diff --git a/StgSharp.RegularAnalysis/Abstraction/ISyntaxNode.cs b/StgSharp.RegularAnalysis/Abstraction/ISyntaxNode.cs
index 0d816001..86ab7806 100644
--- a/StgSharp.RegularAnalysis/Abstraction/ISyntaxNode.cs
+++ b/StgSharp.RegularAnalysis/Abstraction/ISyntaxNode.cs
@@ -36,6 +36,15 @@ using System.Text;
 
 namespace StgSharp.RegularAnalysis.Abstraction
 {
+    public interface ITrinitySyntaxNode<TNode, TLabel> : ISyntaxNode<TNode, TLabel>
+        where TNode : ISyntaxNode<TNode, TLabel>
+        where TLabel : unmanaged
+    {
+
+        TNode Third { get; set; }
+
+    }
+
     public interface ISyntaxNode<TNode, TLabel> where TNode : ISyntaxNode<TNode, TLabel>
         where TLabel : unmanaged
     {
diff --git a/StgSharp.RegularAnalysis/Text/RegexInterpreter.IROptimize.cs b/StgSharp.RegularAnalysis/Text/RegexInterpreter.IROptimize.cs
index 6a5de231..2e517d2d 100644
--- a/StgSharp.RegularAnalysis/Text/RegexInterpreter.IROptimize.cs
+++ b/StgSharp.RegularAnalysis/Text/RegexInterpreter.IROptimize.cs
@@ -87,7 +87,7 @@ namespace StgSharp.RegularAnalysis.Text
                         break;
                     }
 
-                    if (TryFlatternCount(opt_win_sp, opt_ir)) { }
+                    if (TryFlattenCount(opt_win_sp, opt_ir)) { }
                 } while (false);
 
                 index += process_count;
@@ -150,7 +150,7 @@ namespace StgSharp.RegularAnalysis.Text
 
             #endregion
 
-        private static bool TryFlatternCount(
+        private static bool TryFlattenCount(
                             Span<RegexIR> window,
                             List<RegexIR> opt
         )
diff --git a/future/ngra-regex-ir.md b/future/ngra-regex-ir.md
index 6e1e8224..de81e805 100644
--- a/future/ngra-regex-ir.md
+++ b/future/ngra-regex-ir.md
@@ -66,3 +66,7 @@ COUNT_SEQ(a,0,+inf,greedy)
 5. a{m,n}特征，展开为
 MATCH_SEQ(aa...m个a....aa)
 COUNT_SEQ(a,0,n,greedy)
+
+6. .*?a*$
+
+7. 对于一般的.*?，应该将后续的所有东西全都放进一个巨大的try，code gen时使用最慢的一位一位offset直到直到第一个匹配rest的东西
\ No newline at end of file