deactivate froll tests

[ben-schwen]

Towards #6400

Current froll checking times would not pass CRAN.

Change is structured in such a way that releases do not run froll.Rraw. One can still force the execution by setting Sys.setenv("RunAllDataTableTests"="yes")

What still needs to be discussed is whether we want to disable main test suite on CRAN...

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 99.07%. Comparing base (df7fa80) to head (d6c1411).
⚠️ Report is 8 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #7429      +/-   ##
==========================================
- Coverage   99.13%   99.07%   -0.07%     
==========================================
  Files          85       85              
  Lines       16618    16609       -9     
==========================================
- Hits        16474    16455      -19     
- Misses        144      154      +10     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[jangorecki]

Why disable the whole test script. I think that all non-batch tests should finish under 10s and they are very useful, as show by recently detected breaking changes detected by revdeps check.

[jangorecki]

Marking as request change ls because we want to have at least the basic tests run each time. Only batch testing are the issue here. So either a branching them based on env car, or interactive(), or moving to another test script seems much better way to address the issue.

[jangorecki]

Thank you for enabling tests back.
As for env bars, we, and base R always uses capital case with underscores. Any good reason to use camel case instead?

[ben-schwen]

Thank you for enabling tests back. As for env bars, we, and base R always uses capital case with underscores. Any good reason to use camel case instead?

Nah that was just shamelessly copied from Dirk.

Should we maybe run batch tests only in Gitlab and RMD-check-occasional?

[github-actions]

• HEAD=deactivate_tests_on_cran slower P<0.001 for memrecycle regression fixed in #5463
• HEAD=deactivate_tests_on_cran slower P<0.001 for setDT improved in #5427
  

Generated via commit d6c1411

Download link for the artifact containing the test results: ↓ atime-results.zip

Task	Duration
R setup and installing dependencies	4 minutes and 54 seconds
Installing different package versions	10 minutes and 25 seconds
Running and plotting the test cases	2 minutes and 58 seconds

[jangorecki]

I would run them as well for interactive session. Maybe we can detect interactive inside test.data.table() and set env var there?
The purpose is to avoid it on CRAN only. We could avoid it as well in --as-cran CI jobs, so we have a workflow that can be checked to confirm if the mechanism works as expected.

[ben-schwen]

I would run them as well for interactive session. Maybe we can detect interactive inside test.data.table() and set env var there? The purpose is to avoid it on CRAN only.

We could but if you run them interactively you gonna type test.data.table(script="frollBatch.Rraw") anyway, hence leaving the optional to its default?