object_usage_linter doesn't match R CMD check

[MichaelChirico]

As seen here: #931 (comment)

Basically our tools::checkUsage() call doesn't match that of R CMD check. In this case, it's because the default checkUsage() doesn't set skipWith=TRUE, so objects seen inside with() calls can trip the linter.

My understanding is the main goal of object_usage_linter is to help match R CMD check, so we should adjust our code to match.

But maybe we think of this as a bit more general, and could expose an option instead (e.g. match_r_check, I would set TRUE by default).

[MichaelChirico]

PS I think this linter should be tagged as package_development

[AshesITR]

Is there a legitimate true positive caused by skipWith = FALSE?
I reckon most with() calls are using columns from a data frame and thus always trip checkUsage() unless the used columns are declared outside of the data frame as well.

BTW:

checkUsage(fun, name = "<anonymous>", report = cat, all = FALSE, 
           suppressLocal = FALSE, suppressParamAssigns = !all,
           suppressParamUnused = !all, suppressFundefMismatch = FALSE,
           suppressLocalUnused = FALSE, suppressNoLocalFun = !all,
           skipWith = FALSE, suppressUndefined = dfltSuppressUndefined,
           suppressPartialMatchArgs = TRUE) 

suppressPartialMatchArgs = FALSE sounds interesting.
Also, suppressLocalUnused = symbols_in_glue_calls seems like the way to support glue::glue().