align fasta.PrecedenceInfo.precedence with analyzer.TokenType.precedence

pkg/front_end/lib/src/scanner/token.dart

Index: pkg/front_end/lib/src/scanner/token.dart
diff --git a/pkg/front_end/lib/src/scanner/token.dart b/pkg/front_end/lib/src/scanner/token.dart
index 2824fdcb2eb660bef8e07206022008b5bb427260..09dea64a32394ca9249a19e0be9626a757164d67 100644
--- a/pkg/front_end/lib/src/scanner/token.dart
+++ b/pkg/front_end/lib/src/scanner/token.dart
@@ -1009,7 +1009,7 @@ class TokenType {
       const TokenType._('MINUS_EQ', TokenClass.ASSIGNMENT_OPERATOR, '-=');
 
   static const TokenType MINUS_MINUS =
-      const TokenType._('MINUS_MINUS', TokenClass.UNARY_PREFIX_OPERATOR, '--');
+      const TokenType._('MINUS_MINUS', TokenClass.UNARY_POSTFIX_OPERATOR, '--');

[Brian Wilkerson, 2017/03/15 13:50:33]

I'm not sure what the right answer is here. Note that both '--' and '++' are
both prefix and postfix operators. I don't know whether using the smaller or
larger of the two is the right solution, nor am I convinced that it makes a
difference.

I'm confident that it doesn't matter for analyzer's parser, because it doesn't
use any precedence information for anything. I'd have to do a code search to see
where analyzer / server does use this information.

 
   static const TokenType OPEN_CURLY_BRACKET =
       const TokenType._('OPEN_CURLY_BRACKET', TokenClass.NO_CLASS, '{');
@@ -1039,7 +1039,7 @@ class TokenType {
       const TokenType._('PLUS_EQ', TokenClass.ASSIGNMENT_OPERATOR, '+=');
 
   static const TokenType PLUS_PLUS =
-      const TokenType._('PLUS_PLUS', TokenClass.UNARY_PREFIX_OPERATOR, '++');
+      const TokenType._('PLUS_PLUS', TokenClass.UNARY_POSTFIX_OPERATOR, '++');
 
   static const TokenType QUESTION =
       const TokenType._('QUESTION', TokenClass.CONDITIONAL_OPERATOR, '?');