rxdropped

rxdropped is a term used in reactive programming to describe a situation where items emitted by a source observable are discarded before they reach downstream observers. This typically arises when the producer emits more items than the consumer can process, a scenario known as backpressure, or when bounded buffers overflow.

In backpressure-aware streams, dropping is one of several strategies to prevent unbounded memory growth. Certain operators implement dropping semantics to actively discard items that cannot be observed in time. For example, in RxJava, the onBackpressureDrop operator discards items when the downstream is overwhelmed. Other related backpressure strategies include onBackpressureBuffer, which buffers items (and may overflow), and onBackpressureLatest, which keeps only the most recent item. While dropping reduces memory pressure, it also results in data loss, which may be acceptable in some domains (such as sampling sensors) and unacceptable in others.

Across different implementations, related concepts involve rate limiting and window-based sampling. In RxJS and other Rx libraries, similar outcomes are achieved with operators that throttle, debounce, or sample emissions, effectively dropping intermediate events to meet a desired throughput.

Implications and usage considerations include the need to understand and document whether and when dropped items occur, and how this affects data integrity and system behavior. Designers should choose an appropriate backpressure strategy based on domain requirements, consider observability of drop rates, and balance latency, memory usage, and data completeness.

See also: backpressure, drop strategy, onBackpressureDrop, buffer/overflow, latest retention.