1242643 - [e10s] Fix and re-enable /js/xpconnect/tests/browser_dead_object.js

Status: RESOLVED FIXED

(Core :: DOM: Core & HTML, defect)

Description

[Tracy Walker [:tracy]]

Not sure this test case can be made to work in e10s.  Here is what I have.  Will isDeadWrapper work on a browser tab?  My attempts to do this in a ContentTask did not work either.

add_task(function* test() {

  let newTab = yield BrowserTestUtils.openNewForegroundTab(gBrowser, "http://example.com");
  yield BrowserTestUtils.removeTab(newTab);
  SimpleTest.executeSoon(function() {
    ok(Components.utils.isDeadWrapper(newTab),
        'Browser should be dead.');
    });
});

Comment 1

[Tracy Walker [:tracy]]

Bill,

Is it possible to make this test case e10s compatible?

For bug 773980, test that Components.utils.isDeadWrapper works as expected.

From skip if comment in browser.ini: 
  skip-if = e10s # Bug ?????? - test touches content (Components.utils.isDeadWrapper(contentWindow) - or maybe isDeadWrapper should take CPOWs into account?)

Comment 2

[Bill McCloskey [inactive unless it's an emergency] (:billm)]

I think there are a couple ways to do this. One way would be to use a ContentTask, but change how the test makes the wrapper dead. Instead of removing the tab, you would navigate to a different page. The structure would look sort of like this:

- Start up a content task.
- The content task would first get a reference to some part of the old page. Maybe content.document.
- Then it would navigate the page, perhaps with "content.location = 'http://example.com/a';".
- Then you would test that the reference to the old page is dead with isDeadWrapper.

However, one crucial piece is that the old page needs to have an onunload event handler. Otherwise it will go in the bfcache and we can't nuke it. So you'll need to make a test-specific page to start with and add an onunload event handler.

Comment 3

[Tracy Walker [:tracy]]

Attachment: bug_1242643.patch

Attach the patch in try, that fails on Linux.

https://treeherder.mozilla.org/#/jobs?repo=try&revision=b3fe6042437d

Comment 4

[Tracy Walker [:tracy]]

Attachment: bug_1242643[2].patch

Solved issue of failures on Linux  (had file path case issue)

https://treeherder.mozilla.org/#/jobs?repo=try&revision=f9a3d53cfcb8

Comment 5

[Jim Mathies [:jimm]]

Comment on attachment 8725866 [details] [diff] [review]
bug_1242643[2].patch

Review of attachment 8725866 [details] [diff] [review]:
-----------------------------------------------------------------

::: js/xpconnect/tests/browser/browser.ini
@@ +1,4 @@
>  [DEFAULT]
> +support-files =
> +  browser_deadObjectOnUnload.html
> +  

nit - whitespace line below the line with browser_deadObjectOnUnload.html

::: js/xpconnect/tests/browser/browser_dead_object.js
@@ +16,5 @@
> +    yield promise;
> +    return Components.utils.isDeadWrapper(doc);
> +  });
> +  is(contentDocDead, true, "wrapper is dead");
> +  yield BrowserTestUtils.removeTab(newTab); 

nit - trailing whitespace

Comment 6

[Tracy Walker [:tracy]]

Attachment: bug_1242643[2].patch

fixed nits on white space.

Comment 7

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/d0d03be48b9e

Comment 8

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/d0d03be48b9e

Comment 9

[Pulsebot]

https://hg.mozilla.org/integration/fx-team/rev/d45c2b6da860