embind doesn't always need the full std::type_info record. if EMSCRIPTEN_HAS_UNBOUND_TYPE_NAMES=0, then use a lighter type identifier. This shaves 175 KB off of our engine's minified JavaScript.

Chad Austin · chadaustin · commit 2bbdb0cd396d · 2014-04-13T09:08:27.000-07:00
diff --git a/system/include/emscripten/bind.h b/system/include/emscripten/bind.h
@@ -813,7 +813,7 @@ namespace emscripten {
         // NOTE: this returns the class type, not the pointer type
         template<typename T>
         inline TYPEID getActualType(T* ptr) {
-            return reinterpret_cast<TYPEID>(&typeid(*ptr));
+            return getLightTypeID(*ptr);
         };
     }
 
diff --git a/system/include/emscripten/wire.h b/system/include/emscripten/wire.h
@@ -15,22 +15,66 @@
 #define EMSCRIPTEN_ALWAYS_INLINE __attribute__((always_inline))
 
 namespace emscripten {
+    #ifndef EMSCRIPTEN_HAS_UNBOUND_TYPE_NAMES
+    #define EMSCRIPTEN_HAS_UNBOUND_TYPE_NAMES 1
+    #endif
+
+
+    #if EMSCRIPTEN_HAS_UNBOUND_TYPE_NAMES
+    constexpr bool has_unbound_type_names = true;
+    #else
+    constexpr bool has_unbound_type_names = false;
+    #endif
+
     namespace internal {
         typedef void (*GenericFunction)();
 
-        typedef const struct _TYPEID* TYPEID;
+        typedef const struct _TYPEID {}* TYPEID;
+
+
+        // We don't need the full std::type_info implementation.  We
+        // just need a unique identifier per type and polymorphic type
+        // identification.
+        
+        template<typename T>
+        struct CanonicalizedID {
+            static TYPEID get() {
+                static _TYPEID c;
+                return &c;
+            }
+        };
+
+        template<typename T>
+        struct Canonicalized {
+            typedef typename std::remove_cv<typename std::remove_reference<T>::type>::type type;
+        };
+
+        template<typename T>
+        struct LightTypeID {
+            static TYPEID get() {
+                typedef typename Canonicalized<T>::type C;
+                if (has_unbound_type_names || std::is_polymorphic<C>::value) {
+                    return reinterpret_cast<TYPEID>(&typeid(C));
+                } else {
+                    return CanonicalizedID<C>::get();
+                }
+            }
+        };
+
+        template<typename T>
+        const TYPEID getLightTypeID(const T& value) {
+            typedef typename Canonicalized<T>::type C;
+            if (has_unbound_type_names || std::is_polymorphic<C>::value) {
+                return reinterpret_cast<TYPEID>(&typeid(value));
+            } else {
+                return LightTypeID<T>::get();
+            }
+        }
 
-        // This implementation is technically not legal, as it's not
-        // required that two calls to typeid produce the same exact
-        // std::type_info instance.  That said, it's likely to work
-        // given Emscripten compiles everything into one binary.
-        // Should it not work in the future: replace TypeID with an
-        // int, and store all TypeInfo we see in a map, allocating new
-        // TypeIDs as we add new items to the map.
         template<typename T>
         struct TypeID {
             static TYPEID get() {
-                return reinterpret_cast<TYPEID>(&typeid(T));
+                return LightTypeID<T>::get();
             }
         };
 
@@ -53,7 +97,7 @@ namespace emscripten {
         template<typename T>
         struct TypeID<AllowedRawPointer<T>> {
             static TYPEID get() {
-                return reinterpret_cast<TYPEID>(&typeid(T*));
+                return LightTypeID<T*>::get();
             }
         };
         
diff --git a/system/lib/embind/bind.cpp b/system/lib/embind/bind.cpp
@@ -14,26 +14,32 @@ using namespace emscripten;
 
 extern "C" {
     const char* __attribute__((used)) __getTypeName(const std::type_info* ti) {
+        if (has_unbound_type_names) {
 #ifdef USE_CXA_DEMANGLE
-        int stat;
-        char* demangled = abi::__cxa_demangle(ti->name(), NULL, NULL, &stat);
-        if (stat == 0 && demangled) {
-            return demangled;
-        }
+            int stat;
+            char* demangled = abi::__cxa_demangle(ti->name(), NULL, NULL, &stat);
+            if (stat == 0 && demangled) {
+                return demangled;
+            }
 
-        switch (stat) {
-            case -1:
-                return strdup("<allocation failure>");
-            case -2:
-                return strdup("<invalid C++ symbol>");
-            case -3:
-                return strdup("<invalid argument>");
-            default:
-                return strdup("<unknown error>");
-        }
+            switch (stat) {
+                case -1:
+                    return strdup("<allocation failure>");
+                case -2:
+                    return strdup("<invalid C++ symbol>");
+                case -3:
+                    return strdup("<invalid argument>");
+                default:
+                    return strdup("<unknown error>");
+            }
 #else
-        return strdup(ti->name());
+            return strdup(ti->name());
 #endif
+        } else {
+            char str[80];
+            sprintf(str, "%p", ti);
+            return strdup(str);
+        }
     }
 }
 
diff --git a/tests/embind/embind.test.js b/tests/embind/embind.test.js
@@ -1646,6 +1646,10 @@ module({
     if (typeof INVOKED_FROM_EMSCRIPTEN_TEST_RUNNER === "undefined") { // TODO: Enable this to work in Emscripten runner as well!
 
     BaseFixture.extend("unbound types", function() {
+        if (!cm.hasUnboundTypeNames) {
+            return;
+        }
+
         function assertMessage(fn, message) {
             var e = assert.throws(cm.UnboundTypeError, fn);
             assert.equal(message, e.message);
diff --git a/tests/embind/embind_test.cpp b/tests/embind/embind_test.cpp
@@ -2147,6 +2147,8 @@ struct BoundClass {
 };
 
 EMSCRIPTEN_BINDINGS(incomplete) {
+    constant("hasUnboundTypeNames", emscripten::has_unbound_type_names);
+
     function("getUnboundClass", &passThrough<UnboundClass>);
 
     class_<HasUnboundBase, base<UnboundClass>>("HasUnboundBase")

Original file line number	Diff line number	Diff line change
`@@ -813,7 +813,7 @@ namespace emscripten {`
`813`	`813`	`// NOTE: this returns the class type, not the pointer type`
`814`	`814`	`template<typename T>`
`815`	`815`	`inline TYPEID getActualType(T* ptr) {`
`816`		`- return reinterpret_cast<TYPEID>(&typeid(*ptr));`
	`816`	`+ return getLightTypeID(*ptr);`
`817`	`817`	`};`
`818`	`818`	`}`
`819`	`819`