Conversation
aa10eb8 to
77719cd
Compare
It was calling into decNumberExp with the wrong signature, so it didn't work at all. I also fixed the decNumber tests for exp so that they'll actually run if point at the exp.decTest file, and I added a super basic test for ln and exp. It seems a bit too easy to not run some of the decNumber test by accident right now; maybe it would be good to expect every test file to have some non-igored tests, and then just delete the test files for actually unimplemented functionality?
77719cd to
d2efc74
Compare
| Precision: 34 | ||
| maxExponent: 6144 | ||
| minExponent: -6143 | ||
| clamp: 1 |
There was a problem hiding this comment.
Why do we need this? This test should not change at all.
There was a problem hiding this comment.
Because of these lines in the run-test script:
let d128_env = Environment {
precision: Some(34),
rounding: Some(Rounding::HalfEven),
max_exponent: Some(6144),
min_exponent: Some(-6143),
extended: true,
clamp: true,
};
if *env != d128_env {
return TestResult::Ignored(test);
}Without that change, the tests were silently skipped (and none of the other matched either).
Maybe I should have gone in the other direction and defined a different Environment just for these tests? I assume that the check is there to make sure that it only runs applicable tests, but since the exp tests pass with the different clamping mode, I assume that all of those test cases are still valid with the different clamping mode.
There was a problem hiding this comment.
Is it possible to change run_test() to run ops with the specified environment which will make this change unnecessary and increase test coverage?
It was calling into decNumberExp with the wrong signature, so it didn't work at all.
I also fixed the decNumber tests for exp so that they'll actually run if point at the exp.decTest file, and I added a super basic test for ln and exp.
It seems a bit too easy to not run some of the decNumber test by accident right now; maybe it would be good to expect every test file to have some non-igored tests, and then just delete the test files for actually unimplemented functionality?