stats, utils are Imports

[MichaelChirico]

data.table/NAMESPACE

Line 156 in e26cf1b

importFrom(utils, capture.output, contrib.url, download.file, flush.console, getS3method, head, packageVersion, tail, untar, unzip)

data.table/NAMESPACE

Line 160 in e26cf1b

importFrom(stats, as.formula, na.omit, setNames, terms)

The importFrom() directives clearly mark these out as Imports IMO.

[jangorecki]

If we go this route I would argue then we should have a job that runs tests with stats and utils not attached to the session, then we can actually test if those imports are properly (sufficiently) defined. If they wouldn't we should expect a NOTE, but still.

[MichaelChirico]

R_DEFAULT_PACKAGES=null still wouldn't find this, you actually have to uninstall {stats} for it to cause any issue as of today (which is how I found it 😅)

Still, it should produce an R CMD check note, it's just that it's not enforced: https://cran.r-project.org/doc/manuals/r-devel/R-exts.html#Package-Dependencies

The ‘Imports’ field lists packages whose namespaces are imported from... Ideally this field will include all the standard packages that are used

[ben-schwen]

The ‘Imports’ field lists packages whose namespaces are imported from... Ideally this field will include all the standard packages that are used

Apparently R Core members are really doing this see e.g. Matrix from Martin Maechler or slam from Kurt.

[MichaelChirico]

We don't even need to go to package space, the packages are citing each other too, e.g. {stats}

https://github.com/r-devel/r-svn/blob/a43a1dec67bdba665ebf71e528e1c8a556f3c278/src/library/stats/DESCRIPTION.in#L10-L11

Imports: utils, grDevices, graphics
Suggests: MASS, Matrix, SuppDists, methods, stats4