Fix a XHR layout test that may cause a falkiness of other tests.

Fix a XHR layout test that may cause a falkiness of other tests.

The flakiness dashboard results after [1] shows the flakiness is not caused
by the flaky test itself, but caused by another test. The test logs say
xmlhttprequest-responseText-exception.html is the culprit.
This CL fixes the flakiness caused by repeated readystatechange events
on loading.
This CL makes the test use js-test.js, as recommended in [2].

1: https://src.chromium.org/viewvc/blink?view=rev&revision=190652
2: https://docs.google.com/document/d/1Yl4SnTLBWmY1O99_BTtQvuoffP8YM9HZx2YPkEsaduQ/edit#heading=h.ui2te0d6ongo

BUG=415872
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=190716

[yhirano, 2015-02-24 02:13:35]

yhirano@chromium.org changed reviewers:
+ hiroshige@chromium.org

[yhirano, 2015-02-24 02:13:36]

https://codereview.chromium.org/950063004/diff/1/LayoutTests/http/tests/xmlht...
File
LayoutTests/http/tests/xmlhttprequest/xmlhttprequest-responseText-exception.html
(right):

https://codereview.chromium.org/950063004/diff/1/LayoutTests/http/tests/xmlht...
LayoutTests/http/tests/xmlhttprequest/xmlhttprequest-responseText-exception.html:5:
description('XMLHttpRequest.responseText should not throw even when not (loading
or done).');
I changed this description because it was wrong: responseText does not throw
when xhr is not loading and not done.
https://xhr.spec.whatwg.org/#the-responsetext-attribute

[hiroshige, 2015-02-24 03:12:10]

On 2015/02/24 02:13:36, yhirano wrote:
>
https://codereview.chromium.org/950063004/diff/1/LayoutTests/http/tests/xmlht...
> File
>
LayoutTests/http/tests/xmlhttprequest/xmlhttprequest-responseText-exception.html
> (right):
> 
>
https://codereview.chromium.org/950063004/diff/1/LayoutTests/http/tests/xmlht...
>
LayoutTests/http/tests/xmlhttprequest/xmlhttprequest-responseText-exception.html:5:
> description('XMLHttpRequest.responseText should not throw even when not
(loading
> or done).');
> I changed this description because it was wrong: responseText does not throw
> when xhr is not loading and not done.
> https://xhr.spec.whatwg.org/#the-responsetext-attribute

Is this CL for
- trying to fix the flakiness by refactoring?
- or trying to get more detailed logs for investigation?
- or both?
Clarifying it will be good because this CL refactors the test more than expected
from the current CL description.

[Optional] This CL refactors the test largely. Doesn't it cause investigation
harder (e.g. harder to determine which part of the changes makes tests green)?

[yhirano, 2015-02-24 03:34:41]

On 2015/02/24 03:12:10, hiroshige wrote:
> On 2015/02/24 02:13:36, yhirano wrote:
> >
>
https://codereview.chromium.org/950063004/diff/1/LayoutTests/http/tests/xmlht...
> > File
> >
>
LayoutTests/http/tests/xmlhttprequest/xmlhttprequest-responseText-exception.html
> > (right):
> > 
> >
>
https://codereview.chromium.org/950063004/diff/1/LayoutTests/http/tests/xmlht...
> >
>
LayoutTests/http/tests/xmlhttprequest/xmlhttprequest-responseText-exception.html:5:
> > description('XMLHttpRequest.responseText should not throw even when not
> (loading
> > or done).');
> > I changed this description because it was wrong: responseText does not throw
> > when xhr is not loading and not done.
> > https://xhr.spec.whatwg.org/#the-responsetext-attribute
> 
> Is this CL for
> - trying to fix the flakiness by refactoring?
> - or trying to get more detailed logs for investigation?
> - or both?
> Clarifying it will be good because this CL refactors the test more than
expected
> from the current CL description.
> 
This CL aims to fix the flakiness. It is not a refactoring because it changes
the behavior.
I fixed the issue description.

> [Optional] This CL refactors the test largely. Doesn't it cause investigation
> harder (e.g. harder to determine which part of the changes makes tests green)?
I know the old test causes the flakiness on the succeeding test.
Even without the flakiness, the new test is easier to read and maintain. If the
flakiness persists, I will investigate more on the new test.

[hiroshige, 2015-02-24 03:58:10]

lgtm.
The CL description is now clear. So it was a bug in this layout test script (not
XHR implementation), right?
Let's see whether the bots turn green.

[yhirano, 2015-02-24 04:02:49]

Thanks.

On 2015/02/24 03:58:10, hiroshige wrote:
> The CL description is now clear. So it was a bug in this layout test script
(not XHR implementation), right?
Yes, I think so.

[yhirano, 2015-02-24 04:03:15]

The CQ bit was checked by yhirano@chromium.org

[commit-bot: I haz the power, 2015-02-24 04:04:09]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/950063004/1

[commit-bot: I haz the power, 2015-02-24 05:23:41]

Committed patchset #1 (id:1) as
https://src.chromium.org/viewvc/blink?view=rev&revision=190716