rollup 的汇总行和下面的总和不相等 是什么原因呢？？？

Question

select decode(grouping(l.nareated), 1, '日期合计', to_char(l.nareated)) nareated,
       count(distinct(case
                        when l.isindex = 0 then
                         l.userid
                        else
                         null
                      end)) viewpeople, -- 这个字段 汇总行和下面数据行总和不相等                       
       count(distinct(case
                        when l.ischarge = 1 and l.msid is not null then
                         l.msid || l.productid
                        else
                         null
                      end)) bookpeople, -- 这个字段 汇总行和下面数据行总和不相等  
       sum(case
             when l.isnewuser = 0 then
              1
             else
              0
           end) newusers,
       sum(case
             when l.isindex = 0 then
              1
             else
              0
           end) assces,
       count(case
               when l.ischarge = 1 and l.msid is not null then
                l.userid
               else
                null
             end) downs,
       sum(case
             when l.ischarge = 1 then
              l.price
             else
              0
           end) / 10 moneys
  from t_wp_accesslog l, t_declare_dproduct p
where l.productid = p.productid
   and l.msid is not null
   and l.nareated between to_number('20080217') and to_number('20080218')
group by rollup(l.nareated)
order by nareated desc
-----------------------------------------------------------------------------------
结果如下：
NAREATED        VIEWPEOPLE        BOOKPEOPLE        NEWUSERS        ASSCES        DOWNS        MONEYS
   日期合计                9288                       3563                   10732                 24057      9695    19082
20080217        4659                       1794                       5202                     11589        4514         8871.5
20080216        4989                       1880                       5530                     12468        5181        10210.5

Answer 1

调整下格式：
-----------------------------------------------------------------------
NAREATED    VIEWPEOPLE  BOOKPEOPLE   NEWUSERS   ASSCES  DOWNS   MONEYS
日期合计        9288      3563         10732    24057   9695     19082
20080217        4659      1794         5202     11589   4514     8871.5
20080216        4989      1880         5530     12468   5181     10210.5

Answer 2

我现在知道是distinct引起的问题，因为l.userid有重复，
但是这里我需要用distinct来区分用户，计算用户数
根据我的需求我需要怎么改才能得到我要的统计结果呢？