Bug 1076975 needed a test. Carrying over r+ from khuey.

Backed out in https://hg.mozilla.org/integration/mozilla-inbound/rev/469dbf49630e - at the very least, you need whatever the actual condition is for "fails-if e10s", dunno if there will wind up being more failures (and you separately broke hell out of Windows debug, so we probably aren't going to know whether or not it passes there).

Also, dunno how frequent it will be, but https://treeherder.mozilla.org/logviewer.html#?job_id=5140596&repo=mozilla-inbound is it hanging on b2g debug emulator.

And https://treeherder.mozilla.org/logviewer.html#?job_id=5142604&repo=mozilla-inbound is it failing on b2g opt emulator.

And https://treeherder.mozilla.org/logviewer.html#?job_id=5144741&repo=mozilla-inbound is it failing on 10.6 debug

(In reply to Phil Ringnalda (:philor) from comment #2) > Backed out Yeah, there was a small race in the test.

Still failing on e10s. Backed out. https://hg.mozilla.org/integration/mozilla-inbound/rev/f34d4207cb44 https://treeherder.mozilla.org/logviewer.html#?job_id=5162783&repo=mozilla-inbound

And Android: https://treeherder.mozilla.org/logviewer.html#?job_id=5169380&repo=mozilla-inbound

I can't actually land this until we figure out what to do in bug 1068838. Also, since bug 1076975 landed with a test already there's no reason this needs to be related any more.

This version uses the temp dir to work around the fact that child processes can't currently access the profile directory. It still doesn't work though because the file picker in the child process hands out a file-backed blob.