WebKit does this correctly but there is no test for this.
Created attachment 38622 [details] Proposed fix.
Comment on attachment 38622 [details] Proposed fix. I'd like to commit this myself so I can do the changes needed for the Chromium LayoutTestController before committing.
Comment on attachment 38622 [details] Proposed fix. The WK-cross-origin cookie is already set by existing tests - what ensures that the tests won't conflict? Does anything reset the cookie accept policy back after the test finishes?
(In reply to comment #3) > (From update of attachment 38622 [details]) > The WK-cross-origin cookie is already set by existing tests - what ensures that > the tests won't conflict? It works due to the different value but having a different name would be best. > Does anything reset the cookie accept policy back after the test finishes? Right now that is done in resetDefaultsToConsistentValues() which is where a lot of these things seem to get set back to default values. (I had done this in the destructor at one point.)
Comment on attachment 38622 [details] Proposed fix. new patch in progress.
Created attachment 38630 [details] Proposed fix.
Comment on attachment 38630 [details] Proposed fix. r=me
Committed as http://trac.webkit.org/changeset/47792