pushState Test Case

[mdawaffe]

This isn't a useful bit of code, just a test case to ground future pushState work for 3.2. Or be ignored by future pushState work for 3.2 :)

Attached is a test case for determining browser behavior when using pushState.

The numbered links use AJAX and pushState. The "Other" link is a real (non-AJAX) link to an external site.

If the state of the page is refreshed via AJAX, you'll see a var_dump( $_POST ). If the state of the page is refreshed by a page refresh, you won't.

An example of something this test says is broken in Safari 5.0.3 (6533.19.4):

1. Load the page (state = 0: pageload)
2. click "One" (state = 1: ajax)
3. click "Two" (state = 2: ajax)
4. click "Three" (state = 3: ajax)
5. click "Other" (state = external site: pageload)
6. go Back in browser history (state = 3: ajax or just history?)
7. go Back (state = 2: ajax or just history?)
8. go Back (state = 1: ajax or just history?)
9. go Back (state = 0: ajax). Everything is fine up to this point, though it's odd we do an AJAX request to get here.
10. refresh the page once, either by CTRL-R or by hitting enter in the URL bar (state = 3: ajax). That's wrong. What happened? Why are we here?
11. refresh the page a second time (state = 0: pageload). As expected.

An example of something the test says is broken in Chrome10.0.642.2 dev:

1. Do pretty much anything. Chrome Dev fires a popState event on the first page load.

It could be a badly written test, or a badly written implementation of push/popState handling, or both. These are the kinds of scenarios we'll need to test, though.

[markjaquith]

Chrome Dev fires a popState event on the first page load.

That is expected behavior. The HTML5 (or just HTML, now) spec changed in August to make this clear: ​http://html5.org/tools/web-apps-tracker?from=5345&to=5346

Safari is probably the misbehaving one here.

6. go Back in browser history (state = 3: ajax or just history?)

History. This was the last state of the document before we jumped over to example.com

7. go Back (state = 2: ajax or just history?)

Ajax

8. go Back (state = 1: ajax or just history?)

Ajax

10 just sounds like a Safari bug.

[markjaquith]

Both these Safari issues are fixed in the latest Webkit nightly. It behaves like Chrome. I suspect that an upcoming update to Safari will push those fixes out.

[dd32]

Closing for now. I'm not sure what the 3.2 PushState work was, but the testing information in this ticket is now several browser-generations out of date.