Checking package with python dependencies on CRAN

lorenzwalthert · November 2, 2019, 1:35pm

I want to submit my package precommit to CRAN, and it has unit tests in that install packages with conda (via reticulate) and use command line executables via system2. Is there a way to install cona on CRAN machines? Or do I have to skip the test? There are packages that have similar issues like keras, for which unit tests are disabled on CRAN:

github.com

rstudio/keras/blob/master/tests/testthat.R

library(testthat)

if (Sys.getenv("TENSORFLOW_EAGER") == "TRUE")
  tensorflow::tfe_enable_eager_execution()

library(keras)

if (identical(Sys.getenv("NOT_CRAN"), "true")) {
  test_check("keras") 
}

jdblischak · November 4, 2019, 5:01pm

I don't recommend trying to install conda on the CRAN machines. They will not like that. Do you have tests that can be run that don't require conda/reticulate/system2()? If yes, I recommend only skipping specific tests using testthat::skip_on_cran().

system · May 15, 2020, 8:00am

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.