My patch in bug 668728 requires that when each test is finished, its location won't change. The test for bug 622088 breaks this invariant, by using history.replaceState. I have a simple patch to fix this.

Comment on attachment 559603 [details] [diff] [review] Patch (v1) Review of attachment 559603 [details] [diff] [review]: ----------------------------------------------------------------- Fine, but again, I'd really prefer the original test authors to r+ these.

Copy-pasted from bug 685788 comment 3: (In reply to Ehsan Akhgari [:ehsan] from comment #0) > My patch in bug 668728 requires that when each test is finished, its > location won't change. It's not a problem if the test fails (e.g. throws an exception) and leaves the location different than it was initially? ed. to add: The failure mode for many of these tests is to hang and not get to the SimpleTest.finish() call.

(In reply to Justin Lebar [:jlebar] from comment #3) > Copy-pasted from bug 685788 comment 3: > > (In reply to Ehsan Akhgari [:ehsan] from comment #0) > > My patch in bug 668728 requires that when each test is finished, its > > location won't change. > > It's not a problem if the test fails (e.g. throws an exception) and leaves > the location different than it was initially? > > ed. to add: The failure mode for many of these tests is to hang and not get > to the SimpleTest.finish() call. That should be fine, as my patch in bug 668728 would catch that as an explicit failure. So we'll have one timeout, and one explicit failure. Which is even better than only a timeout. ;-)

Comment on attachment 559603 [details] [diff] [review] Patch (v1) Cool; sounds like a plan!