Named windows with WINDOW clause in PostgreSQL

Introduction

Named windows let you reuse the same window definition in multiple places in your query. This saves time and keeps your query neat.

When you want to apply the same sorting and partitioning rules to several window functions.

When your query has many window functions that share the same frame or partition.

When you want to make your SQL easier to read and maintain by avoiding repeated code.

Syntax

PostgreSQL

WINDOW window_name AS (window_definition)

The WINDOW clause is placed after the FROM clause and before ORDER BY.

You can define multiple named windows separated by commas.

Examples

This defines a window named w that groups rows by department and orders them by salary descending.

PostgreSQL

WINDOW w AS (PARTITION BY department ORDER BY salary DESC)

Uses the named window w in the RANK() function to rank employees by salary within each department.

PostgreSQL

SELECT name, salary, RANK() OVER w FROM employees WINDOW w AS (PARTITION BY department ORDER BY salary DESC);

Defines two named windows: w1 partitions by category, w2 orders by date.

PostgreSQL

WINDOW w1 AS (PARTITION BY category), w2 AS (ORDER BY date)

Sample Program

This example creates a sales table, inserts some data, and then uses a named window w to calculate total sales and rank within each region.

PostgreSQL

CREATE TABLE sales (
  id SERIAL PRIMARY KEY,
  region TEXT,
  amount INT
);

INSERT INTO sales (region, amount) VALUES
('North', 100),
('South', 150),
('North', 200),
('South', 120),
('East', 130);

SELECT region, amount,
  SUM(amount) OVER w AS total_region_sales,
  RANK() OVER w AS rank_in_region
FROM sales
WINDOW w AS (PARTITION BY region ORDER BY amount DESC);

Important Notes

Named windows help avoid repeating the same partition and order clauses multiple times.

You can refer to a named window inside another window definition by using WINDOW new_name AS (existing_name [frame]).

Summary

Named windows let you define window specifications once and reuse them.

This makes queries shorter, clearer, and easier to maintain.

Use the WINDOW clause after FROM and before ORDER BY.

Practice

1. What is the main purpose of using the WINDOW clause in PostgreSQL?

easy

A. To join multiple tables using window functions

B. To create temporary tables for complex queries

C. To define named window specifications for reuse in the query

D. To specify the order of rows in a SELECT statement

Solution

1. Step 1: Understand the role of WINDOW clause

   The WINDOW clause allows defining named window specifications once for reuse.

2. Step 2: Compare options with WINDOW clause purpose

   Options A, B, and C describe other SQL features, not the WINDOW clause.

3. Final Answer:

   To define named window specifications for reuse in the query -> Option C

4. Quick Check:

   WINDOW clause = named window reuse [OK]

Hint: WINDOW clause defines reusable window specs [OK]

Common Mistakes:

• Confusing WINDOW with temporary tables
• Thinking WINDOW orders rows directly
• Mixing WINDOW with JOIN operations

2. Which of the following is the correct syntax to define a named window called w1 that orders rows by date ascending?

easy

A. WINDOW w1 (ORDER BY date ASC)

B. WINDOW w1 AS (ORDER BY date ASC)

C. WINDOW w1 BY (ORDER date ASC)

D. WINDOW AS w1 ORDER BY date ASC

Solution

1. Step 1: Recall WINDOW clause syntax

   The correct syntax is WINDOW name AS (window_definition), e.g., WINDOW w1 AS (ORDER BY date ASC).

2. Step 2: Check each option's syntax

   WINDOW w1 AS (ORDER BY date ASC) matches correct syntax. Options A, B, and D have incorrect keyword order or missing AS.

3. Final Answer:

   WINDOW w1 AS (ORDER BY date ASC) -> Option B

4. Quick Check:

   WINDOW name AS (definition) syntax [OK]

Hint: Use WINDOW name AS (definition) syntax [OK]

Common Mistakes:

• Omitting AS keyword
• Using BY instead of ORDER BY
• Misplacing parentheses

3. Given the query:

SELECT department, employee, salary, AVG(salary) OVER w AS avg_salary
FROM employees
WINDOW w AS (PARTITION BY department ORDER BY employee)
ORDER BY department, employee;

What does the avg_salary column show?

medium

A. The average salary of employees in the same department up to the current employee

B. The average salary of all employees in the company

C. The average salary of employees partitioned by department without order

D. The average salary of employees ordered by employee name across all departments

Solution

1. Step 1: Analyze the WINDOW clause

   The window named w partitions rows by department and orders by employee.

