LEFT JOIN vs Stacked Left Join - sql

I wanted to ask whats the difference between those two queries:
SELECT
Customers.CustomerID, Customers.CustomerName, Orders.OrderID,
OrderDetails.Quantity, Products.ProductName
FROM
Customers
LEFT JOIN
(Orders
LEFT JOIN
(OrderDetails
LEFT JOIN
Products ON Products.ProductID = OrderDetails.ProductID
) ON OrderDetails.OrderID = Orders.OrderID
) ON Customers.CustomerID = Orders.CustomerID
GROUP BY
Customers.CustomerName;
Vs
SELECT
Customers.CustomerID, Customers.CustomerName, Orders.OrderID,
OrderDetails.Quantity, Products.ProductName
FROM
Customers
LEFT JOIN
Orders ON Orders.CustomerID = Customers.CustomerID
LEFT JOIN
OrderDetails ON OrderDetails.OrderID = Orders.OrderID
LEFT JOIN
Products ON Products.ProductID = OrderDetails.ProductID
GROUP BY
Customers.CustomerName;
Tested here
https://www.w3schools.com/sql/trysql.asp?filename=trysql_select_join
From what I can see one selects the first of multiple entries, one selects the last of multiple entries, but is that all?
From my point of view the not nested LEFT Join is way easier to read and to understand. Is there any downside of using it?

Your problem is the incorrect use of GROUP BY. The only unaggregated columns in the SELECT should be in the GROUP BY.
The rest of this answer addresses the point about joins.
Your second query is interpreted as:
FROM (((Customers c LEFT JOIN
Orders o
ON o.CustomerID = c.CustomerID
) LEFT JOIN
OrderDetails
ON od.OrderID = o.OrderID
) LEFT JOIN
Products p
ON p.ProductID = od.ProductID
The parentheses can affect the interpretation. But what effect? Essentially, you have:
(((c left join o) left join od) left join p)
versus
c left join (o left join (od left join p)))
Both keep all records in c, regardless of matches in the second. In this case, the two versions do the same thing. But for a particular reason -- the on conditions are strictly chained (that is, c to o, o to od, od to p). If p where joined to o instead of od, then subtle differences can occur.
What are the subtle differences? Two things can differ:
Whether columns from a particular table are NULL or have values.
Whether rows get duplicated, due to multiple matches between two tables.
In practice, I don't fine parentheses particularly useful. If I can about JOIN order, I use an explicit subquery or CTE>

Related

SQL - How can you use WHERE instead of LEFT/RIGHT JOIN?

since I am a bit rusty, I was practicing SQL on this link and was trying to replace the LEFT JOIN completly with WHERE. How can i do this so it does the same thing as the premade function in the website?
What I tried so far is:
SELECT Customers.CustomerName, Orders.OrderID
FROM Customers, Orders
WHERE Customers.CustomerID = Orders.CustomerID OR Customers.CustomerID != Orders.CustomerID
Order by Customers.CustomerName;
Thanks in advance for your help.
You are trying to replace
SELECT Customers.CustomerName, Orders.OrderID
FROM Customers
LEFT JOIN Orders ON Customers.CustomerID = Orders.CustomerID
with
SELECT Customers.CustomerName, Orders.OrderID
FROM Customers, Orders
WHERE ???
this is doomed to failure. Consider Customers has two rows and Orders has zero. The outer join will return two rows.
The cross join (FROM Customers, Orders) will return zero rows.
In standard SQL a WHERE clause can only reduce the rows from that - not increase them so there is nothing you can put for ??? that will give your desired results.
Before ANSI-92 joins were introduced some systems used to have proprietary operators for this, such as *= in SQL Server but this was removed from the product.
This may work for you.
SELECT
c.CustomerName,
o.OrderID
FROM Customers c
LEFT JOIN Orders o
on c.CustomerID = o.CustomerID
Order by c.CustomerName;
If you are trying to replace this:
SELECT c.CustomerName, o.OrderID
FROM Customers c LEFT JOIN
Orders o
ON c.CustomerID = o.CustomerID
ORDER BY c.CustomerName;
Then you can use UNION ALL:
SELECT c.CustomerName, o.OrderID
FROM Customers c JOIN
Orders o
ON c.CustomerID = o.CustomerID
UNION ALL
SELECT c.CustomerName, o.OrderID
FROM Customers c
WHERE NOT EXIST (SELECT 1 FROM Orders o WHERE c.CustomerID = o.CustomerID)
ORDER BY CustomerName
However, the LEFT JOIN is really a much better way to go.

