Skip to contents

Contributing to OECDscaleup

Source: CONTRIBUTING.md

Thanks for contributing. This package is a behaviour-preserving translation of the OECD Scale Up project’s Stata pipeline, used by national statistical institutes across the OECD to compute internationally-comparable high-growth-firm statistics — correctness and faithfulness to the documented methodology matter more here than in most packages.

Before you start

  • For anything beyond a small fix, open an issue first to discuss the change — especially anything touching indicator definitions (build_accounts_indicators()), the classification rules in compute_persistence()/compute_predictors(), or the disclosure-control logic (dominance_share(), approx_median()).
  • If you’re changing behaviour rather than just refactoring, check whether the change should also be reflected in Pseudocode/*.md and Analysis/stata.md in the source repository this package was translated from, so the two stay in sync.

Setup

renv::restore()               # install the pinned dependency versions
devtools::load_all()          # load the package for interactive development
devtools::test()              # run the test suite

Making a change

  1. Create a branch for your change.
  2. Make the change, adding or updating tests in tests/testthat/ — every exported function should have test coverage for its normal case, at least one edge case (missing/NA inputs are the most common source of subtle bugs in this codebase — see stata_num()/stata_replace()’s documentation for why), and, for anything touching Stata-derived semantics, a test that documents why the expected value is what it is (not just that it matches).
  3. Run devtools::document() if you added or changed any roxygen comments, and commit the regenerated files under man/ and NAMESPACE.
  4. Run devtools::test() and devtools::check() locally before opening a pull request — both also run in CI on every push and pull request.
  5. Update NEWS.md with a one-line summary of user-facing changes.

Code style

  • Follow the tidyverse style already used throughout (snake_case, dplyr-based data wrangling, roxygen2 with markdown enabled).
  • Prefer adding a well-named, tested helper in R/utils-panel.R over inlining a workaround for a Stata/R semantic gap — stata_num() and stata_replace() both started this way, and a third such gap is likely to recur elsewhere in the codebase.
  • Don’t add speculative configuration options, abstractions, or generality beyond what an actual, current requirement needs.

Reporting bugs

Please include: the function involved, a minimal reproducible example (synthetic data only — never attach real firm microdata to an issue, even internally), and, if you believe the translated behaviour diverges from the original Stata pipeline, which Pseudocode/*.md step you believe it diverges from.