rtest files

[genedan]

Can anyone shed light on what the rtest files do? They seem to test for parity between the Python and R versions of chainladder, but they aren't run during testing, and they also negatively impact the test coverage metric. For example, see 18 missed lines for rtest_correlation.py:

https://app.codecov.io/github/casact/chainladder-python/tree/main/chainladder%2Fcore%2Ftests

• Should we exclude the rtest files from coverage?
• Should we find a way to execute them?
• Should we exclude test files in general in the coverage metric?

I'm leaning towards including test files in coverage, but doing something about the rtest files so that they're either run or stashed away. Reason being, I wouldn't have found out about them (and that they weren't being run) if they weren't included in the coverage metric:

https://softwareengineering.stackexchange.com/questions/309015/when-does-it-make-sense-to-include-test-code-in-coverage

[henrydingliu]

i'm assuming john has set it up locally to run these through rpy2.

we can turn these into static tests, i.e. someone runs the R code locally and hardcode the values into these tests. we can even keep the R code as comments for complete documentation

[genedan]

Sounds like a plan, I'll create an issue.