Make String::Empty inlineable.

src/api.cc

Index: src/api.cc
diff --git a/src/api.cc b/src/api.cc
index 8cdd3912390e302807f78e4901efc46be9794582..0bc93c2ff252635e7d5fe4dafcb2b8b06f63eb7d 100644
--- a/src/api.cc
+++ b/src/api.cc
@@ -4630,7 +4630,9 @@ void* External::Value() const {
 
 Local<String> v8::String::Empty() {
   i::Isolate* isolate = i::Isolate::Current();
-  EnsureInitializedForIsolate(isolate, "v8::String::Empty()");
+  if (!EnsureInitializedForIsolate(isolate, "v8::String::Empty()")) {

[Sven Panne, 2012/04/25 06:56:22]

Just out of curiosity: Why do we need this change? Or asked another way: Why do
we *not* need it in tons of other places? Or should we test the result of
EnusreInitializedForIsolate everywhere? api.cc is extremely inconsistent...

[Michael Starzinger, 2012/04/25 08:18:31]

Yes, IMHO we need this change in tons of other places. Because
EnsureInitializedForIsolate() can return even though the Isolate is not
initialized or is already dead, in cases where the embedder installed a fatal
error handler with SetFatalErrorHandler() that doesn't terminate the
application.

This just wasn't tested for String::Empty() before. I'll try to fix this
incrementally with my changes. We could also fix this for all functions in one
sweep, I you think that's worth the trouble.

+    return v8::Local<String>();
+  }
   LOG_API(isolate, "String::Empty()");
   return Utils::ToLocal(isolate->factory()->empty_symbol());
 }