incorrectly.
  • A/T support is not consistent or reliable.
  • Browser support for some attributes is incomplete or incorrect.

Note: this does not mean all A/T need to behave identically in order for support to be reliable and consistent.

Impact

  • Utility of ARIA falls far short of its realistic potential.
  • Accessible web GUIs are unnecessarily complex to engineer and use.
  • Sites aiming for broad audiences have to avoid certain types of GUI elements or offer alternatives.

2014 strategy for resolution

  1. Eliminate confusion with clear authoring guidance, including a library of realistic examples.
  2. Help A/T developers leverage the library as a way of measuring and testing the quality of their support for ARIA.

2014 - Today

100% focused on step one of strategy: Eliminate confusion with clear authoring guidance.

Approach:

  • Revived interest in W3C WAI-ARIA Authoring Practices
  • Built authoring practices task force.
  • Adopted W3C review and conflict resolution processes.
  • Establish guiding principles for ARIA pattern design.
  • Deprecated/rewrote all previous draft content using these principles and processes.
  • Focus on building example library; examples were external references in early drafts.
  • Established consistent documentation template for examples.
  • Built automated regression test system to ensure stability and reliability of all ARIA claims.

ARIA-AT Objectives

  1. Help assistive technology developers converge on a set of clear norms for baseline support of WAI-ARIA.
  2. Help web developers understand the current state of support for WAI-ARIA by assistive technologies.

Workstreams

Propose three workstreams:

  1. Project and process design
  2. Infrastructure development
  3. Assessment methodology development

Project and process Design Workstream

  • What are the deliverables and their scope?
  • What are use cases for the deliverables of this project?
  • How is assessment and feedback data controlled and managed?
  • What processes are their for vetting data?
  • How do processes ensure vendor neutrality?
  • What are the processes for correcting assessment errors?
  • How are disagreements resolved?

Infrastructure Development Workstream

Design and build systems for:

  • Compiling tests and testing instructions for each test case (ARIA design pattern example)
  • Collecting assessment data
  • Aggregating assessment data and calculating assessment scores
  • Publishing assessments
  • Gathering and responding to feedback

Assessment Methodology Development

  • For each combination of browser and A/T, define the following for each ARIA design pattern example:
    • Minimal, must-have expectations
    • Desirable should-have or may-have expectations
    • Test conditions
    • Test instructions
  • Define how the assessment score will be calculated for each example?
Clone this wiki locally