Question

This is a pretty complex query so i am having difficulties here. If you help me i appreciate it.

select AvgLevel,TotalCount,tblPokedex.PokemonId,Name,Type1,Type2,Hp,tblPokedex.Attack,tblPokedex.Defense,tblPokedex.SpAttack,tblPokedex.SpDefense,tblPokedex.Speed,(Hp+tblPokedex.Attack+tblPokedex.Defense+tblPokedex.SpAttack+tblPokedex.SpDefense+tblPokedex.Speed) as TotalStats 
from tblPokemonStats,tblAvailablePokemons,tblPokedex 
left join tblUsersPokemons on tblPokedex.PokemonId=tblUsersPokemons.PokemonId 
where tblPokemonStats.PokemonId=tblPokedex.PokemonId 
group by tblPokedex.PokemonId,tblPokedex.Name,tblPokedex.Type1,tblPokedex.Type2,tblPokedex.Hp,tblPokedex.Attack,tblPokedex.Defense,tblPokedex.SpAttack,tblPokedex.SpDefense,tblPokedex.Speed,tblPokemonStats.AvgLevel,tblPokemonStats.TotalCount  
order by PokemonId asc

Alright so what i want to do is for example select between top 100 and 150

How can i do that ?

Answer 1

Here’s one way – tack on a row_number column and filter on that:

select AvgLevel /*...*/
from (
    select AvgLevel /*...*/, row_number() over (order by PokemonId) rownum 
    from tblPokemonStats,tblAvailablePokemons,tblPokedex 
    left join tblUsersPokemons on tblPokedex.PokemonId=tblUsersPokemons.PokemonId 
    where tblPokemonStats.PokemonId=tblPokedex.PokemonId 
    group by /* etc ... */
) A
where A.rownum > 100 and A.rownum <= 150

row_number does what it sounds like – it numbers your rows.

Here’s another possibility – say you want rows X through Y. You can select the TOP Y rows according to your order. Then use that a subquery and select the TOP Y-X according to the opposite order.
Assuming X = 100 and Y = 150:

select top 50 *
from (
    select top 150 PokemonId
    from tblPokemonStats
    order by PokemonId
) A
order by PokemonId desc

As far as performance, I’m not sure which would be better. My gut says that the ORDER BY approach is better for pages near the ‘front’ so to speak, but row_number() would be better for later pages (high X and Y). That’s just a guess though – run some tests and see what works for you.