IndexedDB should respect SchemeRegistry's database access setting

[Martin Hock]

As with WebSQL, IndexedDB should respect SchemeRegistry's database access setting in private browsing mode.

[Martin Hock]

Created attachment 236537 [details]
patch

[Martin Hock]

One of the mac-wk2 runs flaked on storage/indexeddb/mozilla/cursor-update-updates-indexes.html but mac-wk2 is running ML and this test is somewhat flaky there:
http://webkit-test-results.appspot.com/dashboards/flakiness_dashboard.html#tests=storage%2Findexeddb%2Fmozilla%2Fcursor-update-updates-indexes.html

I'll run it a few more times locally.

[Alexey Proskuryakov]

Comment on attachment 236537 [details]
patch

Looks fine. I don't like the function name (isContextValid), but it wasn't particularly appropriate before this change either.

Please add a regression test ASAP (could be in a separate bug if this is urgent to land).

[Martin Hock]

Created attachment 236604 [details]
added test

[Alexey Proskuryakov]

Comment on attachment 236604 [details]
added test

View in context: https://bugs.webkit.org/attachment.cgi?id=236604&action=review

> LayoutTests/storage/indexeddb/private-browsing-affects-indexeddb.html:7
> +if (window.testRunner) {
> +    testRunner.dumpAsText();
> +}

You could have used js-test harness to get this and logging for free.

> LayoutTests/storage/indexeddb/private-browsing-affects-indexeddb.html:23
> +    if (!indexedDB.open('test1')) {
> +        log("Could not open test IndexedDB, exiting test.");
> +        return;
> +    }

I'm not sure if this is useful. We have all the other tests to verify that IndexedDB works, and dynamic changes to private browsing mode are not really meaningful.

[Martin Hock]

Created attachment 236616 [details]
cleaned up test

[Martin Hock]

Committed r172603: <http://trac.webkit.org/changeset/172603>