Add tests certifying offset isn't broken after a recent revert

[mgol]

Basically, we need tests certifying #2828 & #2836 work after the revert done in 49833f7. We should land them on all 3 branches.

[markelog]

We should land them on all 3 branches.

mm, doesn't seem worth it, but okay

[mgol]

Thanks. :)

[mgol]

@markelog Those tests are broken on all browsers, both on 2.2-stable and on 1.12-stable. Could you fix? I'll reopen for now.

[mgol]

This seems to be caused by using $ instead of jQuery. I'm surprised it works on master, is this test using a correct jQuery version?

[mgol]

Also, I've noticed this only now but the variable to which you assign assert.async() should be called done, not stop, to better align with other tests (in the previous nomenclature it would be called start, not stop so it was different than usual either way).

[markelog]

For some reason that does work with amd, that's reason i missed it.

should be called done, not stop

I think it is fine as is.

[mgol]

I think it is fine as is.

It's not stopping anything, it just notifies the test runner it's finished. It was called start in the old API as it was resuming the test suite (after calling stop() before).

It's also almost universally called done, both by other test runners and in our code that has switched to the new API. This is now the only place in the whole suite where it's called stop.

[markelog]

it just notifies the test runner it's finished

And that whole process is called stopping, whereas "stop" is synonym of "finish".

I think we are free to choice the name of our variables as long as those names sufficiently reflects their purpose.