Question

I’m having a problem when i refactoring the legacy join syntax to ANSI join syntax’s, because most of the procedures in our database is using the legacy join. this is the code I need to change it to use the SQL ‘JOIN’ syntax rather than the normal legacy join. can any one suggest me how it can be done?

    select 
          a.userkey, 
          a.username,
          c.monthd ,
          b.currencykey 
     from 
         #users a,
         #invoicedata b,
          #revdate c
     where 
         (a.userkey >= b.userkey or a.userkey <= b.userkey) 
          and b.idate between c.startdate and c.enddate
      group by 
           a.userkey,
           a.username,
           c.monthd,
           b.currencykey

       order by c.id,a.username,b.currencykey

Answer 1

This part (a.userkey >= b.userkey or a.userkey <= b.userkey) doesn’t seem correct and you can’t order by columns (at least without agregate functions) that aren’t in the group by.

So, possibly something like this:

select 
      a.userkey, 
      a.username,
      c.monthd ,
      b.currencykey 
 from #users a
 inner join #invoicedata b on a.userkey = b.userkey
 inner join #revdate c on b.idate between c.startdate and c.enddate
 group by 
      a.userkey,
      a.username,
      c.monthd,
      b.currencykey

I’ve also replaced nested joins with inner joins but that wasn’t necessary 🙂

---

On the other hand, you may want a cross join:

select 
      a.userkey, 
      a.username,
      c.monthd ,
      b.currencykey 
 from #users a
 cross join #invoicedata b
 inner join #revdate c on b.idate between c.startdate and c.enddate
 group by 
      a.userkey,
      a.username,
      c.monthd,
      b.currencykey