Changeset 28309 in webkit for trunk/JavaScriptCore/bindings/runtime_root.cpp

Timestamp:

Dec 1, 2007, 3:56:56 PM (18 years ago)

Author:

[email protected]

Message:

JavaScriptCore:

Reviewed by Beth Dakin.

Reversed the ownership relationship between Interpreter and JSGlobalObject.
Now, the JSGlobalObject owns the Interpreter, and top-level objects
that need the two to persist just protect the JSGlobalObject from GC.

Global object bootstrapping looks a little odd right now, but it will
make much more sense soon, after further rounds of refactoring.

• bindings/runtime_root.h: Made this class inherit from RefCounted, to avoid code duplication.

• kjs/collector.cpp: (KJS::Collector::collect): No need to give special GC treatment to Interpreters, since we mark their global objects, which mark them.

• kjs/interpreter.cpp: (KJS::Interpreter::mark): No need to mark our global object, since it marks us.
• kjs/interpreter.h: Don't inherit from RefCounted -- JSGlobalObject owns us directly.

• kjs/testkjs.cpp: Modified to follow the new rules. (createGlobalObject): (runWithScripts):

JavaScriptGlue:

Reviewed by Beth Dakin.

Modified to follow new JSGlobalObject/Interpreter ownership rules
in JavaScriptCore.

• JSRun.cpp: (JSRun::JSRun): (JSRun::GetInterpreter): (JSRun::Evaluate): (JSRun::CheckSyntax):
• JSRun.h:
• JSValueWrapper.cpp: (unprotectGlobalObject): (initializeGlobalObjectKey): (getThreadGlobalExecState):

WebCore:

Reviewed by Beth Dakin.

Modified WebCore to follow the new JSGlobalObject/Interpreter ownership
rules in JavaScriptCore.

• bindings/js/kjs_binding.cpp:
• bindings/js/kjs_binding.h: Removed stale, unused interpreterForGlobalObject().

• bindings/js/kjs_proxy.cpp: Changed to store a global object, rather than an interpreter. (WebCore::KJSProxy::finishedWithEvent): Need to NULL check m_globalObject here because we no longer unnecessarily instantiate it.

• bindings/js/kjs_window.cpp: (KJS::ScheduledAction::execute):
• bindings/js/kjs_window.h: Removed redundant and less efficient interpreter() function -- global objects have direct access to their interpreters now.

Changed these functions to pass around JSGlobalObjects instead of
Interpreters:

• page/Frame.cpp: (WebCore::Frame::bindingRootObject): (WebCore::Frame::createRootObject):
• page/Frame.h:
• page/mac/WebCoreFrameBridge.mm: (createRootObject):

File:

• trunk/JavaScriptCore/bindings/runtime_root.cpp (modified) (4 diffs)

trunk/JavaScriptCore/bindings/runtime_root.cpp

@@ -26 +26 @@
 #include "runtime_root.h"
 
+#include "JSGlobalObject.h"
 #include "object.h"
 #include "runtime.h"
 #include "runtime_object.h"
-
 #include <wtf/HashCountedSet.h>
 #include <wtf/HashSet.h>
@@ -196 +196 @@
 #endif
 
-PassRefPtr<RootObject> RootObject::create(const void* nativeHandle, PassRefPtr<Interpreter> interpreter)
-{
-    return new RootObject(nativeHandle, interpreter);
-}
-
-RootObject::RootObject(const void* nativeHandle, PassRefPtr<Interpreter> interpreter)
-    : m_refCount(0)
-    , m_isValid(true)
+PassRefPtr<RootObject> RootObject::create(const void* nativeHandle, JSGlobalObject* globalObject)
+{
+    return new RootObject(nativeHandle, globalObject);
+}
+
+RootObject::RootObject(const void* nativeHandle, JSGlobalObject* globalObject)
+    : m_isValid(true)
     , m_nativeHandle(nativeHandle)
-    , m_interpreter(interpreter)
-{
-    ASSERT(m_interpreter);
+    , m_globalObject(globalObject)
+{
+    ASSERT(globalObject);
     rootObjectSet()->add(this);
 }
@@ -233 +232 @@
 
     m_nativeHandle = 0;
-    m_interpreter = 0;
+    m_globalObject = 0;
 
     ProtectCountSet::iterator end = m_protectCountSet.end();
@@ -285 +284 @@
 { 
     ASSERT(m_isValid);
-    return m_interpreter.get(); 
+    return m_globalObject->interpreter();
 }