Make multi-line comments not count when checking whether --> is first on a line.

src/scanner-base.cc

Index: src/scanner-base.cc
diff --git a/src/scanner-base.cc b/src/scanner-base.cc
index 89591baa243e003247e44ae4145b819590b8efce..b9f15239dc7f4a82d53c8df5a44806eafd20ffa7 100644
--- a/src/scanner-base.cc
+++ b/src/scanner-base.cc
@@ -80,6 +80,7 @@ JavaScriptScanner::JavaScriptScanner(UnicodeCache* scanner_contants)
 Token::Value JavaScriptScanner::Next() {
   current_ = next_;
   has_line_terminator_before_next_ = false;
+  next_is_first_on_line_ = false;

[William Hesse, 2011/06/21 11:37:41]

Could we call these things something more parallel, like:
has_line_terminator_before_next and
has_line_terminator_outside_comment_before_next
or
saw_line_terminator_outside_comment
saw_line_terminator_inside_comment?

The meanings and algorithm would be clearer.

Or better yet:
newline_before_current_token  and
multiline_comment_before_current_token.
The condition for handling --> would be newline_before...
and the condition for automatic semicolon insertion
would be newline_before... || multiline_comment_before...

   Scan();
   return current_.token;
 }
@@ -109,6 +110,7 @@ bool JavaScriptScanner::SkipWhiteSpace() {
         // Ignore line terminators, but remember them. This is necessary
         // for automatic semicolon insertion.
         has_line_terminator_before_next_ = true;
+        next_is_first_on_line_ = true;
       }
       Advance();
     }
@@ -117,7 +119,7 @@ bool JavaScriptScanner::SkipWhiteSpace() {
     // line (with only whitespace in front of it), we treat the rest
     // of the line as a comment. This is in line with the way
     // SpiderMonkey handles it.
-    if (c0_ == '-' && has_line_terminator_before_next_) {
+    if (c0_ == '-' && next_is_first_on_line_) {
       Advance();
       if (c0_ == '-') {
         Advance();
@@ -167,7 +169,9 @@ Token::Value JavaScriptScanner::SkipMultiLineComment() {
     }
     // If we have reached the end of the multi-line comment, we
     // consume the '/' and insert a whitespace. This way all
-    // multi-line comments are treated as whitespace.
+    // multi-line comments are treated as whitespace (except
+    // when checking whether there is non-whitespace before a
+    // --> comment).
     if (ch == '*' && c0_ == '/') {
       c0_ = ' ';
       return Token::WHITESPACE;
@@ -211,6 +215,7 @@ void JavaScriptScanner::Scan() {
       case '\n':
         Advance();
         has_line_terminator_before_next_ = true;
+        next_is_first_on_line_ = true;
         token = Token::WHITESPACE;
         break;
 
@@ -289,7 +294,7 @@ void JavaScriptScanner::Scan() {
         Advance();
         if (c0_ == '-') {
           Advance();
-          if (c0_ == '>' && has_line_terminator_before_next_) {
+          if (c0_ == '>' && next_is_first_on_line_) {
             // For compatibility with SpiderMonkey, we skip lines that
             // start with an HTML comment end '-->'.
             token = SkipSingleLineComment();