/**==** 2. 随机排序中的newid()问题 **==**/–测试数据declare @t table(name char(1),value int)insert into @tselect a,1union all select a,2union all select b,3union all select b,4union all select c,5union all select c,6
–要求,随机排序,但name相同的要排在一齐select a.*from @t a join( select aa=newid(),name from(select distinct name from @t) a) b on a.name=b.nameorder by b.aa,newid()
/*–测试结果之一,并没有达到要求.如果将join改成left right full join,就正确name value —- ———– b 4b 3c 5a 1a 2c 6
(所影响的行数为 6 行)–*/
