Use of processR in FIMS/tests/concurrent

[kellijohnson-NOAA]

I was looking at updating the DESCRIPTION file to update some of the packages that are included. One of them is {processR}. This package is being used in tests/concurrent but it is not listed in the DESCRIPTION file. I was going to add it to Suggests but it is currently not available on CRAN, where it was taken off 2023-02-02 because it requires archived {predict3d}. Is there a different package that we can use instead of {processR}? Because this is in tests/concurrent should I just not worry about adding it to the DESCRIPTION file in suggests? Thanks for the discussion about this 😉. @Bai-Li-NOAA

[Bai-Li-NOAA]

@kellijohnson-NOAA, no need to add {processR} to the DESCRIPTION file since it's not included in the {testthat} framework or run in any GHA checks. Tagging @msupernaw to confirm the availability of the {processR}, pretty sure it's a package he developed.

I’d love to hear more about the goals of these files from @msupernaw and where you all think we should store the concurrent files. To me, they could be turned into a vignette if the idea is to share comparison results with developers or users. Right now, they’re not in any GHA checks since they aren’t written using {testthat}, and we often forget to update them with the latest {FIMS}.

P.S. We do have one {testthat} test that checks if FIMS runs in parallel with {snowfall}, so that’s covered!

[kellijohnson-NOAA]

@Bai-Li-NOAA can you please edit this comment to indicate which of the following directories/files inside of "tests" are used in testing, i.e., are ran with check() or in GitHub actions?

1. "compilation_tests"
2. "concurrent"
3. "google_benchmark"
4. "gtest"✅
5. "integration"
6. "test_plan"
7. "testthat" ✅
8. "testthat.R" ✅

[Bai-Li-NOAA]

updated the comment. We can check with the team to see if anyone wants to keep the files that aren't being used in GHA testing. For example:

• I could move /google_benchmark to the collaboration workflow book, and we can bring it back to FIMS when we're ready to write some benchmark tests.
• I can archive /test_plan and start using {testdown} to auto-generate a higher-level test report.

[kellijohnson-NOAA]

{processR} is not actually on CRAN, instead another package called {processR} is archived on CRAN.

• @msupernaw is going to rename the package in nmfs-fish-tools

Second, {processR} is beneficial to think about using in FIMS in the future because

• {processR} does not require openMPI, which is a third-party requirement that snowfall and other R packages require.
• {processR} has shared memory to allow read and write at the same time making it where you do not have to transfer data back and forth