Singlepass

Singlepass, or single-pass, is a term used in computer science to describe processes that process input data with a single traversal, without requiring a second pass over the same data. This approach contrasts with multi-pass methods, where information is revisited or accumulated across several stages. Single-pass techniques emphasize streaming and online processing, enabling low latency and reduced memory usage, but often impose design constraints that trade off some flexibility or optimization opportunities.

Common contexts for single-pass design include compilers or interpreters for simple or domain-specific languages, streaming parsers that consume input characters as they are read, and data processing pipelines that perform filtering, transformation, or aggregation in one pass. In these settings, outputs are typically produced incrementally, and systems are optimized for continuous rather than batch workloads.

Advantages of single-pass approaches include lower memory footprint, better suitability for large or real-time data sources, and simpler pipeline architectures. Limitations include restricted ability to handle forward references or perform global optimizations, and potentially less informative error reporting because certain semantic checks may require awaiting additional input.

In practice, true single-pass systems are most common in constrained environments with lightweight languages or real-time requirements. They may use incremental symbol tables or progressive output construction, and employ buffering only when necessary to resolve essential dependencies.

Related concepts include streaming algorithms, one-pass parsers, and incremental compilation.