Plan a reverse-dependency check — revdep

Discover reverse dependencies and estimate the preparation work without downloading, building, or checking packages. Direct reverse dependencies are always selected. Recursive-strong-only targets are opt-in and can be limited with a reproducible sample.

Usage

revdep_plan(
  package,
  recursive = FALSE,
  max_recursive = NULL,
  sample_seed = NULL,
  cache = NULL,
  repos = NULL
)

Arguments

package: Path to the development package checkout.
recursive: Include recursive strong reverse dependencies as candidates.
max_recursive: Optional maximum number of recursive-only targets to select. All direct targets remain selected.
sample_seed: Optional non-negative integer used to choose a different reproducible recursive sample. It can be supplied only with max_recursive.
cache: Cache directories to inspect for compatible binaries. NULL uses the ordinary crancache directory and the current runner-managed binary cache when they exist; character() disables cache inspection. An explicit value replaces default discovery.
repos: Named source repository base URLs. NULL combines the configured repositories with the standard Bioconductor repositories. An explicit value is used exactly. When a named CRAN repository is present, reverse targets come from CRAN while dependencies can come from every configured repository.

Value

A revdep_plan list with summary, targets, requirements, unavailable, repository_alternates, and workload tables. Per-target requirement counts overlap; use summary for totals over unique selected targets and their installation dependencies in requirements. Declared system requirements are metadata clues, not a platform-readiness check. Repository-unavailable Suggests remain in unavailable, but do not enter preparation requirements.

Details

workload explains the selected targets and their direct dependency roots. Its columns are target, relationship, root, requirements, native_requirements, and source_builds. Rows with relationship = "target" count the target itself and its installation dependencies. Other rows count a direct dependency root and its reachable installation dependencies, using the same frozen dependency policy. Base packages, the supplied subject, and unavailable packages do not count as preparation requirements. Counts overlap across roots and targets; they are not estimates of packages saved by dropping a dependency. Native counts mean packages declaring NeedsCompilation: yes; source-build counts reflect cache availability at planning time.

Preparation creates reusable dependency binaries. Comparison still runs source checks of each selected target against both released and development versions of the subject, including target compilation when needed.

Examples

if (FALSE) { # \dontrun{
plan <- revdep_plan("/path/to/package")
plan$summary

sampled <- revdep_plan(
  "/path/to/package",
  recursive = TRUE,
  max_recursive = 20
)
sampled$targets
} # }