[R-pkg-devel] How to create a macos-arm64 virtual machine/cloud environment to debug R package?
Sorry, I now see that this may actually be a timezone specific issue which I can now reproduce in Windows by changing the timezone of the timeseries to "Pacific/Auckland". This would indicate that macos-arm64 CRAN machines are based in a different timezone than the other machines, not the arm64 OS as such. This reminds me that I ran into a similar issues earlier this year where I struggled to get GitHub Actions to run the macos workflow in a non-default timezone, which I raised elsewhere: https://community.rstudio.com/t/upgrading-github-action-workflow-to-address-no-package-called-rcmdcheck-but-unclear-how-to-account-for-timezone/166876 ... I should have thought about that first. Vincent On Wed, 6 Dec 2023 at 10:28, Vincent van Hees <vincentvanhees at gmail.com> wrote:
Hello, What is the recommended way to create a virtual machine or cloud based R environment for debugging R package issues specific to OS macos-arm64? Details: CRAN results show macos-arm64 specific problems for my package: https://cran.r-project.org/web/checks/check_results_GGIR.html. I am able to reproduce them with https://mac.r-project.org/macbuilder/submit.html. I suspect something goes wrong with the timestamp creation or handling as that has caused issues with this OS for me in the past. All this relates to this <https://github.com/wadpac/GGIR/blob/master/tests/testthat/test_part5_analyseRest.R> test for this <https://github.com/wadpac/GGIR/blob/master/R/g.part5.analyseRest.R> function, that in the process calls this <https://github.com/wadpac/GGIR/blob/master/R/POSIXtime2iso8601.R> short function. I have been trying some edits locally and re-submitting them to macbuilder a couple of times to no avail. To debug this efficiently it would be good to have access to an R environment for this OS. I do not have a physical macos myself. Vincent