做项目时,无意间发现了分页上的一个bug,在此记录一下:
首先手动将后台输出的sql语句复制进oracle中查看,以便排查错误,对比以下视图前10条的结果集与10到20条的结果集,发现大部分记录出现重复现象,SQL语句如下:
--前10条记录
select *
from (select row_.*, rownum rownum_
from (select t.idcard, count(1)
from sampling.v_unvoucher_blacklist t
where 1 = 1
group by t.idcard
order by count(1) asc) row_
where rownum <= 10)
where rownum_ > 0;
--第11条-第20条记录
select *
from (select row_.*, rownum rownum_
from (select t.idcard, count(1)
from sampling.v_unvoucher_blacklist t
where 1 = 1
group by t.idcard
order by count(1) asc) row_
where rownum <= 20)
where rownum_ > 10;
出现以上错误时,先是反复增加/删除where之后的条件,不断测试,貌似发现了一些端倪,错误记录都发生在count(1)的值相同时,于是在网上好一阵搜索,可还是没有找到解决办法;
没办法了,最后还是硬着头皮去找官网的一些demo,从中观察用法,探索者、摸索着,就有了以下的版本,经测试可以返回正确结果集,修改后如下:
select *
from (select row_1.*, rownum rownum_
from (select * from (select t.idcard idcard, count(1) total
from sampling.v_unvoucher_blacklist t
where 1 = 1
group by t.idcard)
order by total,rownum asc) row_1
where rownum <= 10)
where rownum_ > 0;
总结:order by语句应当遵循条件唯一性原则,否则oracle不保证每次查询的结果集都相同。
详细介绍请参考oracle官网:
http://www.oracle.com/technetwork/issue-archive/2006/06-sep/o56asktom-086197.html
转载于:https://blog.51cto.com/jiyanle/1640988