Refactoring with sequenced expectations

[sjgriffin7]

I'm having some issues with refactoring sequenced calls. (Note: I'm using Catch2 as the unit test framework, and currently using C++11)

Ideally, I'd be able to take a test like this:

TEST_CASE("Refactoring","[refactoring]")
{
    Obj* obj = new Obj(); // Assume Obj::MethodA() and Obj::MethodB() are defined

    SECTION("Original")
    {
        // - - - Expectation - - - //

        trompeloeil::sequence seq;

        REQUIRE_CALL_V(  obj, MethodA(),
            .IN_SEQUENCE( seq )
        );

        REQUIRE_CALL_V(  obj, MethodB(),
            .IN_SEQUENCE( seq )
        );

        // - - - Execution - - - //

        obj->MethodA();
        obj->MethodB();
    }

}

And add a helper function to reuse REQUIRE_CALL_V() statements:

void REQUIRE_MethodB(Obj* obj_, trompeloeil::sequence &seq)
{
    REQUIRE_CALL_V( obj_, MethodB(), .IN_SEQUENCE( seq ) );
}

TEST_CASE("Refactoring","[refactoring]")
{
    Obj* obj = new Obj(); // Assume Obj::MethodA() and Obj::MethodB() are defined

    SECTION("Original")
    {
        // - - - Expectation - - - //

        trompeloeil::sequence seq;

        REQUIRE_CALL_V(  obj, MethodA(),
            .IN_SEQUENCE( seq )
        );

        REQUIRE_CALL_V(  obj, MethodB(),
            .IN_SEQUENCE( seq )
        );

        // - - - Execution - - - //

        obj->MethodA();
        obj->MethodB();
    }

    SECTION("Refactored")
    {
        // - - - Expectation - - - //

        trompeloeil::sequence seq;

        REQUIRE_CALL_V(  obj, MethodA(),
            .IN_SEQUENCE( seq )
        );

        REQUIRE_MethodB(obj, seq);

        // - - - Execution - - - //

        obj->MethodA();
        obj->MethodB();
    }

}

My issue arises in the SECTION("Refactored") results, which reports unfulfilled expectations for MethodA() and MethodB().

Is this something that should work, or are there other recommendations for refactoring? Any details on the inner workings of trompeloeil::sequence are also welcome.

[rollbear]

REQUIRE_CALL() and its siblings are valid until the end of the scope, so your function REQUIRE_MethodB() sets a requirement for a call, that immediately fails because the requirement is checked at the end of the function. I don't expect the sequence to be of importance here.

See NAMED_REQUIRE_CALL.

Example: https://godbolt.org/z/fY6v39vKa

[sjgriffin7]

Excellent, that was the piece of info I was missing. Thanks!