Rely on 'memcpy' in WTF::copyChars

Source/wtf/text/StringImpl.h

Index: Source/wtf/text/StringImpl.h
diff --git a/Source/wtf/text/StringImpl.h b/Source/wtf/text/StringImpl.h
index 28a0095340766995c6054bb6c44b9dc22b1139ec..afc6d95aab8b704fb55085ac36e94c941e2ea468 100644
--- a/Source/wtf/text/StringImpl.h
+++ b/Source/wtf/text/StringImpl.h
@@ -300,25 +300,7 @@ public:
             return;
         }

[abarth-chromium, 2014/03/24 16:46:54]

Should we remove this branch too?

[Mikhail, 2014/03/25 10:30:26]

This makes a bit better assembly, on the other hand I do not think we can ever
feel it in the real world, so IMO branch can be removed indeed for the code
simplification, wdyt?

 
-        // FIXME: Is this implementation really faster than memcpy?
-        if (numCharacters <= s_copyCharsInlineCutOff) {
-            unsigned i = 0;
-#if (CPU(X86) || CPU(X86_64))
-            const unsigned charsPerInt = sizeof(uint32_t) / sizeof(T);
-
-            if (numCharacters > charsPerInt) {
-                unsigned stopCount = numCharacters & ~(charsPerInt - 1);
-
-                const uint32_t* srcCharacters = reinterpret_cast<const uint32_t*>(source);
-                uint32_t* destCharacters = reinterpret_cast<uint32_t*>(destination);
-                for (unsigned j = 0; i < stopCount; i += charsPerInt, ++j)
-                    destCharacters[j] = srcCharacters[j];
-            }
-#endif
-            for (; i < numCharacters; ++i)
-                destination[i] = source[i];
-        } else
-            memcpy(destination, source, numCharacters * sizeof(T));
+        memcpy(destination, source, numCharacters * sizeof(T));
     }
 
     ALWAYS_INLINE static void copyChars(UChar* destination, const LChar* source, unsigned numCharacters)
@@ -440,9 +422,6 @@ private:
         return sizeof(StringImpl) + length * sizeof(CharType);
     }
 
-    // This number must be at least 2 to avoid sharing empty, null as well as 1 character strings from SmallStrings.

[abarth-chromium, 2014/03/24 16:46:54]

I take it this comment is no longer accurate?  It doesn't make much sense to
me...

-    static const unsigned s_copyCharsInlineCutOff = 20;
-
     template <class UCharPredicate> PassRefPtr<StringImpl> stripMatchedCharacters(UCharPredicate);
     template <typename CharType, class UCharPredicate> PassRefPtr<StringImpl> simplifyMatchedCharactersToSpace(UCharPredicate, StripBehavior);
     NEVER_INLINE unsigned hashSlowCase() const;