- SQL code
select * from t_standard_dtl d where trim(d.acc) = '111111' and trim(d.sub) = '6' and d.txn_type = '2' and d.confirm_flag = '0' and d.txn_date || d.txn_time = (select max(txn_date || txn_time) from t_standard_dtl where trim(acc) = '111111' and trim(sub) = '6' and d.txn_type = '2' and d.confirm_flag = '0')
------解决方案--------------------
有些看不明白,但说说我的想法。
1、max()是根据ascii进行逐个比较,所以其得到的值应该是唯一的;
2、外面的条件和里面的条件一样,
所以可否用order by 呢?
- SQL code
select * from (select * from t_standard_dtl d where trim(d.acc) = '111111' and trim(d.sub) = '6' and d.txn_type = '2' and d.confirm_flag = '0' order by max(txn_date || txn_time) desc) where rownum <2;
------解决方案--------------------
同楼上,是返回2条日期最大的两条记录吧
------解决方案--------------------
不要让别人接受你的思路,建议把原数据和结果数据贴出来!
------解决方案--------------------
- SQL code
select * from t_standard_dtl d where trim(d.acc) = '111111' and trim(d.sub) = '6' and d.txn_type = '2' and d.confirm_flag = '0' and d.txn_date || d.txn_time = (select max(txn_date || txn_time) from t_standard_dtl where trim(acc) = '111111' and trim(sub) = '6' and d.txn_type = '2' and d.confirm_flag = '0')-- 上面的SQL可以看出:既然父查询和子查询是查的同一个表,-- 那么你这样的话,需要析取两次:父查询析取一次、子查询析取一次!(如果你的表有1000万行记录,析取后,只有1000条记录)-- 我们可以用表别名,这样只需要对原表析取一次with a as ( select * from t_standard_dtl d where trim(d.acc) = '111111' and trim(d.sub) = '6' and d.txn_type = '2' and d.confirm_flag = '0' )select a1.*from a a1 where exists (select 1 from a a2 having max(a2.txn_date)=a1.txn_date and max(a2.txn_time)=a1.txn_date );