Premature Optimisation

Also Known As

TL;DR

Explanation

Donald Knuth's famous observation — 'premature optimisation is the root of all evil' — warns against sacrificing code clarity for performance improvements that are speculative or negligible. Most performance problems lie in a small fraction of the code; optimising the wrong parts wastes time and produces harder-to-maintain code. The correct approach: write clear, correct code first; measure with a profiler to find real bottlenecks; optimise only what the data confirms is slow. In PHP, micro-optimisations (avoiding function calls, hand-rolling loops) almost never matter — I/O and query patterns almost always do.

Common Misconception

Why It Matters

Common Mistakes

• Micro-optimising loop constructs (for vs foreach vs while) without benchmarking actual impact.
• Using static methods or avoiding OOP for 'performance' in application code where the overhead is negligible.
• Caching results before measuring whether the operation is actually slow.
• Choosing a complex algorithm over a simple one for scale that the application will never reach.

Code Examples

// Micro-optimised before profiling — complex and wrong
public function getActiveUsers(): array {
    // Avoid array_filter — 'too slow'
    \$result = [];
    \$users  = \$this->users;
    \$count  = count(\$users); // cache count 'for speed'
    for (\$i = 0; \$i < \$count; \$i++) {
        if (\$users[\$i]->active) \$result[] = \$users[\$i];
    }
    return \$result;
}

// Write the clear version first — profile before optimising
public function getActiveUsers(): array {
    return array_values(array_filter(\$this->users, fn(\$u) => \$u->active));
}

// If this is actually slow (profiler shows it) — fix the query, not the PHP:
// SELECT * FROM users WHERE active = 1  (let the DB filter, not PHP)

References

• ↗ https://en.wikipedia.org/wiki/Program_optimization#When_to_optimize

Tags

Related Terms