Add strict mode, sloppy mode and strong mode UseCounters

src/parser.h

Index: src/parser.h
diff --git a/src/parser.h b/src/parser.h
index 35a6f3af20bdc7022e7ff5ac065b132533afe4c6..b75059fe6f1cf3265a6fcdc3d30ea23c43820f0f 100644
--- a/src/parser.h
+++ b/src/parser.h
@@ -1200,6 +1200,9 @@ class Parser : public ParserBase<ParserTraits> {
   Expression* SpreadCallNew(Expression* function,
                             ZoneList<v8::internal::Expression*>* args, int pos);
 
+  void SetLanguageMode(Scope* scope, LanguageMode mode);
+  void EnableLanguageMode(LanguageMode mode);

[adamk, 2015/11/03 22:23:15]

These names don't help me understand the difference.

As noted in my comments in parser.cc, I was sort of expecting the use counts to
just be inlined into the prologue parsing, rather than causing a parser
refactor.

[Dan Ehrenberg, 2015/11/03 22:38:34]

I'd like to measure how many webpages are pure strict-mode. I don't see how I
could do that with just prologue parsing.

EnableLanguageMode is a separate cleanup refactoring. I can break it out into a
separate patch if you'd prefer, or if you think it makes things worse rather
than better, I'll just get rid of it. Do you have naming suggestions?

[adamk, 2015/11/03 23:04:17]

I don't follow: how can a page be "pure strict" without a "'use strict'"
prologue?
It doesn't improve readability for me, anyway; the existing code seemed clear,
so I don't really have naming suggestions.

+
   Scanner scanner_;
   PreParser* reusable_preparser_;
   Scope* original_scope_;  // for ES5 function declarations in sloppy eval