Which direction need for INNER JOIN?

If I check this example: https://www.w3schools.com/sql/sql_join_inner.asp
This make sense:
SELECT Orders.OrderID, Customers.CustomerName
FROM Orders
INNER JOIN Customers ON Orders.CustomerID = Customers.CustomerID;
but this would work? I do not think so, right?
SELECT Orders.OrderID, Customers.CustomerName
FROM Customers
INNER JOIN Orders ON Orders.CustomerID = Customers.CustomerID;
I should always do child INNER JOIN parent and never parent INNER JOIN child, right?
You can do the inner join in either order. The only difference for an inner join is aesthetic. The ordering doesn't event affect performance (because the database decides on the best execution path).
I do recommend table aliases either way:
SELECT o.OrderID, c.CustomerName
FROM Customers c INNER JOIN
Orders o
ON o.CustomerID = c.CustomerID;

How to Organize Multiple Joins SQL

In SQL, how should I be joining tables together when I do multiple joins in one query. Should I join on only one table - in this case the Customers table or is it okay to do what I have done (joining on different tables as new keys are needed)?
SELECT O.OrderID, O.OrderDate, C.City, C.Country, C.PostalCode, C.ContactName, O.CustomerID, O.ShipperID, D.ProductID, COUNT(D.ProductID) ProductCount, S.SupplierID
FROM Customers C
INNER JOIN Orders O
ON O.CustomerID = C.CustomerID
INNER JOIN OrderDetails D
ON O.OrderID = D.OrderID
INNER JOIN Products P
ON D.ProductID = P.ProductID
INNER JOIN Suppliers S
ON S.SupplierID = P.SupplierID
WHERE 1 = 1
GROUP BY O.OrderID
ORDER BY OrderDate DESC
I am using W3Schools SQL TryIt editor to test this, not sure what DB engine it is!
Thanks!
Of course you can join on multiple tables in a query. That is a big part of the power of SQL.
In your particular case, you don't need the join to the Suppliers table, because the column is already in Products.
Also, you need to be careful about your SELECT and GROUP BY clauses. In general, you should put all non-aggregated columns in the GROUP BY:
SELECT O.OrderID, O.OrderDate, C.City, C.Country, C.PostalCode, C.ContactName,
O.CustomerID, O.ShipperID, D.ProductID,
COUNT(D.ProductID) as ProductCount,
P.SupplierID
FROM Customers C INNER JOIN
Orders O
ON O.CustomerID = C.CustomerID INNER JOIN
OrderDetails D
ON O.OrderID = D.OrderID INNER JOIN
Products P
ON D.ProductID = P.ProductID
GROUP BY O.OrderID, O.OrderDate, C.City, C.Country, C.PostalCode, C.ContactName,
O.CustomerID, O.ShipperID, D.ProductID, P.SupplierId
ORDER BY OrderDate DESC;
The WHERE 1=1 is also unnecessary.
I wonder if this query really does what you want. However, you don't state what you actually want the query to do, so I'm merely speculating.
The way you have done it is find, don't forget that for each inner join, your record set may reduce by the number of non matching keys in each additional join.
you could also just use the JOIN syntax.

