BorderContainer_complex.html test problems

[bill]

I'm seeing a number of problems with this test:

1. When run on FF3.6/win I get an error message about dojo.global.dijit._curFocus.id is not defined.

2. Test should not be using console.error() to report errors, use the DOH assertions. I see that there is a doh.t(passed) at the very end of each test but why have that "passed" variable at all? doh.t(), doh.f(), and doh.is() are standard.

3. need comments for all assertions, otherwise if there's a failure for example in this set of tests we don't know which one failed:

doh.t(dijit.byId("tab2").selected);
doh.f(dijit.byId("tab1").selected);
doh.f(dijit.byId("tab3").selected);
doh.f(dijit.byId("tab4").selected);
doh.f(dijit.byId("tab5").selected);
doh.f(dijit.byId("tab6").selected);

All those doh.t() and doh.f() calls should have a second argument like "tab2", "tab1", etc.

4. dojo coding style: need brackets for all if(), else, while(), etc. There were also a lot of extra tabs inside of every doh.robot.sequence(), you only need one tab to indent.

[Katie Vance]

You are right, this test was one of the first that I automated and it did not follow the standards. Attached is how the test should have been written. I did not remove dojo.global.dijit._curFocus.id because it is working for me. Many other dijit robot tests use the same method for determining what has focus. Let me know if you still have problems with that.

[bill]

Thanks, that looks much better. Using dojo.global.dijit._curFocus is fine, it's just that I was getting a test failure for some reason. Maybe robot flakiness ignoring the first click event, not sure, it seems to be working now.

[bill]

(In [23227]) Test cleanup from Katie, fixes #11969, thanks!

[bill]

(In [23863]) Avoid spurious timing failures in AccordionContainer tests plus other cleanup, refs #11969.