remove mismatched delete

src/utils/debugger/SkObjectParser.cpp

Index: src/utils/debugger/SkObjectParser.cpp
diff --git a/src/utils/debugger/SkObjectParser.cpp b/src/utils/debugger/SkObjectParser.cpp
index 504cd085c92fc38ad8924c24370ed4dd6879df95..85ac4d3839145343a6eb528b7a78406f86de613a 100644
--- a/src/utils/debugger/SkObjectParser.cpp
+++ b/src/utils/debugger/SkObjectParser.cpp
@@ -336,10 +336,10 @@ SkString* SkObjectParser::TextToString(const void* text, size_t byteLength,
         case SkPaint::kUTF16_TextEncoding: {
             decodedText->append("UTF-16: ");
             size_t sizeNeeded = SkUTF16_ToUTF8((uint16_t*)text, byteLength / 2, NULL);
-            char* utf8 = new char[sizeNeeded];
+            SkAutoTArray<char> autoArray(sizeNeeded);
+            char* utf8 = autoArray.get();

[bungeman-skia, 2013/08/22 20:03:09]

I like

SkAutoSTMalloc<0x100, char> utf8(sizeNeeded);

Using the array makes it seem like we're going to use it 'array like', but we're
really just getting a hunk of memory. Most text runs are pretty short, so they
should mostly fit on the stack. Also, this way there's an implicit conversion to
T*, so we don't need another name (autoArray).

             SkUTF16_ToUTF8((uint16_t*)text, byteLength / 2, utf8);
             decodedText->append(utf8, sizeNeeded);
-            delete utf8;
             break;
         }
         case SkPaint::kUTF32_TextEncoding: {