2. Step 2: Understand AVG() OVER w behavior

   AVG(salary) OVER w calculates the average salary for rows in the same department up to the current employee in order.

3. Final Answer:

   The average salary of employees in the same department up to the current employee -> Option A

4. Quick Check:

   Partition + order = running average per department [OK]

Hint: Partition + order means running calculation per group [OK]

Common Mistakes:

• Ignoring ORDER BY in window
• Assuming average over entire table
• Confusing partition with no partition

4. Consider this query:

SELECT name, salary, SUM(salary) OVER w
FROM employees
WINDOW w AS (PARTITION BY department ORDER BY name BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW);

It returns an error. What is the likely cause?

medium

A. The window frame clause requires ROWS or RANGE keyword explicitly

B. The window frame clause is invalid without RANGE keyword

C. The WINDOW clause must come after ORDER BY clause

D. The window name cannot be used in the OVER clause

Solution

1. Step 1: Check window frame syntax

   The frame clause must specify ROWS or RANGE explicitly before frame boundaries.

2. Step 2: Identify error cause

   The query has ORDER BY name BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW, missing the required ROWS or RANGE keyword before BETWEEN.

3. Final Answer:

   The window frame clause requires ROWS or RANGE keyword explicitly -> Option A

4. Quick Check:

   Window frame needs ROWS or RANGE keyword [OK]

Hint: Always specify ROWS or RANGE in window frame [OK]

Common Mistakes:

• Placing WINDOW clause after ORDER BY
• Omitting ROWS or RANGE keyword
• Using window name incorrectly in OVER

5. You want to calculate the cumulative sum of sales per region ordered by date, and reuse this window in multiple queries. Which is the best way to define and use a named window for this purpose?

hard

A. Define WINDOW sales_win AS (PARTITION BY region ORDER BY date ROWS UNBOUNDED PRECEDING) and use SUM(sales) OVER sales_win

B. Use SUM(sales) OVER (PARTITION BY region ORDER BY date) without WINDOW clause

C. Define WINDOW sales_win AS (ORDER BY date PARTITION BY region) and use SUM(sales) OVER sales_win

D. Define WINDOW sales_win AS (PARTITION BY region ORDER BY date RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) and use SUM(sales) OVER sales_win

Solution

1. Step 1: Understand cumulative sum window frame

   Cumulative sum requires a frame from the start to current row, so use RANGE or ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW.

2. Step 2: Check window definition correctness

   Define WINDOW sales_win AS (PARTITION BY region ORDER BY date RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) and use SUM(sales) OVER sales_win correctly defines WINDOW with PARTITION BY region, ORDER BY date, and RANGE frame for cumulative sum.

3. Step 3: Compare with other options

   Define WINDOW sales_win AS (PARTITION BY region ORDER BY date ROWS UNBOUNDED PRECEDING) and use SUM(sales) OVER sales_win uses ROWS UNBOUNDED PRECEDING which sums the entire partition, not cumulatively. Define WINDOW sales_win AS (ORDER BY date PARTITION BY region) and use SUM(sales) OVER sales_win has wrong order of clauses. Use SUM(sales) OVER (PARTITION BY region ORDER BY date) without WINDOW clause does not reuse named window.

4. Final Answer:

   Define WINDOW sales_win AS (PARTITION BY region ORDER BY date RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) and use SUM(sales) OVER sales_win -> Option D

5. Quick Check:

   Named window with correct frame = Define WINDOW sales_win AS (PARTITION BY region ORDER BY date RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) and use SUM(sales) OVER sales_win [OK]

Hint: Use RANGE UNBOUNDED PRECEDING for cumulative sums [OK]

Common Mistakes:

• Omitting frame clause for cumulative sum
• Incorrect order of PARTITION BY and ORDER BY
• Not using named window for reuse