614079 - test-windows.testActiveWindow fails with "Non-browser windows aren't handled by this module"

Status: RESOLVED FIXED

(Add-on SDK Graveyard :: General, defect, P1)

Description

[Matěj Cepl]

info: executing 'test-windows.testActiveWindow'
console: [JavaScript Error: "gBrowser is not defined" {file: "chrome://browser/c
ontent/browser.js" line: 330}]
console: [JavaScript Error: "gBrowser is not defined" {file: "chrome://browser/c
ontent/browser.js" line: 330}]

...

 test-windows.testActiveWindow: timed out

Comment 1

[Matěj Cepl]

I have different error with 6f113e4f and Mozilla/5.0 (X11; Linux x86_64; rv:2.0b9) Gecko/20110114 Firefox/4.0b9 (Fedora native build):

info: executing 'test-windows.testActiveWindow'
info: pass: Correct number of browser windows
info: pass: Correct number of windows returned by iterator
error: TEST FAILED: test-windows.testActiveWindow (failure)
error: fail: Non-browser windows aren't handled by this module (({on:function on() {[native code]}, removeListener:function removeListener() {[native code]}, get title function () {[native code]}, close:function close() {[native code]}, activate:function activate() {[native code]}, get tabs function () {[native code]}}) != ({on:function on() {[native code]}, removeListener:function removeListener() {[native code]}, get title function () {[native code]}, close:function close() {[native code]}, activate:function activate() {[native code]}, get tabs function () {[native code]}}))
info: Traceback (most recent call last):
  File "resource://testpkgs-addon-kit-tests/test-windows.js", line 278, in focusListener
    nextStep();
  File "resource://testpkgs-addon-kit-tests/test-windows.js", line 253, in nextStep
    testSteps.shift()();
  File "resource://testpkgs-addon-kit-tests/test-windows.js", line 216, in null
    test.assertEqual(windows.activeWindow, window2, "Non-browser windows aren't handled by this module");
  File "resource://testpkgs-api-utils-lib/unit-test.js", line 195, in assertEqual
    this.fail(message);
  File "resource://testpkgs-api-utils-lib/unit-test.js", line 113, in fail
    console.trace();

Comment 2

[Matěj Cepl]

Still present on FF4b9 (Mozilla/5.0 (X11; Linux x86_64; rv:2.0b9) Gecko/20110121 Firefox/4.0b9) and 0f9e14ea9d9f3251cbe925e2b69facd31376be8a.

Comment 3

[Brian Warner [:warner :bwarner]]

FWIW, I'm not seeing this test failure against the same jetpack revision and FF4.0b9 on OS-X. test-widget.testConstructor fails, but test-windows.testActiveWindow appears to pass.

Comment 4

[Matěj Cepl]

Yes, I can confirm that I cannot reproduce it with Mozilla/5.0 (X11; Linux x86_64; rv:2.0b10) Gecko/20110125 Firefox/4.0b10 and 0f9e14ea9d9f3251cbe925e2b69facd31376be8a

Comment 5

[Myk Melez [:myk] [@mykmelez]]

I see this problem on tip with the latest Firefox 4 nightly build on 32bit Linux, but it's intermittent.

Comment 6

[Matěj Cepl]

Attachment: output of cfx testall -v

Again could reproduce with Mozilla/5.0 (X11; Linux x86_64; rv:2.0b11) Gecko/20110213 Firefox/4.0b11 and 30f4b497aa5267526b93fb615bfd539c2ff3fa7a

Comment 7

[Alexandre Poirot [:ochameau]]

Attachment: Modify the test to have the failure at all executions

Here is a simple patch that add a setTimeout in order to have this failure at all execution.
It appears that instead of having window2 we get window3.
For now, I didn't found why we get window3 and how to force to have it.

FYI, this failure doesn't exists on windows.

Comment 8

[Alexandre Poirot [:ochameau]]

Attachment: Fix

Here is a proposal fix. I'm not really satisfied with this solution,
but I didn't managed to force window2 to be the active window ...

I tried :
- Calling blur before focus, but it didn't work this time.
https://bug606007.bugzilla.mozilla.org/attachment.cgi?id=499172
- Using nsIWindowWatcher.activeWindow without any success.
https://bug606007.bugzilla.mozilla.org/attachment.cgi?id=509111
- Waiting on focus event bubble instead of capturing.

