Requires patch in bug 574872.

Comment on attachment 454635 [details] [diff] [review] Patch, v1 For comments with context, please see http://reviews.visophyte.org/r/454635/. on file: dom/indexedDB/IDBEvents.cpp line 345 > NS_IMETHODIMP > IDBSuccessEvent::GetResult(JSContext* aCx, > jsval* aResult) > { > if (!mResult) { > *aResult = JSVAL_VOID; > return NS_OK; > } > > nsIXPConnect* xpc = nsContentUtils::XPConnect(); > NS_ENSURE_STATE(xpc); > > JSAutoRequest ar(aCx); > JSObject* scope = JS_GetGlobalObject(aCx); > NS_ENSURE_STATE(scope); > > nsresult rv = xpc->VariantToJS(aCx, scope, mResult, aResult); > NS_ENSURE_SUCCESS(rv, rv); > > return NS_OK; > } Hmmm, why are we still using a variant for this. Isn't that slower? on file: dom/indexedDB/nsIIDBCursor.idl line 45 > %{C++ > #include "jsapi.h" > %} Is this needed for the context? If so, could we not just forward declare it? r=sdwilsh

(In reply to comment #1) > Hmmm, why are we still using a variant for this. Isn't that slower? Removing nsIVariant is another task on my list, not attempted here. > Is this needed for the context? If so, could we not just forward declare it? And for the jsval. Since jsval is special (not a class, not passed by address) we can't just forward declare.