test: some suggestions for making failing test file more readable

[maelle]

👋 @quishqa!

I don't know yet why the Windows build is failing (I am on Ubuntu and can't run the tests locally yet see #9) however I think this test file can be made more informative when it fails by using more specific expectations. Some of them might be new to you, I think going over https://testthat.r-lib.org/reference/index.html once in a while is useful as new expectations appear over time (it's useful for me at least!).

Anyway as a summary:

• no need to explicitly load the internal data.
• I had trouble understanding the failure param_means["ws"] > 1 (actual) not equal to param_means["ws"] < 2 (expected)., using expect_lt() and expect_gt() instead will give a clearer failure.
• I'd recommend moving the expectations around the class of pinheiros and its columns before the calculations.
• These expectations around classes should use https://testthat.r-lib.org/reference/inheritance-expectations.html although for the columns I have added a comment, I think it'd make sense to create the vector of all class names then use expect_setequal() https://testthat.r-lib.org/reference/expect_setequal.html (it does not check the order of columns).

[maelle]

Now it's failing everywhere, clearly I did something wrong. Happy to run the tests locally once I have more instructions 🙈

[maelle]

@quishqa friendly reminder 😸

[quishqa]

Hi @maelle!
Sorry for the late answer, I'm back to maintaining qualR 🦾 .

• By looking the commits, I think the errors are happening by removing the internal data which contains an example credential to test the functions (probably not a good practice).
• I agree using expect_lt, expect_gt, expect_<class> functions, it produce a cleaner code.
  I believe I do need to read again testthat documentation 😞

[maelle]

Don't hesitate to ping me to ask any question!

[maelle]

@quishqa any help needed? 😸

[maelle]

@quishqa any update or help needed? 😸

[quishqa]

Hi @maelle, sorry for the delay. I think the error happens because in 5ca83eb the deleted file has the credentials to log in and download the data...

[maelle]

Do you need to cache/mock some stuff?

[maelle]

@quishqa any update / any help needed?

[maelle]

Current results on https://ropensci.r-universe.dev/builds

  Running the tests in ‘tests/testthat.R’ failed.
  Last 13 lines of output:
    Differences:
    `actual`:   FALSE
    `expected`: TRUE 
    
    ── Failure ('test-monitor_ar_retrieve_pol.R:15:3'): monitor_ar_retrieve_pol works! ──
    Expected `pol_means["pm10"] > 22` to equal `pol_means["pm10"] < 23`.
    Differences:
    `actual`:   FALSE
    `expected`: TRUE 
    
    
    [ FAIL 4 | WARN 0 | SKIP 0 | PASS 129 ]
    Error:
    ! Test failures.
    Execution halted

Any help needed?

[quishqa]

Hi @maelle, Sorry for the delay. I solved the test error in my last PR, and also do a lot of the proposed suggestions. All test passed except for macOS...do you have any ideia?

[maelle]

I see the error involves the XML package. That package is no longer maintained (it's just kept alive by the CRAN team) so it might make sense to switch to xml2. I don't promise it'd solve the error but it's a good thing to do in any case 😸

[maelle]

Happy to help with xml2 if you have any question!