Argument 'local' of Future() to become defunct [breaks your package]

[HenrikBengtsson]

Background

Argument local for future::Future() is currently defunct for local = FALSE. This argument will soon become completely defunct, i.e. if it is specified, an error is produced.

Issue

R CMD check on your package fails when the above change is effective. This is because, in your CivisFuture(), you are currently passing the default local = TRUE to future::Future();

civis-r/R/civis_future.R

Lines 84 to 99 in 51c2e22

	future <- future::Future(expr = expr,
	envir = env,
	substitute = substitute,
	globals = gp$globals,
	packages = unique(c(packages, gp$packages)),
	lazy = lazy,
	local = local,
	gc = gc,
	earlySignal = earlySignal,
	label = label,
	version = "1.8", # see: https://github.com/civisanalytics/civis-r/issues/168
	# extra args to scripts_post_containers
	required_resources = required_resources,
	docker_image_name = docker_image_name,
	docker_image_tag = docker_image_tag, ...)

Action

Please drop this local argument everywhere. It already now has no effect.

[HenrikBengtsson]

Here is how your package tests will fail:

• checking tests ...

    Running ‘testthat.R’
   ERROR
  Running the tests in ‘tests/testthat.R’ failed.
  Last 50 lines of output:
     1. └─civis::CivisFuture(quote(2 + 3)) at test_civis_future.R:24:2
     2.   └─future::Future(...)
     3.     └─base (local) dfcn(...)
    ── Error ('test_civis_future.R:61'): run and value work ────────────────────────
    <defunctError/error/condition>
    Error: Argument 'local' is defunct as of future 1.31.0 (2023-??-??)
  ...
    Error: Argument 'local' is defunct as of future 1.31.0 (2023-??-??)
    Backtrace:
        ▆
     1. └─civis::CivisFuture(quote(2 + 2)) at test_civis_future.R:118:2
     2.   └─future::Future(...)
     3.     └─base (local) dfcn(...)
    
    [ FAIL 8 | WARN 4 | SKIP 0 | PASS 1041 ]
    Error: Test failures
    Execution halted
  

Source: https://github.com/HenrikBengtsson/future/blob/d70df1c2e18b21a64e11aa341172e9c4e7b5038f/revdep/problems.md?plain=1#L339-L378

[HenrikBengtsson]

To reproduce the above check error with the latest future 1.31.0 on CRAN, set:

export R_FUTURE_CHECK_IGNORE_CIVIS=false

before running R CMD check.

[pcooman]

I submitted an updated version of the civis package (v3.1.0) to CRAN for approval. In this new version, the "local" argument to the CivisFuture() function has been deprecated, and we are no longer passing it on the the future::Future() function. To confirm, I ran R CMD CHECK with R_FUTURE_CHECK_IGNORE_CIVIS set to false and all checks passed.

The new submission passed CRAN's automated checks and is now waiting for manual review, which should be completed within ~5 days. We should be ready on our end for you to make the proposed changes to the future package.

Thank you again for bringing this to our attention and for your patience!

[HenrikBengtsson]

It passes my revdep checks. Thanks.