":=" syntax and assignment expressions: what and why?

Question

PEP 572 introduces assignment expressions (colloquially known as the Walrus Operator), implemented for Python 3.8. This seems like a really substantial new feature since it will allow this form of assignment within comprehensions and lambda functions.

What exactly are the syntax, semantics, and grammar specification of assignment expressions?

Why is this new (and seemingly quite radical concept) being introduced, when a similar idea in PEP 379 on "Adding an assignment expression" was rejected before?

Answer 1

PEP 572 contains many of the details, especially for the first question. I'll try to summarise/quote concisely arguably some of the most important parts of the PEP:

Rationale

Allowing this form of assignment within comprehensions, such as list comprehensions, and lambda functions where traditional assignments are forbidden. This can also facilitate interactive debugging without the need for code refactoring.

Recommended use-case examples

a) Getting conditional values

for example (in Python 3):

command = input("> ") while command != "quit":     print("You entered:", command)     command = input("> ") 

can become:

while (command := input("> ")) != "quit":     print("You entered:", command) 

Similarly, from the docs:

In this example, the assignment expression helps avoid calling len() twice:

if (n := len(a)) > 10:     print(f"List is too long ({n} elements, expected <= 10)") 

b) Simplifying list comprehensions

for example:

stuff = [(lambda y: [y,x/y])(f(x)) for x in range(5)]

can become:

stuff = [[y := f(x), x/y] for x in range(5)]

Syntax and semantics

In any context where arbitrary Python expressions can be used, a named expression can appear. This is of the form name := expr where expr is any valid Python expression, and name is an identifier.

The value of such a named expression is the same as the incorporated expression, with the additional side-effect that the target is assigned that value

Differences from regular assignment statements

In addition to being an expression rather than statement, there are several differences mentioned in the PEP: expression assignments go right-to-left, have different priority around commas, and do not support:

• Multiple targets

x = y = z = 0  # Equivalent: (z := (y := (x := 0))) 

• Assignments not to a single name:

# No equivalent a[i] = x self.rest = [] 

• Iterable packing/unpacking

# Equivalent needs extra parentheses  loc = x, y  # Use (loc := (x, y)) info = name, phone, *rest  # Use (info := (name, phone, *rest))  # No equivalent  px, py, pz = position name, phone, email, *other_info = contact 

• Inline type annotations:

# Closest equivalent is "p: Optional[int]" as a separate declaration p: Optional[int] = None 

• Augmented assignment is not supported:

total += tax  # Equivalent: (total := total + tax) 

Answer 2

A couple of my favorite examples of where assignment expressions can make code more concise and easier to read:

if statement

Before:

match = pattern.match(line) if match:     return match.group(1) 

After:

if match := pattern.match(line):     return match.group(1) 

Infinite while statement

Before:

while True:     data = f.read(1024)     if not data:         break     use(data) 

After:

while data := f.read(1024):     use(data) 

There are other good examples in the PEP.