Parallel styling of files by style_pkg & style_dir (#277)

[RichardJActon]

Implements parallel styling of files by the style_pkg & style_dir functions addressing issue #277 using the future and future.apply packages.

Example of how a user would get parallel styling:

library(future)
plan(multisession, workers = 4)
style_pkg()

[lorenzwalthert]

Thanks @RichardJActon. A few minor things. Can you also add a test or two? I don't know how many core the CI machines have, but you can detect them with parallel::detectCores() (to avoid importing the future package explicitly, see also https://stackoverflow.com/questions/28954991/whether-to-use-the-detectcores-function-in-r-to-specify-the-number-of-cores-for). Since the APIs messages are different, you could implement one for style_file(). Best add them in styler/tests/testthat/test-public_api.R and make sure to first capture active future strategy and restore it afterwards.

[lorenzwalthert]

This should actually go under a new # styler 1.3.2.9000 header. Also, please reference the PR, not the issue. I also believe that style_file() will also benefit (since it is vectorized over file). Can you adapt this as well?

[lorenzwalthert]

Can we make sure the new argument root is also documented? You can @inheritParams here from transform_file and document it there. You can also add a sentence like "required because working directory set with withr::with_dir() not respected by future.apply::future_sapply().

[lorenzwalthert]

also @inheritParams from `transform_files() to get path documented.

[lorenzwalthert]

Please also use root as argument name and then @inheritParams from transform_files

[lorenzwalthert]

For the example with librar(future), we'd also need to declare future as an import explicitly (although it will be available anyways since we import future.apply)

[RichardJActon]

I'm working on some tests. I'm not quite sure what you mean by 'make sure to first capture active future strategy and restore it afterwards' could you clarify?

[lorenzwalthert]

Withint a testthat test

test_that(..., {

old_plan <- future::plan()
on.exit(future::plan(old_plan)
# you test (including setting a plan)
})

Not sure it works as expected but try 😄

[lorenzwalthert]

Any progress on that @RichardJActon. I can fix the merge conflicts after you're done.

[lorenzwalthert]

Closing due to inactivity and new developments in #682.