permit using pipelining option at query level

pipelining documentation

documentation/pipelining.md

@@ -0,0 +1,45 @@
+## Option "pipelining"
+
+With traditional synchronous drivers, queries are send one by one, waiting for result before sending next one.
+Communication with server following a request–response messaging pattern.
+
+That is not very effective when having to request many query at the same time.
+Node.js is efficient because of asynchrone concept, let's follow this rules.
+
+Pipelining is "optimistic send": Connector will send queries one after another, preserving FIFO order. 
+This is particularly efficient when client is distant from server.
+
+#### Example :
+create a basket with 3 items
+
+```javascript
+connection.beginTransaction();
+connection.query("INSERT INTO BASKET(customerId) values (?)", [1], (err, res) => {
+  //must handle error if any
+  const basketId = res.insertId;
+  try {
+    connection.query("INSERT INTO BASKET_ITEM(basketId, itemId) values (?, ?)", [basketId, 100]);
+    connection.query("INSERT INTO BASKET_ITEM(basketId, itemId) values (?, ?)", [basketId, 101]);
+    connection.query("INSERT INTO BASKET_ITEM(basketId, itemId) values (?, ?)", [basketId, 102], (err) => {
+      //must handle error if any
+      connection.commit();
+    });
+  } catch (err) {
+    connection.rollback();
+    //handle error
+  }
+});
+```
+#### Network exchanges :
+<p align="center">
+    <img src="./misc/pipelining.png">
+</p>
+
+
+Using standard client-server protocol (aka "ping-pong"), driver communicate with database following a request–response messaging pattern.
+When sending a command, connector won't send any until response is available from input socket.
+
+Using option "pipelining", commands are send by bulk, saving network latency.
+The Inconvenient is that if an error occur on first/second command, following command are already send to database.
+In that sense that pipelining is "optimistic". 
+

documentation/readme.md

@@ -103,7 +103,7 @@ default: false
 * `timezone`: string. force using indicated timezone, not current node.js timezone. possible value are 'Z' (fot UTC), 'local' or '±HH:MM' format    
 * `nestTables`: boolean/string. resultset are presented by table to avoid results with colliding fields. default: false 
 * `rowsAsArray`: boolean. default rows are defined as a JSON object. when active row is an array. default false 
-* `pipelining`: boolean. will send query one by one, but without waiting the results of previous entry. default true
+* `pipelining`: boolean. will send query one by one, but without waiting the results of previous entry ([detail information](/documentation/pipelining.md)). default true
  
 ## Query
 `connection.query(sql[, values][,callback])`