Lab report showing incorrect information

[ferrao]

Lab report tells me that a condition in my code will always evaluate to true, but I believe this is not correct, bellow some sample code to validate.

lib/index.js

module.exports = function(response, options = {}) {
    const { key } = options;

    if (!Array.isArray(response) && !key) {
        throw Error('missing key')
    }

    if (key && !response[key]) {
        throw Error('wrong key')
    }
    
    // if (key) {
    if (key && !Array.isArray(response)) {
        return 1;
    }

    return 0;
}

test/test.js

const test = require("../lib");

describe("Testing lab coverage", () => {
  it("1 - valid object and key", () => {
    expect(test({ someKey: "xpto" }, { key: "someKey" })).to.equal(1);
  });

  it("2 - valid object and no key", () => {
    expect(() => test({ someKey: "xpto" })).to.throw();
  });

  it("3 - array and no key", () => {
    expect(test([])).to.equal(0);
  });

  it("4 - array and key", () => {
    expect(() => test([], { key: "someKey" })).to.throw();
  });

  it('5 - array and numeric key', () => {
       expect(test([1,2], { key: 1 })).to.equal(0);
  });
});

If I run the following test code through lab with ./node_modules/.bin/lab test -v -t 100 I am getting a missing coverage on line if (key && !Array.isArray(response)) { and if I generate a report with ./node_modules/.bin/lab test -v -t 100 -r html I get on the same line an indication regarding !Array.isArray(response) - Condition always true.

As is, I can in fact remove the right side of the boolean expression and all tests will still pass.

If I uncomment the last test in the suite I now get 100% coverage and no indication regarding the condition.

But with this last test, I can not remove the right side of the boolean expression, as this last test will now fail, making my point that such condition is NOT always true.

[ferrao]

This old case #401 is probably related?

[Marsup]

I'm not sure I follow you, without your last test of course it's missing coverage. Either your function or your tests logic is flawed, but everything seems normal to me.

[ferrao]

What I am claiming is that without the last test, lab reports Condition always true which is wrong. If condition was always true as lab claims, I could remove the redundant condition from my code.

But I can not, because if I do it will make that last test fail.

[Marsup]

But it is always true. There is never a case where you reach that if with a key and an array. Your test 3 has an array but no key.

[ferrao]

Forget about it... I was trying to come up with a more simple example that the actual code to explain what was happening, but I failed miserably at my example and I believe you are right that lab is doing the right thing (with my actual code and not the example I mean)....

[Marsup]

To explain further, lab checks logical coverage.
So with if (a && b) the possible scenarios are:

a	b
true	true
true	false
false	not important

You have case 1 and 3 but you're clearly missing the 2nd.

[ferrao]

Yes, I understand and it is how I expected coverage to work, but let me make another try at explaining my point, I have renamed the tests for better clarity.

What seems wrong to me is not the coverage at all, but the Condition always true message I am getting on the coverage report, but perhaps I am misunderstanding the meaning of it.

Lets say I have code a with condition if (key && !Array.isArray(response)) and code b with condition if (key).

Running the exact same 5 tests above on code a and code b produces different results. It passes all tests with b and fails test 5 with a . So these two pieces of code can not be interchangeable.

That said, if I simply comment test 5, the coverage report now informs me that code a contains a condition which is always true, which leads me to think I can replace code a with code b, which I know that I can not as I concluded they are not the same on the previous step.

Anyway, thank you for your time on this @Marsup, it is by no means a big issue to me, more just a curiosity, so if you believe that my point is not valid let's just close the issue.

[Marsup]

Well, when it is reached, it's always true, when it's not the value has no point since it virtually doesn't exist. In terms of coverage, I think it's harder to differentiate between a value that's reached and false, and not reached, we could have an execution counter for each elements of conditionals but I'm not sure it would make it clearer.

[lock]

This thread has been automatically locked due to inactivity. Please open a new issue for related bugs or questions following the new issue template instructions.