On-Merge With Named Windows
功能：对window中的insert、update、delete操作进行组合运用。
格式：
| 1 | "; |
| 14 | |
| 15 | System.out.println("start Name Window select"); |
| 16 | EPOnDemandQueryResult sResult=epRuntime.executeQuery(select); |
| 17 | EventBean[] arryBeans =sResult.getArray(); |
| 18 | for (EventBean bean:arryBeans){ |
| 19 | System.out.println("select:"+bean.getUnderlying()); |
| 20 | } |
| 21 | |
| 22 | System.out.println("start Name Window update"); |
| 23 | EPOnDemandQueryResult uResult=epRuntime.executeQuery(update); |
| 24 | arryBeans =uResult.getArray(); |
| 25 | for (EventBean bean:arryBeans){ |
| 26 | System.out.println("select:"+bean.getUnderlying()); |
| 27 | } |
| 28 | |
| 29 | System.out.println("start Name Window delete"); |
| 30 | EPOnDemandQueryResult dResult=epRuntime.executeQuery(delete); |
| 31 | arryBeans =dResult.getArray(); |
| 32 | for (EventBean bean:arryBeans){ |
| 33 | System.out.println("select:"+bean.getUnderlying()); |
| 34 | } |
Explicitly Indexing Named Windows
功能：对Named Window中存放的事件属性建立索引。
格式：
| 1 | create [unique] index index_name on window_or_table_name (property [hash|btree] |
| 2 | [, property] [hash|btree] [,...] ) |
格式说明：
1、Unique:表示建立唯一索引，如果插入了重复的行，则会抛出异常并阻止重复行插入。如果不使用此关键字，则表示可以插入重复行；
2、Index_name:索引名称；
3、后面的括号中包含named window中的属性以及索引类型。索引类型分两种，hash索引不会排序，如果有=操作，建议使用此类型索引。btree索引基于排序二叉树，适合<, >, >=, <=, between, in等操作。如果不显式声明hash或者btree，则默认为hash索引;
Dropping or Removing Named Windows
功能：
注销named window，方式是直接调用EPStatement对象的destroy方法。虽然注销，但是named window的名字仍然被占用着，所以你只能重新建立和之前的named window一样结构的window，否则会抛出异常
示例子：
| 1 | // Create DropWindow |
| 2 | create window DropWindow.win:keepall() as select * from DropEvent |
| 3 | |
| 4 | // Destroy DropWindow |
| 5 | EPStatement state = admin.createEPL("create window DropWindow.win:keepall() as select * from DropEvent"); |
| 6 | state.destroy(); |
| 7 | |
| 8 | // Create DropEvent again(different with prior epl) |
| 9 | create window DropWindow.win:keepall() as select name from DropEvent |
| 10 | // throw Exception |