[patch] [cla] Slider: use doh.robot to test keyboard a11y

[Joseph Scheuhammer]

It is time consuming to manually determine whether dijit.form.Slider is keyboard accessible, and that it implements the keystrokes as documented. It would be better if such testing could be automated.

Actually, this should be done for all the widgets, but one has start somewhere.

[Joseph Scheuhammer]

The attached unit test file, "test_keyboard_Slider.html", tests putting focus on a slider via Tab and Shift+tab keystrokes, and manipulating the values of a slider using Home, End, PgUp, PgDn, and the arrow keys. It should be located in ".../dijit/tests/form/robot/test_keyboard_Slider.html".

It was tested against r15370 of trunk, and by loading the test file with FF3 and IE7.

[Joseph Scheuhammer]

The patch file, "7774.patch", makes it such that loading ".../dijit/tests/robot.hml" will run a11y and robot unit tests for slider.

The patch contains a file "test_a11y_Slider.html" that replaces the previous attachment "test_keyboard_Slider.html". The differences are:

• uses robot mouse events to put focus on the slider (previously used element.focus(), but that sometimes fails to put focus on the slider before robot keypresses are queued up).
• tests tab focussing of slider.
• tests keystroke manipulation of slider.
• tests that ARIA role is properly set (i.e. "slider").
• tests that ARIA valuemin, valuemax, and valuenow are set or updated as the slider is manipulated.

The aria tests do not involve doh.robot, and could be placed in the current ".../dijit/tests/test_Slider.html" file. It depends on whether it's useful to put all a11y tests in one place or not.

[Joseph Scheuhammer]

• Merged tests from separate "test_a11y_Slider.html" with existing "test_Slider.html"; see attachment "7774a.patch".
• a11y tests applied to all but the disabled sliders (previously, tested only a single slider).
• run with FF3/Mac OS X and IE7/WinXP.

[Joseph Scheuhammer]

Attaching another patch, "7774b.patch", to fix a couple of bugs:

• the assertEquals(...) test for a sequence of expected vs. actual values in a11yKeystrokeTest() was not, in fact, comparing anything. Fixed to compare real expected and actual values.
• added keyStrokeSetup() to pull common intializing code out of a11yKeystrokeTest().
• removed slider3Keystrokes test fixture as the actual values are mysteriously in error.

[Becky Gibson]

fixed in changeset [15564]