The Linux Foundation

 
Accessibility/IAccessible2/Minutes/20070424

From The Linux Foundation

Minutes for 2007/04/24

  1. Regrets
  2. Attendees
    • Pete Brunet, Itoh-san (Takashi Itoh), Michael Stewart, Andres Gonzalez, Barry Feigenbaum, Mike Squillace, Larry Weiss, Cathy Laws, George Kraft, Rich Schwerdtfeger
  3. Testing of IA2 implementations - George Kraft
    • LSB is the most successful ABI interface - used to determine if an app will run on all Linux distros
    • 3 aspects: spec - sample implementation - test suite
    • For the test suite, test assertions need to be written.
    • How can we test that an AT is using IA2 correctly or an app is implementing IA2 correctly?
    • What technique is needed to test IA2: a formal test suite? a validation tool?
    • The Linux Foudation could create a compliance standard, e.g. "Open Accessibility for Windows Compliant".
    • Rich, Janina, and George recently talked to Jim Zemlin. This would be part of the IA2 branding.
    • On Linux generally libs are being tested, but APIs like X have also been tested.
    • LSB uses POSIX testing methodolgy.
    • Want a structured testing harness with automation and journaling.
    • The Linux a11y group is currently testing app side of libatk.
    • They have written a few interfaces as a prototype, 3-4 PMs - all of libatk is probably a PY effort.
    • Testing event triggering is important; testing interface behavior is also important.
    • ABI testing is important on Linux because of many distros, but on Win there is only XP (and maybe Vista).
    • However, there are multiple apps. Each one needs to consistently implement IA2 so ATs don't need special casing.
    • The validation feature of AcTF (Accessibility Tools Framework) to be donated to Eclipse later this year could be used instead of a testing suite. Valudation rules need to be written.
    • Although a core set of validation rules can always be applied, additional AT specific validation rules are needed for JAWS, Window-Eyes, NVDA, etc.
    • Validation rules covering timing or sequencing of events are needed.

[Article] [Discussion] [View source] [History]