Stack-wide conventions (naming, style, git policy, API principles) are unified in optimal2050 CONVENTIONS.md; this file covers the workflow of this repo.
Thanks for your interest! timescales is part of the optimal2050 modeling stack and is developed alongside its spatial companion geoscales.
The repository is public and contributions are welcome — issues, ideas, and pull requests alike. The package is pre-1.0, so APIs may still move; open an issue first for larger changes.
Languages and phases
- Phase 1 (current) — pure R implementation.
-
Phase 2 — C++17 core under
cpp/; R bindings via Rcpp insrc/. -
Phase 3 — Python bindings under
python/via nanobind + scikit-build-core.
A single semantic version (in VERSION) drives all language artifacts.
Repository layout
timescales/
├── DESCRIPTION, NAMESPACE, R/, man/, tests/, vignettes/ # R package (root)
├── inst/include/timescales/ # C++ headers (Phase 2)
├── src/ # Rcpp glue (Phase 2)
├── cpp/ # standalone C++ core (Phase 2)
├── python/ # Python package (Phase 3)
├── docs/ # unified Quarto site
├── specs/ # cross-language golden tests
├── benchmark/ # cross-language benchmarks
└── .github/workflows/ # CIDevelopment workflow (R)
# from the repo root
devtools::load_all()
devtools::test() # fast tier (default)
Sys.setenv(SCALES_TEST_TIER = "full"); devtools::test() # backend sweep
devtools::document()
devtools::check()Testing system: see tests/README.md (tiers, @covers tags, harnesses) and dev/TESTING.md (recipes and traps). New behaviour lands with tests; coverage gaps show in tests/coverage/matrix-summary.md (Rscript tools/coverage/build_matrix.R).
Style: follow tidyverse style; 2-space indent; snake_case function names.
Cross-language consistency
Language-agnostic input/output pairs live under specs/golden/. Each language implementation must reproduce them identically. When you change observable behaviour, update both the spec and the per-language tests in the same PR.