Left, right, inner and outer joins should behave correctly. That said, we generally aren't trying to join ALL data into a single mammoth table. Instead, we're doing more implied joins by simply naming fields the same.
You could think of it as a relational database system, where one table has the key to another table. So I might have an order item with a "customer" field on it. I also have a customer table with "customer" as the key.
In SQL, I might grab related data from both tables by doing this:
SELECT O.OrderItem, O.OrderQuantity, C.Customer, C.CustomerName
FROM Orders O
, Customers C
WHERE C.CUSTOMER = O.CUSTOMER
In QlikView, we CAN do that:
[Order Items and Customers]:
LOAD *
;
SQL
SELECT O.OrderItem, O.OrderQuantity, C.Customer, C.CustomerName
FROM Orders O
, Customers C
WHERE C.Customer = O.Customer
;
But I more typically load tables with separate SQL, and let QlikView handle the connection on Customer. Our QlikView table layout then mimics our relational database table layout. That isn't required to make QlikView work, but may be convenient for helping you see the connections between it and your source data.
[Order Items]:
LOAD *
;
SQL
SELECT OrderItem, OrderQuantity, Customer
FROM Orders
;
[Customers]:
LOAD *
;
SQL
SELECT Customer, CustomerName
FROM Customers
;
Or even if I want them to be one table in QlikView, I'll typically do the join inside of QlikView instead of SQL, partially because I usually use QVDs that are often (though not always) 1 to 1 with our source tables.
[Order Items and Customers]:
LOAD *
;
SQL
SELECT OrderItem, OrderQuantity, Customer
FROM Orders
;
INNER JOIN ([Order Items and Customers])
LOAD *
WHERE EXISTS(Customer)
;
SQL
SELECT Customer, CustomerName
FROM Customers
;
