Recently I switched to a multi-session approach here to run behat tests consistently, without focus issues and not interfering my normal work.

So, apart from my "stronk7" session, I've four "testing0X" sessions, each one running selenium in different port.

Then from my account I can init & run behat jobs against those selenium servers, each one starting the browsers in own session.

And it works perfectly, zero focus problems and 100% passes under Chrome. But, with Firefox there are 14 constant failures happening all the time (and few more random that are passed on rerun).

The 3.2 release testing matrix is plenty of examples (macos/firefox row).

Here there are some examples (of the 14 failures):

• http://pastebin.com/tiAFECCv
• http://pastebin.com/C55tEH6t

And, no matter I rerun N times, they continue failing all the time.

BUT, if I launch the run and quickly switch to that "testing0X" user to see the execution while it happens, then the 14 errors pass without a problem, see:

• http://pastebin.com/3UenYST3

So, while it's possible to workaround the probem and execute the --reruns in current session, it would be amazing to discover why those errors happen when running in another session and, if possible, make them pass. I really don't know what's in common with the failures.

That's it. Not critical at all, as far as there is a workaround, but would be great to allow runs being executed in other sessions to pass (I bet something similar will happen with Windows).

Ciao