Always preload all tokens before parsing

Source/core/html/parser/HTMLDocumentParser.cpp

Index: Source/core/html/parser/HTMLDocumentParser.cpp
diff --git a/Source/core/html/parser/HTMLDocumentParser.cpp b/Source/core/html/parser/HTMLDocumentParser.cpp
index 20f5c4f44d18ec3e9105989a5f2161a94f952af2..02750630ae6a12df8d506b0d4dde1d2aea4f8bdc 100644
--- a/Source/core/html/parser/HTMLDocumentParser.cpp
+++ b/Source/core/html/parser/HTMLDocumentParser.cpp
@@ -353,7 +353,13 @@ void HTMLDocumentParser::didReceiveParsedChunkFromBackgroundParser(PassOwnPtr<Pa
     RefPtrWillBeRawPtr<HTMLDocumentParser> protect(this);
 
     ASSERT(m_speculations.isEmpty());
-    chunk->preloads.clear(); // We don't need to preload because we're going to parse immediately.
+    // This preload call is redundant, but we plan to change treebuilding to
+    // always operate asynchronously after a message-loop round trip after delivery
+    // to the main thread, thus we'll always use the preloader. Before making
+    // that change, we're sending all chunks through the preloader right
+    // before treebuilding (changes 15 layout tests).
+    // https://codereview.chromium.org/258013009/
+    m_preloader->takeAndPreload(chunk->preloads);
     m_speculations.append(chunk);

[tonyg, 2014/05/29 23:27:25]

These two lines are now identical to the other code path above. Is there a
cleaner way to structure this or is it so temporary that it doesn't matter.

     pumpPendingSpeculations();
 }