This is a follow-up for bug 949325. We need to fix the test of: is(cursor.store, gStore, "Cursor.store is the store"); in the file_sync_worker.js. The tricky part is cursor.store will return a newly created WorkerDataStore shell which proxie to the real DataStore. However, we shouldn't create a new WorkerDataStore to return. Instead, we should return the WorkerDataStore which owns the WorkerDataStoreCursor. I already had a patch for this. Will upload it tomorrow.

Forgot to clean up the CC list when cloning. Sorry for the noise.

Comment on attachment 8418492 [details] [diff] [review] Part 1, minor code clean-up Review of attachment 8418492 [details] [diff] [review]: ----------------------------------------------------------------- ::: dom/workers/DataStore.cpp @@ +360,5 @@ > } > }; > > +// A DataStoreRunnable to run DataStore::Clear(...) on the main thread. > +class DataStoreClearRunnable MOZ_FINAL : public DataStoreRunnable I just move DataStoreClearRunnable to a better place which collects all the runnables, so that the real API implementations can be listed consecutively and to be viewed easily.

Comment on attachment 8418492 [details] [diff] [review] Part 1, minor code clean-up Review of attachment 8418492 [details] [diff] [review]: ----------------------------------------------------------------- ::: dom/workers/DataStore.cpp @@ +36,5 @@ > { > JSContext* cx = aGlobal.GetContext(); > WorkerPrivate* workerPrivate = GetWorkerPrivateFromContext(cx); > MOZ_ASSERT(workerPrivate); > workerPrivate->AssertIsOnWorkerThread(); Why all of this if you want to call NS_NOTREACHED immediately after?

(In reply to Andrea Marchesini (:baku) from comment #5) > Comment on attachment 8418492 [details] [diff] [review] > Part 1, minor code clean-up > > Review of attachment 8418492 [details] [diff] [review]: > ----------------------------------------------------------------- > > ::: dom/workers/DataStore.cpp > @@ +36,5 @@ > > { > > JSContext* cx = aGlobal.GetContext(); > > WorkerPrivate* workerPrivate = GetWorkerPrivateFromContext(cx); > > MOZ_ASSERT(workerPrivate); > > workerPrivate->AssertIsOnWorkerThread(); > > Why all of this if you want to call NS_NOTREACHED immediately after? Yeap, we can remove these.

Sorry, wrong patch. This one is correct.

Try: https://tbpl.mozilla.org/?tree=Try&rev=98fff4e6c5c0

https://hg.mozilla.org/integration/mozilla-inbound/rev/65ba102d73b2 https://hg.mozilla.org/integration/mozilla-inbound/rev/729515d1cfff

https://hg.mozilla.org/mozilla-central/rev/65ba102d73b2 https://hg.mozilla.org/mozilla-central/rev/729515d1cfff