Programming/Language/Concept/Declarative vs Imperative

• Declarative vs Imperative

  • Overview

    • Programming paradigms can be broadly grouped by whether code describes what to compute or how to compute it.
    • Declarative code expresses the desired outcome or constraint, leaving the runtime or compiler to figure out the execution steps — SQL queries, HTML markup, and CSS rules are canonical examples.
    • Imperative code specifies an explicit sequence of commands that mutate state to arrive at a result — C, shell scripts, and most mainstream loops are imperative at their core.
    • The distinction matters because declarative code trades control over execution for readability, correctness guarantees, and optimization opportunities; imperative code trades those for precise control over performance and side effects.

  • Context

    • The terms come from linguistics: a declarative sentence states a fact; an imperative sentence issues a command. The analogy is apt — a SQL SELECT states what rows we want; a loop with conditionals commands how to find them.
    • Wikipedia characterizes declarative programming as "a programming paradigm … that expresses the logic of a computation without describing its control flow" (Wikipedia: Declarative programming) and imperative programming as a paradigm that "uses statements that change a program's state" (Wikipedia: Imperative programming).
    • Most real languages are multi-paradigm: Python, JavaScript, and Kotlin support both styles; the question is usually which style dominates a given component or layer.
    • Software/Paradigm/Functional is often considered a subset of the declarative family, emphasizing pure functions and expressions over procedure sequences. Programming/Language/Concept/Object-Oriented blends with imperative style but is orthogonal to the declarative/imperative axis.

  • Key Principles

    • What vs How — Declarative: the program states what result is wanted; the runtime determines how. Imperative: the program specifies each step explicitly.
    • State mutation — Imperative programs typically read and write mutable state step by step. Declarative programs tend to avoid or hide mutation behind abstractions.
    • Control flow ownership — Imperative code owns control flow (loops, branches, exceptions). Declarative code delegates it to a runtime engine (query planner, reconciler, constraint solver).
    • Referential transparency (declarative ideal) — A sub-expression always yields the same result for the same inputs, enabling safe reuse and optimization.

  • Mechanism

    • Declarative execution models: query planners (SQL, Datalog), reactive rendering engines (React reconciler, CSS layout engine), logic unification (Prolog), constraint solvers (CSS flexbox, linear programming), and dependency-graph build systems (Make, Nix).
    • Imperative execution models: register-machine instruction sequences, stack-based bytecodes (JVM, CPython), shell pipelines (sequential commands with exit codes), and event-loop callbacks with mutable closures.
    • Hybrid layers are the norm in production code — React component trees are declarative (describe the UI), but event handlers inside them are often imperative (mutate refs, call APIs sequentially).

  • Examples

    • 1. SQL — SELECT name FROM users WHERE active = true is declarative; an equivalent Python loop iterating over rows and filtering is imperative.
    • 1. HTML/CSS — Setting display: flex; justify-content: center is declarative; computing pixel positions in JavaScript is imperative.
    • 1. Nix/NixOS — A Nix expression describes a system configuration; Nix decides how to build and link it. Contrast with shell-script provisioning.
    • 1. Terraform / Kubernetes YAML — Describe desired infrastructure state; the tool reconciles current reality to match the spec.
    • 1. React JSX — Describe what the UI should look like given current state; the reconciler figures out minimal DOM mutations.

  • Misconceptions

    • "Declarative is always better" — False; when execution semantics matter (performance, ordering, error recovery), declarative abstraction can hide costs or make them hard to diagnose. Imperative code gives more control.
    • "Functional = declarative" — Mostly true in spirit but not definitionally identical. Functional programming is a specific discipline (pure functions, immutability) that is highly declarative; declarative programming is the broader category (SQL is declarative but not functional).
    • "Imperative is lower-level" — Oversimplified; OOP is imperative in style but high-level in abstraction. The axis is about who controls flow, not about abstraction height.
    • "You must choose one" — False; most production systems layer declarative configuration over imperative business logic, or use a declarative DSL that compiles to imperative code.