R/bigANNOY-package.R

In bigANNOY: Approximate k-Nearest Neighbour Search for 'bigmemory' Matrices with Annoy

#' bigANNOY: Annoy Search for bigmemory Matrices
#'
#' Approximate nearest-neighbour search for [`bigmemory::big.matrix`]
#' references through Annoy indexes built on disk, with multi-metric Annoy
#' support, explicit loaded-index lifecycle helpers, descriptor-aware
#' file-backed workflows, benchmark helpers, and in-memory or streamed
#' `big.matrix` result writes.
#'
#' Package options:
#' - `bigANNOY.block_size`: default number of rows processed per block while
#'   building an index or reading query matrices. Defaults to `1024L`.
#' - `bigANNOY.backend`: backend used by the public API. Defaults to `"cpp"`.
#'   Set to `"r"` only for debugging/comparison, or `"auto"` to prefer the
#'   native backend when available.
#' - `bigANNOY.progress`: logical flag controlling simple progress messages
#'   during index build and search. Defaults to `FALSE`.
#'
#' @keywords internal
#' @importFrom Rcpp evalCpp
#' @useDynLib bigANNOY, .registration = TRUE
"_PACKAGE"