Tee Ming

Thanks for the reminder @benmccann . I've just learnt what a rebase is. I did a merge before that (hence the force push trying to undo that). Will use rebase...

> @s3812497 this has a merge conflict - can you look at it today? Else I'll look into it later. Sure thing! I'll have a look at it now.

Shoot the webkit checks are failing. Probably broke something

Cancelling `beforeunload` is currently broken in Safari. It only works when the page is in an iframe. EDIT: I have filed a bug report in the Webkit forums https://bugs.webkit.org/show_bug.cgi?id=255837

> or does it "just" not fix it for them? In the latter case I'd be ok with not executing that test until it's fixed in Safari. Yes, exactly that....

> Do we have no other tests currently that check that simply closing the browser triggers our `beforeunload` (which would then already have failed on Safari)? We do have several...

~After merging the latest changes from `master` this seems to be working just fine? (tested on local). Not sure which change was responsible for fixing this. Feel free to close...

> I've tried to update the test to point at an external URL (that we manage inside the test) but I can't seem to get it to run. I was...

The Ubuntu tests are failing because Firefox is navigating to the download URL instead of only downloading the response body. I tried reproducing this on an actual Ubuntu machine and...

I've no idea why the tests are so flaky even with the hacky "wait for idle network" statements. Would like some help with this or maybe I'll take another stab...