Contributing to knbsdata

Thanks for your interest. This project is in early development, so the most valuable contributions right now are source discovery findings and corrections to how the package describes access conditions.

Ground rules

These are not negotiable, and a pull request that breaks one will not be merged:

1. Official sources first. Data must come from KNBS-controlled or KNBS-authorised systems. No third-party mirrors by default.
2. Never bypass access controls. No circumventing registration, licences, approval workflows, authentication, rate limits, robots exclusions or human verification. A restricted dataset may be listed in the catalogue; it must not be downloaded.
3. Never commit secrets. No credentials, tokens or cookies in code, fixtures, registry files, manifests or test snapshots.
4. Fail explicitly. Never return silently incomplete data. Use the classed conditions in R/conditions.R.
5. No silent guessing. Do not infer indicator ids, units, geographic vintages, release dates, missing-value meanings or access permissions.
6. No live network in ordinary tests. Unit tests use fixtures. Live checks belong in the scheduled test-sources workflow.

Development setup

install.packages("devtools")
devtools::load_all()
devtools::test()
devtools::document()
devtools::check()

Style follows the tidyverse style guide: two-space indentation, snake_case, small functions, early validation. lintr runs in CI; the configuration is in .lintr.

Pull requests

• Open an issue first for anything larger than a fix, so the approach can be agreed before you write code.
• One vertical slice at a time. The plan sequences work deliberately; adding a second source before the first is finished and tested creates rework.
• Add tests with fixtures for any new behaviour, including the failure paths.
• Run devtools::document() if you touched roxygen comments, and commit the generated man/ and NAMESPACE.
• Update NEWS.md and dev/IMPLEMENTATION_STATUS.md.

Adding a source adapter

A source adapter is complete only when every item below is satisfied. Please copy this checklist into your pull request.

• Official ownership documented
• Terms and access reviewed
• Supported operations documented
• Public endpoints identified and recorded in dev/source-discovery/
• Registry entry added and validated
• Pagination implemented
• Errors handled with classed conditions
• Rate limiting respected
• Fixtures stored (normal, empty, paginated, missing field, changed field, restricted, error, duplicate, provisional)
• Parser implemented
• Standardiser implemented
• Required schema validated
• Metadata preserved
• Citation supported
• Cache behaviour tested
• Live health check available
• User documentation added
• Source limitations documented
• No credentials committed
• No access controls bypassed

Source discovery contributions

You do not need to write R code to help. Documenting a source in dev/source-discovery/source-audit.md — its landing page, whether it exposes a machine-readable endpoint, its terms of use, its access conditions and its stability — is the current bottleneck. Record public, documented endpoints only, and never anything obtained from a session token or by circumventing a control.

Code of conduct

Please note that this project is released with a Contributor Code of Conduct. By contributing you agree to abide by its terms.