Very similar to what bug 492956 used to be, except the error (now reported through bug 652494) is (a little) different: { [...] ERROR TEST-UNEXPECTED-FAIL | chrome://mochitests/content/a11y/accessible/[...] | an unexpected uncaught JS exception reported through window.onerror - gA11yEventListeners is undefined at chrome://mochitests/content/a11y/accessible/events.js:1438 } ***** Code reminder: http://mxr.mozilla.org/mozilla-central/search?string=gA11yEventListeners&case=1&find=%2Faccessible%2F.*%2Fevents\.js%24 { /accessible/tests/mochitest/events.js * line 1412 -- var gA11yEventListeners = {}; * line 1438 -- var listenersArray = gA11yEventListeners[event.eventType]; * line 1501 -- if (!(aEventType in gA11yEventListeners)) * line 1502 -- gA11yEventListeners[aEventType] = new Array(); * line 1504 -- var listenersArray = gA11yEventListeners[aEventType]; * line 1512 -- var listenersArray = gA11yEventListeners[aEventType]; * line 1523 -- gA11yEventListeners[aEventType] = null; * line 1524 -- delete gA11yEventListeners[aEventType]; } gA11yEventListeners should be defined everywhere: it's a calling context issue (due to executeSoon()/setTimeout()/events/...). No different than bug 492956 case, as I already wrote :-<

This won't help with this bug, but it updates/simplifies code I fixed in bug 492956.

(In reply to Serge Gautherie (:sgautherie) from comment #0) > it's a calling context issue (due to executeSoon()/setTimeout()/events/...). I tried a few tweaks, but without success. In bug 492956, I moved observerService into gA11yEventObserver... Here, I tried moving gA11yEventListeners into gA11yEventObserver too, fwiw: that works but it then complains about 'gLogger'... ;-< I would guess at a bad interaction between TestRunner.js and events.js. (but is it avoidable?)

This is not required (and will not fix this bug), but it helps isolate tests from one another, just in case. Succeeded as: https://tbpl.mozilla.org/?tree=Try&rev=df396ec55c88

Comment on attachment 588695 [details] [diff] [review] (Bv1) accessible/tests/mochitest/events.js: Use Services.jsm [Checked in: See comment 6+7] Review of attachment 588695 [details] [diff] [review]: ----------------------------------------------------------------- ::: accessible/tests/mochitest/events.js @@ +1607,1 @@ > }, do you need to keep trailing comma?

(In reply to Serge Gautherie (:sgautherie) from comment #3) > Created attachment 588746 [details] [diff] [review] > (Cv1) accessible/tests/mochitest/events.js: Call SimpleTest.finish() from > SimpleTest.executeSoon() > > This is not required (and will not fix this bug), but it helps isolate tests > from one another, just in case. what's benefits of tests isolation?

Comment on attachment 588695 [details] [diff] [review] (Bv1) accessible/tests/mochitest/events.js: Use Services.jsm [Checked in: See comment 6+7] https://hg.mozilla.org/mozilla-central/rev/6af8c11d727a

Comment on attachment 588695 [details] [diff] [review] (Bv1) accessible/tests/mochitest/events.js: Use Services.jsm [Checked in: See comment 6+7] (In reply to alexander :surkov from comment #4) > do you need to keep trailing comma? https://hg.mozilla.org/mozilla-central/rev/7edc6c4cf83d (Bv1a-bis) accessible/tests/mochitest/events.js: Fix typo in patch Bv1a.

Comment on attachment 588746 [details] [diff] [review] (Cv1) accessible/tests/mochitest/events.js: Call SimpleTest.finish() from SimpleTest.executeSoon() canceling review until comment #5 is addressed

Sad to see this stall. Serge can you clarify regarding test isolation?

Mass closing mochitest bugs that haven't had activity in the past 5 years. Please re-open or file a new bug with modern context if this is still relevant.