Comment 9

[Myk Melez [:myk] [@mykmelez]]

Comment on attachment 513148 [details] [diff] [review]
Fix

This patch seems to have been truncated at the 80th column.  Perhaps a copy/paste issue?

Comment 10

[Alexandre Poirot [:ochameau]]

Attachment: Valid fix

Oooops sorry for this error. You were right, copy'n paste between VM is not a good idea!

Comment 11

[Myk Melez [:myk] [@mykmelez]]

Comment on attachment 513242 [details] [diff] [review]
Valid fix

Hmm, yeah, it isn't ideal, and we should come up with a more robust fix for window activation on Linux in the long run, but this is a good short term fix.

r=myk! And a=myk for landing during the freeze.

Comment 12

[Myk Melez [:myk] [@mykmelez]]

https://github.com/mozilla/addon-sdk/commit/5c8e072317ddfcd12a20b455d04c6b38df2a7cfa

Comment 13

[Ayan Shah[:ashah]]

Happened for me on Ubuntu with Add-on SDK 1.0b3RC1.

Here are the logs:
http://pastebin.mozilla.org/1088106

SDK: 1.0b3 RC1
Browser: FFx 4.0b11
Platform: Ubuntu

The command that is executed from inside the integration script is:
"cfx testall -b /Applications/Firefox4.0b11/Firefox.app"

Comment 14

[Alexandre Poirot [:ochameau]]

I'm not able to reproduce this one, nor on windows, nor on ubuntu 10.10 VM.
Myk: are you able to reproduce it ?

Ayan: during test execution were you doing anything or just let the test run ? I suppose that this bug is intermitent or did you got it fail every time ?

Comment 15

[Matěj Cepl]

(In reply to comment #14)
> I'm not able to reproduce this one, nor on windows, nor on ubuntu 10.10 VM.
> Myk: are you able to reproduce it ?
> 
> Ayan: during test execution were you doing anything or just let the test run ?
> I suppose that this bug is intermitent or did you got it fail every time ?

This is Mozilla/5.0 (X11; Linux x86_64; rv:2.0b12) Gecko/20100101 Firefox/4.0b12 and I have 100% fail here on this one test (and this test only) with today's master (24c8f4a6). Just running cfx testall -v and nothing else.

Comment 16

[Matěj Cepl]

Just to note ... still alive with 5e251f5d and Mozilla/5.0 (X11; Linux x86_64; rv:2.0) Gecko/20100101 Firefox/4.0 (Fedora/Rawhide package)

Comment 17

[Myk Melez [:myk] [@mykmelez]]

(In reply to comment #14)
> I'm not able to reproduce this one, nor on windows, nor on ubuntu 10.10 VM.
> Myk: are you able to reproduce it ?

Yes, I see it frequently in my 32bit Ubuntu Linux 10.04 VM when running all `addon-kit` tests.  When I run only `windows` tests, however, I don't see it.  So the problem could be in the interaction between the `windows` tests and other tests.

My Linux installation is fairly stock; I haven't made too many changes; so the window manager is the default one that ships with Ubuntu, etc.  But let me know if I can provide any additional details about my configuration that might help you to reproduce the problem!

Comment 18

[Ayan Shah[:ashah]]

I can see this on SDK 1.0b4rc3, Firefox 4.0, ubuntu 9.0 while running cfx testall

The logs are here - http://pastebin.mozilla.org/1190409

Comment 19

[Alexandre Poirot [:ochameau]]

A quick note about bug 599204, that is extremelly close to this one.
But the test from windows-utils seems more robust than the current test from windows API!

Comment 20

[Myk Melez [:myk] [@mykmelez]]

We can either fix the test or disable it, but we have to squash all intermittent test failures so Firefox/Gecko developers can rely on our automated test infrastructure to tell them when they do something that breaks the SDK.

Comment 21

[Alexandre Poirot [:ochameau]]

Attachment: Disable intermittent failing test

As I'm really unable to reproduce it in a any linux VM, I disabled the failing assertion.

Comment 22

[Irakli Gozalishvili [:irakli] [:gozala] [@gozala]]

Comment on attachment 525401 [details] [diff] [review]
Disable intermittent failing test

Looks good.

Comment 23

[Alexandre Poirot [:ochameau]]

Landed:
https://github.com/mozilla/addon-sdk/commit/09d2287b7515b33a63f4f2a82bb5a583162dae70