12.1 Introduction to Sets

What Is a Set?

What Is It?

An unordered collection of unique, hashable values. Written with curly braces, or via set().

>>> unique_ports = {80, 443, 8080}
>>> unique_ports
{80, 443, 8080}

Sets were already introduced briefly in 5.6 Set Data Types; this chapter covers them in full.

Why Does It Matter?

Whenever “does this exist?” or “remove duplicates” matters more than order or position, a set is the right tool — this is a genuinely different job from what a list or tuple is built for.

Characteristics of Sets

• Unordered — no guaranteed position or insertion order (see 12.6 Set Indexing and Ordering).
• No duplicates — adding an existing value is a no-op.
• Mutable — elements can be added/removed after creation (see 12.4 Adding Elements and 12.5 Removing Elements).
• Elements must be hashable — no lists or dicts as members (see 12.17 Common Mistakes); this follows the same rule covered in 5.9 Mutable vs Immutable Types.

Why Use Sets?

Two things a set does better than any other built-in type: automatic deduplication, and near-instant membership testing — O(1) average, versus a list’s O(n) (see 10.6 List Operators for the list-side comparison). Both are covered in depth in 12.13 Performance.

Real-World Applications

• Deduplicating a list of email addresses
• Tracking which users have already been notified
• Finding common tags between two articles
• Validating a value against an allowed set of options

Sets in DevOps

Deduplicating IP addresses from logs, comparing installed vs. required packages, diffing security group rules — 12.16 Sets in DevOps is dedicated entirely to these patterns.

>>> unique_ips = {"10.0.0.1", "10.0.0.2", "10.0.0.1"}
>>> unique_ips
{'10.0.0.1', '10.0.0.2'}

Quick Interview Answer

“A set is an unordered collection of unique, hashable values, built for exactly two jobs: automatic deduplication and fast membership testing. Internally it’s a hash table (see 12.3 Internal Representation), so checking x in s is O(1) average instead of the O(n) a list requires — that performance difference is the entire reason the type exists. The trade-offs that follow directly from being hash-based: no indexing, no guaranteed iteration order, and every element must itself be hashable.”

Common Mistakes

• Reaching for a list and manually checking in repeatedly, when converting to a set once up front would make each check O(1) instead of O(n) (see 12.18 Best Practices).
• Assuming a set preserves insertion order the way a list, tuple, or (since Python 3.7) a dict does — it doesn’t.
• Writing s = {} expecting an empty set — that creates an empty dict instead (see 12.2 Creating Sets).