作了一个测试,就像我们以前一直提醒的一样。在使用非 number datatype 的 数字类型一定要注意。
create table t_float (f float(2),n number(2))
SQL> insert into t_float values (7,7);
1 row created.
SQL> insert into t_float values (11,11);
1 row created.
SQL> insert into t_float values (99,99)
1 row created.
1 row created.
SQL> select * from t_float;
F N
---------- ----------
7 7
10 11
100 99
---------- ----------
7 7
10 11
100 99
Why:
a FLOAT column with a binary precision of 2 bits, is implemented as a NUMBER column with 1 decimal digit of precision and no fixed scale. Thus the number 7, which requires 3 binary bits, will nevertheless be stored exactly, whereas the number 11, which has two decimal digits, will be rounded to 10 because only one decimal digit of precision is allowed.
It is legitimate for a database implementation to use higher precision than requested in this way. Therefore, database applications should always round data values explicitly when required, and should not rely on the precision of the datatype to round data implicitly.
暂时还有理解上面的含义 :(
来自 “ ITPUB博客 ” ,链接:http://blog.itpub.net/34596/viewspace-805109/,如需转载,请注明出处,否则将追究法律责任。
转载于:http://blog.itpub.net/34596/viewspace-805109/