|Submitter||Diff||Changes||Open Issues||Last Updated|
|Error loading review requests:|
The push to mozilla central for bug 1136939 shows that we're failing somewhere in between tests 3 and 4 in the child. Both of those tests stop waiting for notifications, so more logging is needed to see where we stop. As I've been working on this test, I've been getting annoyed with my hand-rolled, generator-based control flow. It's hard to see what event is supposed to trigger when, as the test is based on two (one in the child and one in the parent) global generators that could be triggered by any number of events. I've converted the child to use Task.jsm and the parent to use SpawnTask.js and promises, which at least allows me to see what's supposed to trigger what.
Comment on attachment 8825575 [details] Bug 1330154 - Try to add more error handling. https://reviewboard.mozilla.org/r/103698/#review104608 Phew, this is a pretty complicated test. I wonder if it'd be better to re-write this down the road as a mochitest-browser test. That way, we get ContentTask, which might simplify some of this. Or maybe we should just make ContentTask something that can also work with mochitest-plain with mozbrowser iframes. Anyhow, adding more instrumentation and error handling to help reason about this test is a good thing. r=me!
Pushed by firstname.lastname@example.org: https://hg.mozilla.org/integration/autoland/rev/0bac600f111b Try to add more error handling. r=mconley
Patch landed few hours ago https://hg.mozilla.org/mozilla-central/rev/0bac600f111b