quirks tests run in standards mode on IE9

[bill]

On IE9, the quirksmode tests are running in standards mode, leading to spurious failures (and also it means that we aren't testing quirksmode).

The failures on dojo/core are:

• html_quirks
• html_box_quirks
• uacss.sniffquirks

Failures in dijit/ are:

• TextBox_sizes.tundra.quirks
• TextBox_sizes.claro.quirks
• TextBox_sizes.soria.quirks
• TextBox_sizes.nihilo.quirks
• TextBox_sizes.a11y.quirks

[bill]

Note that this was split off from #12461.

See also ​http://stackoverflow.com/questions/5123326/force-ie9-into-quirks-mode. The iframe's mode seems inextricably tied to the main page's mode, It seems doubtful that even setting up an HTTP header specifying quirksmode would work.

[bill]

I'm not seeing the failures in dijit anymore. Was there a patch to IE9 or something? Mark, do you see the failures anymore?

[haysmark]

I still see systematic failures in the quirks tests in IE9.

[Colin Snover]

In [30009]:

IE9 does actually continue to do the wrong thing in quirks mode, it just doesn't do quirks mode in iframes of standards mode documents. Refs #12342, #14321.

[Colin Snover]

In [30012]:

IE9 does actually continue to do the wrong thing in quirks mode, it just doesn't do quirks mode in iframes of standards mode documents. Refs #12342, #14321. Backport to 1.7.

[Colin Snover]

In [30013]:

IE9 does actually continue to do the wrong thing in quirks mode, it just doesn't do quirks mode in iframes of standards mode documents. Refs #12342, #14321. Backport to 1.8.

[Colin Snover]

Things that don’t work:

1. Setting <meta http-equiv="X-UA-Compatible" content="IE=5"> (or EmulateIE7, or IE8) in the quirks mode child
2. Checking for has("quirks") in the child (it seemed to return true even though the browser is not actually doing all quirks mode stuff)

Therefore, these quirks mode tests are disabled in IE9 in the test runner. This effectively makes quirks mode in IE unsupported since we will never be running automated tests to confirm it is not broken, unless someone can find a working workaround.

[Colin Snover]

In [30014]:

Disable quirks mode tests in IE9+ in the test runner. Fixes #14321. !strict

[Colin Snover]

In [30015]:

Disable quirks mode tests in IE9+ in the test runner. Fixes #14321. Backport to 1.8. !strict

[Colin Snover]

In [30016]:

Disable quirks mode tests in IE9+ in the test runner. Fixes #14321. Backport to 1.7. !strict

[Colin Snover]

In [30017]:

Less convoluted browser test. Refs #14321. Backport to 1.8. !strict

[Colin Snover]

In [30018]:

Less convoluted browser test. Refs #14321. !strict

[Colin Snover]

In [30020]:

Disable more quirks tests in IE9+. Refs #14321. !strict

[Colin Snover]

In [30022]:

Disable more quirks tests in IE9+. Refs #14321. Backport to 1.8. !strict

[Colin Snover]

In [30024]:

Disable more quirks tests in IE9+. Refs #14321. Backport to 1.7. !strict

[bill]

Presumably you meant to mark this for 1.7.5 as 1.7.4 was already released.

[bill]

In [30107]:

If a module uses has("ie") etc., it needs to require() "dojo/sniff". "dojo/has" isn't sufficient. Refs #14321 !strict.

[bill]

This ticket is only half-fixed, because the dijit tests weren't updated. (And that's using the word "fixed" euphemistically, because what we really want is to run the quirks tests, not disable them.)

[bill]

Oh, nevermind, actually the quirks.html file, used by dijit to run the quirks tests, is able to run the tests in quirks mode even on IE9 and IE10, even from runTests.html.

Using quirks.html would also be an option to actually run the dojo core quirks tests, instead of just disabling them. I'll close this ticket again for now though.