[meta] many IE6 regression test failures

[bill]

I see failures in the dijit regression on IE6 in:

• dijit/_base/place - tracking this in #12418
• Calendar_a11y - tracking in #13722
• Tree - didn't reproduce when running Tree.html or dijit/tests/tree/runTests.html by themselves, guess it's OK
• Editor_LinkDialog- didn't reproduce when running Editor_LinkDialog.html or dijit/tests/editor/runTests.html by themselves, guess it's OK

These tests failed for me but when I ran dijit/tests/runTests.html, but when I run dijit.tests.form.module by itself they work, so I guess they are OK:

• ButtonMixin
• ValidationTextBox
• TimeTextBox
• AutoCompleter_mixin
• ComboBox_mouse
• ExpandingTextAreaMixin

These tests failed for me but when I ran dijit/tests/runTests.html, but when I run dijit.tests._BidiSupport.module by itself they work, so I guess they are OK:

• dijit.tests._BidiSupport.form.noTextDirTextWidgets
• dijit.tests._BidiSupport.inheritance.Inher-MarkupContainers
• dijit.tests._BidiSupport.tree.TreeRootlessCustomIcons.html

Marking as high priority since IE6 is still a major platform for our users. I don't personally care though if we fix before or after the release.

[Chris Mitchell]

reducing severity to medium, we agreed on irc that this should not gate rc1

[bill]

Closing this, tracking remaining issues in other tickets.

There might still be a new memory leak on IE that's causing the regression suite when run altogether (in one pass) to fail. Could be from the event changes. Not sure.