因为NPSI是一个单序列列表，很容易看到DB 2如何来回切换，从表到索引、索引到表，然后以ZIPCODE次序返回行，而不需要排序。

Because the NPSI is a single ordered list, it is easy to see how DB2 could just toggle back and forth, index to table, index to table, and return rows in ZIPCODE order without sorting.

youdao

所以如果两次序列化一个单例对象，这两个序列化得到的对象会持有两个不同的标识。

So if you serialize one singleton twice, the two serialized objects take on separate identities.

youdao