Lazy Evaluation

Lazy evaluation is a computation strategy in which expressions are evaluated only when their values are actually needed, rather than at the point of binding or invocation. This seemingly minor technical choice has profound consequences: it enables the manipulation of infinite data structures, decouples producers from consumers through implicit streams, and transforms the programmer's model of time from eager sequencing to demand-driven revelation. Lazy evaluation is the default semantics of Haskell and was central to the design of languages like Miranda and Hope. Its theoretical foundation lies in the distinction between call-by-name and call-by-need — the latter ensuring that an expression is evaluated at most once, with subsequent references sharing the computed result. The trade-off is memory unpredictability: unevaluated expressions accumulate as thunks in the heap, and space leaks remain the characteristic pathology of lazy languages, requiring programmers to develop an intuition for when strictness is the greater virtue.