fix call-calc-coverage workflow

* update the DESCRIPTION file and set Config/testthat/parallel to false.
  This addresses the GHA run-time issue.
* skip parallel tests when running covr::package_coverage(), as these tests
  don't contribute to increasing code coverage and are redundant due to
  the reuse of wrapper functions. Integration tests are sufficient for
  coverage calculation.
* update GHA build triggers and remove the paths-ignore section. In the
  future, if workflow run times become too long, consider setting up a
  paths filter. This would limit the workflow to run only when changes
  are made to specific folders, such as inst/include, src, or R, rather
  than listing all non-code files.

Co-authored-by: Kathryn Doering <[email protected]>

.github/workflows/call-calc-coverage.yml

@@ -4,15 +4,13 @@ name: call-calc_coverage
 # The default is to run the workflow on every push or pull request to every branch.
 on:
   workflow_dispatch:
-    paths-ignore:
-      - .devcontainer
-      - .github
-      - 'LICENSE'
-      - 'README.md'
-      - 'CONTRIBUTING.md'
-      - 'Rbuildignore'
-      - '.gitignore'
-      - 'man'
+  push:
+    branches:
+      - dev  
+  pull_request:
+    branches:
+      - main
+      - dev  
 jobs:
   call-workflow:
     uses: nmfs-fish-tools/ghactions4r/.github/workflows/calc-coverage.yml@main

DESCRIPTION

@@ -89,7 +89,7 @@ LinkingTo:
 VignetteBuilder:
     knitr
 Config/testthat/edition: 3
-Config/testthat/parallel: true
+Config/testthat/parallel: false
 Encoding: UTF-8
 LazyData: true
 NeedsCompilation: yes

tests/testthat/test-parallel-with-snowfall-with-wrappers.R

@@ -8,6 +8,7 @@
 # modes. The parallel execution uses {snowfall} to parallelize the tasks across
 # multiple CPU cores.
 
+testthat::skip_on_covr()
 # Load the model comparison operating model data from the fixtures folder
 load(test_path("fixtures", "integration_test_data.RData"))