inlining

Inlining is a compiler optimization and programming practice in which the body of a function is substituted directly at each call site, replacing the overhead of a function call with a direct sequence of instructions. The technique can reduce runtime overhead and enable further optimizations such as constant folding, register allocation, and better instruction cache utilization. However, inlining can also enlarge the generated code, a drawback known as code growth or code bloat, which may negate performance gains if it harms instruction cache locality.

In languages like C and C++, the inline keyword marks a function as a candidate for inlining and allows its definition to appear in multiple translation units. The keyword, however, does not guarantee that inlining will occur; the compiler ultimately decides based on heuristics and optimization settings. Modern compilers typically perform automatic inlining for small or frequently called functions regardless of the inline specifier. C++ also uses inline semantics to permit the one-definition rule to be satisfied when a function is defined in a header included by multiple translation units. C++17 added inline variables, extending similar inline semantics to variables.

In other programming environments, inlining is primarily a runtime optimization carried out by just-in-time (JIT) compilers or virtual machines. For example, Java and .NET runtimes commonly inline hot methods during execution, often guided by profiling data. Inlined code may later face deoptimization if assumptions change.

Common considerations include the trade-off between speed and code size, the function’s complexity, and the likelihood that inlining will improve locality of reference. Inlining is not a universal performance fix, and its benefits are workload- and architecture-dependent.