XHR test failure on IE8

[bill]

Works in 1.7, fails on trunk:

    _AssertFailure: assertEqual() failed: 	expected
		[object Object]
	but got
		undefined

     ERROR IN: 		 function jsContentHandler(t){
							var jsonObj = {
								foo: "bar",
								baz: [
									{ thonk: "blarg" },
									"xyzzy!"
								]
							};
							t.is(jsonObj,
								dojo._contentHandlers["javascript"]({
									responseText: "("+dojo.toJson(jsonObj)+")"
								})
							);
							t.t(dojo._contentHandlers["javascript"]({
									responseText: "true;"
								})
							);
							t.f(dojo._contentHandlers["javascript"]({
									responseText: "false;"
								})
							);
						} FAILED test: ../../dojo/tests/_base/xhr.html::t::jsContentHandler 0 ms

[bill]

Rawld, this started with [27764].

[Rawld Gill]

In [27791]:

yet another attempt to fix defective dojo.eval without breaking current usages; fixes #14749; refs #14406; !strict

[liucougar]

dojo.eval has always returns a value in IE since the API is available. I think it's more important to keep the behavior of returning a variable for IE than to make IE eval in true global scope: this is a regression from earlier versions of dojo (dojo.eval has always been not able to eval in global scope in IE, but it always returns value in IE)

[Rawld Gill]

Replying to liucougar:

dojo.eval has always returns a value in IE since the API is available. I think it's more important to keep the behavior of returning a variable for IE than to make IE eval in true global scope: this is a regression from earlier versions of dojo (dojo.eval has always been not able to eval in global scope in IE, but it always returns value in IE)

Not sure why you reopened. dojo.eval does return a value in IE with the latest change. Note that this ticket was closed with a changeset that missed a file...that mistake was immediately corrected in [27798]