Twopass

Twopass, or two-pass, describes processing data in two passes over the input. In this approach, information discovered during the first pass informs the second pass, enabling tasks such as resolving forward references and handling dependencies that cannot be settled in a single sweep.

In computing, the two-pass design is common in assemblers and compilers. A typical two-pass assembler first traverses the source to collect labels, macros, and symbol definitions and to build a symbol table. The second pass re-reads the source to emit machine code, substituting symbolic addresses with concrete values. This separation simplifies symbol resolution and adds opportunities for error reporting for undefined symbols.

Two-pass techniques also appear in parsing and data processing, where the first pass gathers statistics or constructs an index, and the second pass performs the main computation or data transformation. The approach can also be used in other domains that require resolving interdependencies before producing final output.

Advantages of the two-pass approach include support for forward references and straightforward handling of complex dependencies. Drawbacks include the need to read the input twice, which can be less efficient, and higher memory usage to retain intermediate data and symbol tables. In modern systems, many tools employ single-pass or incremental strategies with lookahead or lazy evaluation, but two-pass methods remain relevant in resource-constrained environments or simpler implementations.

Related terms include two-pass assembler and two-pass compiler, which describe two-pass implementations within their respective domains.