Skip to content

Latest commit

 

History

History
201 lines (144 loc) · 8.07 KB

assertions.md

File metadata and controls

201 lines (144 loc) · 8.07 KB
Raw

Assertion Macros

Contents
Natural Expressions
Exceptions
Matcher expressions
Thread Safety
Expressions with commas

Most test frameworks have a large collection of assertion macros to capture all possible conditional forms (_EQUALS, _NOTEQUALS, _GREATER_THAN etc).

Catch is different. Because it decomposes natural C-style conditional expressions most of these forms are reduced to one or two that you will use all the time. That said there are a rich set of auxiliary macros as well. We'll describe all of these here.

Most of these macros come in two forms:

Natural Expressions

The REQUIRE family of macros tests an expression and aborts the test case if it fails. The CHECK family are equivalent but execution continues in the same test case even if the assertion fails. This is useful if you have a series of essentially orthogonal assertions and it is useful to see all the results rather than stopping at the first failure.

  • REQUIRE( expression ) and
  • CHECK( expression )

Evaluates the expression and records the result. If an exception is thrown, it is caught, reported, and counted as a failure. These are the macros you will use most of the time.

Examples:

CHECK( str == "string value" );
CHECK( thisReturnsTrue() );
REQUIRE( i == 42 );
  • REQUIRE_FALSE( expression ) and
  • CHECK_FALSE( expression )

Evaluates the expression and records the logical NOT of the result. If an exception is thrown it is caught, reported, and counted as a failure. (these forms exist as a workaround for the fact that ! prefixed expressions cannot be decomposed).

Example:

REQUIRE_FALSE( thisReturnsFalse() );

Do note that "overly complex" expressions cannot be decomposed and thus will not compile. This is done partly for practical reasons (to keep the underlying expression template machinery to minimum) and partly for philosophical reasons (assertions should be simple and deterministic).

Examples:

  • CHECK(a == 1 && b == 2); This expression is too complex because of the && operator. If you want to check that 2 or more properties hold, you can either put the expression into parenthesis, which stops decomposition from working, or you need to decompose the expression into two assertions: CHECK( a == 1 ); CHECK( b == 2);
  • CHECK( a == 2 || b == 1 ); This expression is too complex because of the || operator. If you want to check that one of several properties hold, you can put the expression into parenthesis (unlike with &&, expression decomposition into several CHECKs is not possible).

Floating point comparisons

When comparing floating point numbers - especially if at least one of them has been computed - great care must be taken to allow for rounding errors and inexact representations.

Catch provides a way to perform tolerant comparisons of floating point values through use of a wrapper class called Approx. Approx can be used on either side of a comparison expression. It overloads the comparisons operators to take a tolerance into account. Here's a simple example:

REQUIRE( performComputation() == Approx( 2.1 ) );

Catch also provides a UDL for Approx; _a. It resides in the Catch::literals namespace and can be used like so:

using namespace Catch::literals;
REQUIRE( performComputation() == 2.1_a );

Approx is constructed with defaults that should cover most simple cases. For the more complex cases, Approx provides 3 customization points:

  • epsilon - epsilon serves to set the percentage by which a result can differ from Approx's value before it is rejected. By default set to std::numeric_limits<float>::epsilon()*100.
  • margin - margin serves to set the the absolute value by which a result can differ from Approx's value before it is rejected. By default set to 0.0.
  • scale - scale is used to change the magnitude of Approx for relative check. By default set to 0.0.

epsilon example

Approx target = Approx(100).epsilon(0.01);
100.0 == target; // Obviously true
200.0 == target; // Obviously still false
100.5 == target; // True, because we set target to allow up to 1% difference

margin example

Approx target = Approx(100).margin(5);
100.0 == target; // Obviously true
200.0 == target; // Obviously still false
104.0 == target; // True, because we set target to allow absolute difference of at most 5

scale

Scale can be useful if the computation leading to the result worked on different scale than is used by the results. Since allowed difference between Approx's value and compared value is based primarily on Approx's value (the allowed difference is computed as (Approx::scale + Approx::value) * epsilon), the resulting comparison could need rescaling to be correct.

Exceptions

  • REQUIRE_NOTHROW( expression ) and
  • CHECK_NOTHROW( expression )

Expects that no exception is thrown during evaluation of the expression.

  • REQUIRE_THROWS( expression ) and
  • CHECK_THROWS( expression )

Expects that an exception (of any type) is be thrown during evaluation of the expression.

  • REQUIRE_THROWS_AS( expression, exception type ) and
  • CHECK_THROWS_AS( expression, exception type )

Expects that an exception of the specified type is thrown during evaluation of the expression. Note that the exception type is extended with const& and you should not include it yourself.

  • REQUIRE_THROWS_WITH( expression, string or string matcher ) and
  • CHECK_THROWS_WITH( expression, string or string matcher )

Expects that an exception is thrown that, when converted to a string, matches the string or string matcher provided (see next section for Matchers).

e.g.

REQUIRE_THROWS_WITH( openThePodBayDoors(), Contains( "afraid" ) && Contains( "can't do that" ) );
REQUIRE_THROWS_WITH( dismantleHal(), "My mind is going" );
  • REQUIRE_THROWS_MATCHES( expression, exception type, matcher for given exception type ) and
  • CHECK_THROWS_MATCHES( expression, exception type, matcher for given exception type )

Expects that exception of exception type is thrown and it matches provided matcher (see next section for Matchers).

Please note that the THROW family of assertions expects to be passed a single expression, not a statement or series of statements. If you want to check a more complicated sequence of operations, you can use a C++11 lambda function.

REQUIRE_NOTHROW([&](){
    int i = 1;
    int j = 2;
    auto k = i + j;
    if (k == 3) {
        throw 1;
    }
}());

Matcher expressions

To support Matchers a slightly different form is used. Matchers have their own documentation.

  • REQUIRE_THAT( lhs, matcher expression ) and
  • CHECK_THAT( lhs, matcher expression )

Matchers can be composed using &&, || and ! operators.

Thread Safety

Currently assertions in Catch are not thread safe. For more details, along with workarounds, see the section on the limitations page.

Expressions with commas

Because the preprocessor parses code using different rules than the compiler, multiple-argument assertions (e.g. REQUIRE_THROWS_AS) have problems with commas inside the provided expressions. As an example REQUIRE_THROWS_AS(std::pair<int, int>(1, 2), std::invalid_argument); will fail to compile, because the preprocessor sees 3 arguments provided, but the macro accepts only 2. There are two possible workarounds.

  1. Use typedef:
using int_pair = std::pair<int, int>;
REQUIRE_THROWS_AS(int_pair(1, 2), std::invalid_argument);

This solution is always applicable, but makes the meaning of the code less clear.

  1. Parenthesize the expression:
TEST_CASE_METHOD((Fixture<int, int>), "foo", "[bar]") {
    SUCCEED();
}

This solution is not always applicable, because it might require extra changes on the Catch's side to work.

Home