We expose value through IAccessibleValue as well as through the object attributes (valuetext). In this case: <div aria-valuetext="a" aria-valuenow="0"> We expose "a" through both the accessible value interface, and the valuetext object attribute. We made this decision because we thought the valuetext would usually provide richer information, and we wanted to cover the bases for AT that checked the value interface as well as for AT that check object attributes. The WAI-PF group wants us to expose both so that the AT can decide what gets precedence.

nsAccessible::GetAttributes should be fixed for that.

David: can you or someone elaborate a little further on this bug request? I've been reviewing the relevant code / previous bugs, and tinkering with nsAccessible.cpp and the test_valuechange.html test routine and can't see a problem where data is obscured ... values for aria-valuetext and aria-valuenow seem to be initialized and changed consistently. ( http://www.w3.org/TR/wai-aria/states_and_properties#aria-valuenow ) How do I reproduce / simulate the error condition?

How I'm watching values as set by the <!-- ARIA sliders --> definitions. Testing shows valuenow and valuetext propagations, then propagations based on values and change requests. Passed: 33 Failed: 12 Todo: 0 passed | Wrong value of 'slider_vn' - 11 should equal 11 passed | Wrong current value of 'slider_vn' - 11 should equal 11 passed | Wrong value of 'slider_vt' - t22 should equal t22 passed | Wrong current value of 'slider_vt' - 0 should equal 0 passed | Wrong value of 'slider_vnvt' - t33 should equal t33 passed | Wrong current value of 'slider_vnvt' - 33 should equal 33 failed | ===Change Requests====== passed | aValuetext value of 'slider_vn' - undefined should equal undefined passed | aValuenow value of 'slider_vn' - 6 should equal 6 failed | ===Value/Now Start===== passed | acc.value of 'slider_vn' - 11 should equal 11 passed | acc.currentValue of 'slider_vn' - 11 should equal 11 failed | ====ARIA Text Changed=== passed | acc.value of 'slider_vn' - 11 should equal 11 passed | acc.currentValue of 'slider_vn' - 11 should equal 11 failed | ====ARIA Now Changed==== passed | acc.value of 'slider_vn' - 6 should equal 6 passed | acc.currentValue of 'slider_vn' - 6 should equal 6 passed | test with ID = ' 'slider_vn' value changed' failed. No value change event. failed | ===Change Requests====== passed | aValuetext value of 'slider_vt' - hey! should equal hey! passed | aValuenow value of 'slider_vt' - undefined should equal undefined failed | ===Value/Now Start===== passed | acc.value of 'slider_vt' - t22 should equal t22 passed | acc.currentValue of 'slider_vt' - 0 should equal 0 failed | ====ARIA Text Changed=== passed | acc.value of 'slider_vt' - hey! should equal hey! passed | acc.currentValue of 'slider_vt' - 0 should equal 0 failed | ====ARIA Now Changed==== passed | acc.value of 'slider_vt' - hey! should equal hey! passed | acc.currentValue of 'slider_vt' - 0 should equal 0 passed | test with ID = ' 'slider_vt' value changed' failed. No value change event. failed | ===Change Requests====== passed | aValuetext value of 'slider_vnvt' - sweet should equal sweet passed | aValuenow value of 'slider_vnvt' - 3 should equal 3 failed | ===Value/Now Start===== passed | acc.value of 'slider_vnvt' - t33 should equal t33 passed | acc.currentValue of 'slider_vnvt' - 33 should equal 33 failed | ====ARIA Text Changed=== passed | acc.value of 'slider_vnvt' - sweet should equal sweet passed | acc.currentValue of 'slider_vnvt' - 33 should equal 33 failed | ====ARIA Now Changed==== passed | acc.value of 'slider_vnvt' - sweet should equal sweet passed | acc.currentValue of 'slider_vnvt' - 3 should equal 3 passed | test with ID = ' 'slider_vnvt' value changed' failed. No value change event.

(In reply to David Bolter [:davidb] from comment #0) > The WAI-PF group wants us to expose both so that the AT can decide what gets > precedence. David, please provide details how to fix this bug (this is requirement of good first bug).

Marco, since David is not in the office today, maybe you know decision on how to expose it? via valuetext object attribute?

Mark I'm very sorry - somehow I didn't notice the activity on this bug! Please always feel free to email my bugmail address directly or poke me on irc.mozilla.org #accessibility. I'll try not to miss the bug activity again. We give valuetext precedence at the MSAA level, and it seems to me we expose the numerical value via IA2 (IAccessibleValue). I believe this bug to be invalid. Sorry about the delay, but I need to go back to the spec group before we proceed on this bug. Alexander, Jamie what do you think in general about exposing valuetext as an attribute? (Also please check truth of my previous paragraph).

(In reply to David Bolter [:davidb] from comment #6) > We give valuetext precedence at the MSAA level, and it seems to me we expose > the numerical value via IA2 (IAccessibleValue). I believe this bug to be > invalid. I think that's right.