Query returns cartesian product when not expected

Task: Select all orders having products belonging to ‘Sea Food’ category.
Result: OrderNo, OrderDate, Product Name
I write this query but it returns Cartesian products.
select o.orderid, o.orderdate as "Order Date", p.productname , ct.categoryname from orders o,
order_details od , products p , customers c ,categories ct
where
od.orderid = o.orderid and p.productid = od.productid and ct.categoryid = p.categoryid
and ct.categoryname = 'Seafood';
Question: What is wrong with my query ?
You're doing a CROSS JOIN on customers table since you forgot to specify the connection. This is why you should use explicit JOIN syntax rather than old syntax using commas in WHERE clause.
After translating your query into explicit syntax, you will see that there is no WHERE condition involving customers table:
select
o.orderid,
o.orderdate as "Order Date",
p.productname,
ct.categoryname
from
orders o,
inner join order_details od on od.orderid = o.orderid
inner join products p on p.productid = od.productid
inner join categories ct on ct.categoryid = p.categoryid
cross join customers c -- either you don't need this table, or you need to specify conditions
where
ct.categoryname = 'Seafood'
Basically the reason you got it was that your where clause omitted join condition involving customers table, so you were left with:
from (...), customers -- cross join when joining condition not applied in where clause

SQL Query on SQL Server 2008

I'm trying to get only customers that ordered both a "Gas Range" and a "Washer". I'm getting Customers who ordered a "Gas Range" and not a "Washer" and customers with both. I need the customer that meets both conditions. I'm close but a little stuck. Below is the query that I have so far. Please let me know if you need more information.
My Tables - CUSTOMER(CUST_NUM, CUST_NAME), ORDER_LINE(ORDER_NUM, PART_NUM), ORDERS(ORDER_NUM, CUST_NUM), PART(PART_NUM, PART_DESCRIPTION)
SELECT C.CUST_NAME AS [Customer(s) that ordered a Gas Range and Washer]
FROM CUSTOMER C
INNER JOIN ORDERS O
ON C.CUST_NUM = O.CUST_NUM
INNER JOIN ORDER_LINE OL
ON O.ORDER_NUM = OL.ORDER_NUM
INNER JOIN PART P
ON OL.PART_NUM = P.PART_NUM
WHERE P.PART_DESCRIPTION IN ('GasRange','Washer')
GROUP BY C.CUST_NAME
try the following
SELECT C.CUST_NAME AS [Customer(s) that ordered a Gas Range and Washer]
FROM CUSTOMER C
INNER JOIN ORDERS O
ON C.CUST_NUM = O.CUST_NUM
INNER JOIN ORDER_LINE OL
ON O.ORDER_NUM = OL.ORDER_NUM
INNER JOIN PART P
ON OL.PART_NUM = P.PART_NUM
INNER JOIN ORDERS O2
ON C.CUST_NUM = O2.CUST_NUM
INNER JOIN ORDER_LINE OL2
ON O2.ORDER_NUM = OL2.ORDER_NUM
INNER JOIN PART P2
ON OL2.PART_NUM = P2.PART_NUM
WHERE P.PART_DESCRIPTION IN ('GasRange') and P2.PART_DESCRIPTION IN ('Washer')
GROUP BY C.CUST_NAME
EDIT: Had a further look and I'm afraid that this can't be simplified in any other way than using WITH and complicated aggregate functions, which I would say would be more complicated than this - I think the other solution suggested using WITH won't work - it joins incorrectly. You definitely can't remove order line, and you have to use the order twice as well - if it was used once, it will cover only when the customer ordered it within one order, which is not what you wanted ;)
Try this...
So basically you need to join your Parts table again to ensure the same customer ordered a "Gas Range" and a "Washer". An IN, like in your current query functions as an OR therefore you are not getting the expected result.
WITH CTE AS (
SELECT DISTINCT O.CUST_NUM FROM ORDERS O
INNER JOIN ORDER_LINE OL
ON O.ORDER_NUM = OL.ORDER_NUM
INNER JOIN PART P
ON OL.PART_NUM = P.PART_NUM
INNER JOIN PART P2
ON OL.PART_NUM = P2.PART_NUM
WHERE P.PART_DESCRIPTION IN ('GasRange')
AND P2.PART_DESCRIPTION IN ('Washer')
)
SELECT C.CUST_NAME AS [Customer(s) that ordered a Gas Range and Washer]
FROM CUSTOMER C
INNER JOIN CTE O
ON C.CUST_NUM = O.CUST_NUM