Skip to content

Statically import R functions into packages and projects

License

Unknown, MIT licenses found

Licenses found

Unknown
LICENSE
MIT
LICENSE.md
Notifications You must be signed in to change notification settings

wch/staticimports

Repository files navigation

staticimports

R-CMD-check

staticimports makes it easy to statically import functions into R projects or packages. The “static” importing means that the functions are copied into the project as text, instead of being loaded from a separate package at run time.

The benefits of doing things this way:

  • A project or package that uses staticimports will not take a run-time dependency on the staticimports package.
  • Statically imported functions will not change underneath you unexpectedly. They only will change when you re-import them, and then you can inspect the diff for changes that may cause problems.
  • You can specify the individual functions to import. This is faster and lighter-weight than loading a separate package, especially if the package has much more functionality than you need.
  • There is zero commitment: After you use staticimports once, you never have to use it again. You can even move the imported functions out to other files and treat them like any other functions in your project.

Where can statically imported functions come from?

  • Any package with an inst/staticexports/ directory. This includes the staticimports package itself, which contains many utility functions in inst/staticexports/.
  • A local directory on your computer. For example, you may maintain your own set of functions that you can statically import into a project.
  • (Not implemented yet) A directory in a remote Git repository.

Instead of copying and pasting utility functions from project to project, the utility functions can be centralized in a place where they can be vetted and tested.

The functions in staticimports are designed to be:

  • Fast
  • Simple
  • Have no external dependencies

If your project imports a function from staticimports, and that function changes in a way that has a negative impact on your project, you can simply stop importing it from staticimports, and copy the old version to a separate file.

Installation

You can install the development version of staticimports with:

remotes::install_github("wch/staticimports")

Usage

To use, put a comment block starting with # @staticimports in one of your R source files. For example, your utils.R may have this at the top:

# @staticimports pkg:staticimports
#  os_name %||%
#  map walk

The pkg:staticimports tells it to import from staticimports package. To import from a different package, use pkg:mypackage. It looks in a directory of the package named staticexports to find the objects. To import from a local absolute or relative path, you can use something like @staticimports ../r_utils/.

The following lines name the objects to import from the source. In this case, they are os_name, %||%, map, and walk.

To perform the import, run:

library(staticimports)
import()

By default this will write the functions to a file R/staticimports.R in your project.

You examine the output by writing to stdout() instead of R/staticimports.R. Notice how importing os_name automatically brings in is_windows, is_mac, and is_linux.

import(outfile = stdout())
#> # Generated by staticimports; do not edit by hand.
#> # ======================================================================
#> 
#> `%||%` <- function(a, b) {
#>   if (is.null(a)) b else a
#> }
#> 
#> is_linux   <- function() Sys.info()[['sysname']] == 'Linux'
#> 
#> is_mac     <- function() Sys.info()[['sysname']] == 'Darwin'
#> 
#> is_windows <- function() .Platform$OS.type == "windows"
#> 
#> map <- function(.x, .f, ...) {
#>   lapply(.x, .f, ...)
#> }
#> 
#> os_name <- function() {
#>   if (is_windows()) {
#>     "win"
#>   } else if (is_mac()) {
#>     "mac"
#>   } else if (is_linux()) {
#>     "linux"
#>   } else if (.Platform$OS.type == "unix") {
#>     "unix"
#>   } else {
#>     "unknown"
#>   }
#> }
#> 
#> walk <- function(.x, .f, ...) {
#>   for (i in seq_along(.x)) {
#>     .f(.x[[i]], ...)
#>   }
#>   NULL
#> }

For testing what the output will look like, you can use import_objs() to see what it looks like when you import specific objects by name:

import_objs(c("map", "walk"), outfile = stdout())
#> # Generated by staticimports; do not edit by hand.
#> # ======================================================================
#> 
#> map <- function(.x, .f, ...) {
#>   lapply(.x, .f, ...)
#> }
#> 
#> walk <- function(.x, .f, ...) {
#>   for (i in seq_along(.x)) {
#>     .f(.x[[i]], ...)
#>   }
#>   NULL
#> }

Under the hood

The functions provided by staticimports are in the inst/staticexports directory of the repository. When import() is called, it sources all of those files into a new environment. Then it finds all the internal dependencies among those functions. If a function fn_a is requested, and it uses a function fn_b, then both of those functions will be copied to the project.

If a different set of source files is used, it sources all of the files in the target directory into a new environment and then proceeds the same way.

TODO

  • Figure out and explain licensing issues.
  • Write license info in generated file.

About

Statically import R functions into packages and projects

Resources

License

Unknown, MIT licenses found

Licenses found

Unknown
LICENSE
MIT
LICENSE.md

Stars

Watchers

Forks

Releases

No releases published

Packages