http://trac.webkit.org/changeset/55823 changed database callbacks in JSC, creating a new isolated world for each callback. This seems clearly wrong. A callback should just reuse the world it was created with. This: , m_isolatedWorld(DOMWrapperWorld::create(globalObject->globalData(), true)) should be this: , m_isolatedWorld(globalObject->world())
On it.
Created attachment 52899 [details] patch
Comment on attachment 52899 [details] patch Test? Or does this fix the open-database test that's failing?
No, it needs a new test.
Comment on attachment 52899 [details] patch You heard the man.
Created attachment 52901 [details] patch (In reply to comment #5) > (From update of attachment 52899 [details]) > You heard the man. Yessir!
Comment on attachment 52901 [details] patch r=me
Landed as r57301.
http://trac.webkit.org/changeset/57301 might have broken Tiger Intel Release The following changes are on the blame list: http://trac.webkit.org/changeset/57304 http://trac.webkit.org/changeset/57301 http://trac.webkit.org/changeset/57302 http://trac.webkit.org/changeset/57303
(In reply to comment #9) > http://trac.webkit.org/changeset/57301 might have broken Tiger Intel Release > The following changes are on the blame list: > http://trac.webkit.org/changeset/57304 > http://trac.webkit.org/changeset/57301 > http://trac.webkit.org/changeset/57302 > http://trac.webkit.org/changeset/57303 None of these changes are to blame. The bots just happened to like open-database-creation-callback-isolated-world.html in the previous run, but it failed again when these 4 patches were pulled in.
The bots have been failing for 12 hours, ever since bug 31482 went in. As far as I can tell, this depends on bug 31482, so I think I'll need to roll both out since Dumi is not around.