Local R CMD check --as-cran --no-manual mlumr_0.1.0.tar.gz on macOS
(aarch64-apple-darwin20, R 4.5) reports:
- 0 errors
- 2 warnings (both environmental; see "Environmental notes" below)
- 1 note (expected: new submission + cmdstanr declared via
Additional_repositories)
No tests fail and no examples error.
- local macOS (aarch64-apple-darwin20), R 4.5
- planned before release: win-builder (devel and release), R-hub (Fedora, Windows Server, Ubuntu Linux)
Two warnings surface only on the submitter's macOS / Apple clang toolchain and are not expected to appear on CRAN's build servers:
- Stale Homebrew
ldsearch path:ld: warning: search path '/opt/homebrew/Cellar/r/4.5.1/lib/R/lib' not found. The submitter previously had a Homebrew-installed R; the path is now stale. The build still completes successfully. checkbashismsscript not installed locally: the macOS toolchain does not ship the Debiancheckbashismshelper, so R CMD check reports that "a complete check needs the 'checkbashisms' script." CRAN's Linux servers run this check normally.
- This is a new submission.
- The package embeds six Stan models (binomial / normal / poisson, each with SPFA and relaxed variants). They are compiled at install time via the standard rstantools pipeline; expect several minutes of C++ compilation.
cmdstanris listed inSuggestswithAdditional_repositories: https://stan-dev.r-universe.dev. This follows the same approach adopted by brms and other Stan-ecosystem packages: cmdstanr is an optional backend, and rstan is the default, CRAN-available engine. Examples, tests, and vignettes do not require cmdstanr; cmdstanr comparison code is skipped unless cmdstanr and CmdStan are available.- Developer-only directories and build artifacts are excluded via
.Rbuildignore; the current source tarball weighs about 380 KB. - The installed package is approximately 6.9 MB, dominated by the six
Stan-generated shared libraries (5.4 MB in
libs/). This is in line with other Stan-backed CRAN packages (e.g.rstanarm,brms) and is the unavoidable consequence of shipping pre-compiled Stan models.