test() failure for missed expected warning not ideal

[MichaelChirico]

Observed in #7091

data.table:::test(1, {warning('a'); 2}, 2, warning=c('a', 'b'))
# Test 1 produced 1 warnings but expected 2
# Expected: a
# Observed: a
#  Test 1 produced 1 warnings but expected 2
# Expected: b
# Observed: a

In isolation, it's a bit easier to grok, but in the full test log with potentially more errors, it's harder to mentally "mod out" the first one that does match. Ideally, we can simply not produce any warnings that match as intended, or otherwise signal that there's nothing incorrect with the first case] (i.e., you're not going crazy).

[ben-schwen]

Test 1 produced 1 warnings but expected 2
Expected: a
          b
Observed: a

Is this good enough or should we start with matching them?

[MichaelChirico]

That looks like a good first step yes

[lucaslarson25]

Hi! I'm a student from Igor Steinmacher's open source software course.
I'd like to work on this issue with my team, if that's alright. We will start looking at the testing code and documents, but is there any other information that might be critical for this issue?

Thanks!

[ben-schwen]

Hey Lucas. I assigned your group the issue. The problem AFAIR is that we loop through all warnings and produce the output in the first post instead of cleanly gathering the information first. As Michael also pointed we are actually only interested in the setdiff of expected\observed or observed\expected.

A good starting point would be to make yourself comfortable with our cc() function in .dev since it will make working on this issue a lot easier and smoother since you can fast compile and test the code.