Reduce NMBObjCMatcher usages

[ikesyo]

See also #592.

[ikesyo]

nil should not be converted to false here.

[sharplet]

Oh, good catch. Should we add a test that covers this difference between beFalse() and beFalsy()? We're also accepting values that fail to cast to NSNumber, should we also return nil if the cast fails?

[ikesyo]

That is already covered here:

• Nimble/Tests/NimbleTests/objc/ObjCBeFalseTest.m

  Lines 24 to 29 in 7de0814

  	expectNilFailureMessage(@"expected to be false, got <nil>", ^{
  	expect(nil).to(beFalse());
  	});
  	expectNilFailureMessage(@"expected to not be false, got <nil>", ^{
  	expect(nil).toNot(beFalse());
  	});

• Nimble/Tests/NimbleTests/objc/ObjCBeFalsyTest.m

  Line 13 in 7de0814

  expect(nil).to(beFalsy());

• Nimble/Tests/NimbleTests/objc/ObjCBeFalsyTest.m

  Lines 26 to 28 in 7de0814

  	expectFailureMessage(@"expected to not be falsy, got <nil>", ^{
  	expect(nil).toNot(beFalsy());
  	});

Previously the behavior is ensured by NMBObjCMatcher(canMatchNil: false):

Nimble/Sources/Nimble/Adapters/NMBObjCMatcher.swift

Lines 47 to 52 in 7de0814

	if !canMatchNil {
	if try actualExpression.evaluate() == nil {
	failureMessage.postfixActual = " (use beNil() to match nils)"
	return false
	}
	}

[ikesyo]

What I wanted to say is that the message

nil should not be converted to false here.

is only applicable to the new NMBPredicate impl in this PR, but not to the previous NMBObjCMatcher impl.

[ikesyo]

Testing contain matcher in ObjCEndWithTest.m is odd (and the fact the test is passing is odd as well) 😛

[sharplet]

One comment, but lgtm 👍

[sharplet]

Oh, good catch. Should we add a test that covers this difference between beFalse() and beFalsy()? We're also accepting values that fail to cast to NSNumber, should we also return nil if the cast fails?

[ikesyo]

I will go ahead 🚀

Thanks for the review @sharplet!