Buckets and bucket sources

Description

Handling of large data buffers.

Buckets (tcspc::bucket instances) are used where a single event carries a large quantity of data, such that it is beneficial to avoid copying the data, including to its ultimate destination. Correct use of buckets enables zero-copy writing of processing results to externally-provided destination memory.

Bucket sources provide bucket instances using different backing storage and management thereof.

Topics
	Bucket sources
	Objects producing a series of tcspc::bucket instances to carry data.

Classes
class	tcspc::bucket< T >
	Value-semantic container for array data allowing use of custom storage. More...

Functions
template<typename T>
auto	tcspc::ad_hoc_bucket (std::span< T > s) -> bucket< T >
	Create a tcspc::bucket referencing a span.

Function Documentation

ad_hoc_bucket()

template<typename T>

auto tcspc::ad_hoc_bucket ( std::span< T > s ) -> bucket< T >

nodiscard

Create a tcspc::bucket referencing a span.

This can be used when a bucket is needed but is only used to view existing data and its storage is not important. The storage of the returned bucket cannot be observed or extracted.

Attention

The caller is responsible for ensuring that the data in the span outlives the returned bucket. Usually this means that the returned bucket should only be published (e.g., by emitting to downstream) via const reference and should never be returned or stored by the caller.

Template Parameters

T	the bucket data element type

Parameters

s	the span to wrap as an ad-hoc bucket