Fix #451

Readme.md

@@ -360,7 +360,7 @@ using (var reader = connection.ExecuteReader("select * from Shapes"))
 
 Limitations and caveats
 ---------------------
-Dapper caches information about every query it runs, this allow it to materialize objects quickly and process parameters quickly. The current implementation caches this information in a `ConcurrentDictionary` object. The objects it stores are never flushed. If you are generating SQL strings on the fly without using parameters it is possible you will hit memory issues. We may convert the dictionaries to an LRU Cache.
+Dapper caches information about every query it runs, this allow it to materialize objects quickly and process parameters quickly. The current implementation caches this information in a `ConcurrentDictionary` object. Statements that are only used once are routinely flushed from this cache. Still, if you are generating SQL strings on the fly without using parameters it is possible you may hit memory issues.
 
 Dapper's simplicity means that many feature that ORMs ship with are stripped out. It worries about the 95% scenario, and gives you the tools you need most of the time. It doesn't attempt to solve every problem.