<html lang="en">
<head>
<title>CxxTest User's Guide</title>
<meta name="description" content="CxxTest User's Guide">
<meta name="generator" content="makeinfo 4.3">
</head>
<body>
<h1 class="settitle">CxxTest User's Guide</h1>
<a href="#contents">Table of contents</a>
<h2 class="chapter"><a name="TOC0"></a>1 Introduction</h2>
<p>CxxTest is a <a href="http://junit.org/">JUnit</a>/<a href="http://cppunit.sourceforge.net">CppUnit</a>/<a href="http://xprogramming.com/software.html">xUnit</a>-like framework for C++.
<p>Its advantages over existing alternatives are that it:
<ul>
<li>Doesn't require RTTI
<li>Doesn't require member template functions
<li>Doesn't require exception handling
<li>Doesn't require any external libraries (including memory management, file/console I/O, graphics libraries)
</ul>
In other words, CxxTest is designed to be as portable as possible. Its
only requirements are a reasonably modern C++ compiler and either Perl
or Python. However, when advanced features are supported in your
environment, CxxTest can use them, e.g. catch unhandled exceptions and
even display a GUI.
<p>In addition, CxxTest is slightly easier to use than the C++
alternatives, since you don't need to "register" your tests. It also
features some extras like a richer set of assertions and even support
for a "to do" list (see <a href="#warn"><code>TS_WARN(</code><em></em><code>)</code></a>).
<p>CxxTest is available under the
<h3 class="section"><a name="TOC1"></a>1.1 About this guide</h3>
<p>This guide is not intended as an introduction to Extreme Progamming
<h2 class="chapter"><a name="TOC2"></a>2 Getting started</h2>
<h3 class="section"><a name="TOC3"></a>2.1 Getting CxxTest</h3>
<p>The homepage for CxxTest is <a href="http://cxxtest.sourceforge.net">http://cxxtest.sourceforge.net</a>. You
can always get the latest release from the SourceForge download page,
of this guide is available online at
<a href="http://cxxtest.sourceforge.net/guide.html">http://cxxtest.sourceforge.net/guide.html</a>. A PDF version is also
available at <a href="http://cxxtest.sourceforge.net/guide.pdf">http://cxxtest.sourceforge.net/guide.pdf</a>.
<h3 class="section"><a name="TOC4"></a>2.2 Your first test!</h3>
<p>Here's a simple step-by-step guide:
<ol type=1 start=1>
<li>Tests are organized into "Test Suites".
Test suites are written in header files.
<p>A test suite is a class that inherits from <code>CxxTest::TestSuite</code>.
A test is a public <code>void (void)</code> member function of that class whose name starts with <code>test</code>,
e.g. <code>testDirectoryScanner()</code>, <code>test_cool_feature()</code> and even <code>TestImportantBugFix()</code>.
<pre class="smallexample">
<pre class="verbatim">
#include <cxxtest/TestSuite.h>
class MyTestSuite : public CxxTest::TestSuite
{
public:
void testAddition( void )
{
TS_ASSERT( 1 + 1 > 1 );
TS_ASSERT_EQUALS( 1 + 1, 2 );
}
};
</pre>
</pre>
</p><li>After you have your test suites, you use CxxTest to generate a "test runner" source file:
<pre class="verbatim">
</pre>
<p>or, for those less fortunate:
<pre class="verbatim">
</pre>
</p><li>You would then simply compile the resulting file:
<pre class="verbatim">
# g++ -o runner runner.cpp
</pre>
<p>or perhaps
<pre class="verbatim">
C:\tmp> cl -GX -o runner.exe runner.cpp
</pre>
<p>or maybe even
<pre class="verbatim">
C:\tmp> bcc32 -erunner.exe runner.cpp
</pre>
</p><li>Finally, you run the tests and enjoy a well tested piece of software:
<pre class="verbatim">
# /runner
</pre>
</ol>
<h3 class="section"><a name="TOC5"></a>2.3 Your second test</h3>
<p>Now let's see what failed tests look like.
We will add a failing test to the previous example:
<pre class="smallexample">
<pre class="verbatim">
#include <cxxtest/TestSuite.h>
class MyTestSuite : public CxxTest::TestSuite
{
public:
void testAddition( void )
{
TS_ASSERT( 1 + 1 > 1 );
TS_ASSERT_EQUALS( 1 + 1, 2 );
}
void testMultiplication( void )
{
TS_ASSERT_EQUALS( 2 * 2, 5 );
}
};
</pre>
</pre>
<p>Generate, compile and run the test runner, and you will get this:
<pre class="smallexample">
<pre class="verbatim">
# /runner
Running 2 tests.
MyTestSuite.h:15: Expected (2 * 2 == 5), found (4 != 5)
Failed 1 of 2 tests
Success rate: 50%
</pre>
</pre>
<p>Fixing the bug is left as an excercise to the reader.
<h3 class="section"><a name="TOC6"></a>2.4 Graphical user interface</h3>
<font size=-2>(v3.0.1)</font>
CxxTest can also display a simple GUI. The way to do this is depends on
your compiler, OS and environment, but try the following pointers:
<ul>
</code><em>--gui=X11Gui</em><code> MyTestSuite</code>. You may need to tell the compiler
where to find X, usually something like <code>g++ -o runner
always, compile and link the Qt headers and libraries.
</ul>
<p>See <a href="#GUI">Graphical user interface</a> and <a href="#samples">Running the samples</a> for
more information.
<h2 class="chapter"><a name="TOC7"></a>3 <em>Really</em> using CxxTest</h2>
<p>There is much more to CxxTest than seeing if two times two is four.
You should probably take a look at the samples in the CxxTest distribution.
Other than that, here are some more in-depth explanations.
<h3 class="section"><a name="TOC8"></a>3.1 What can you test</h3>
<p>Here are the different "assertions" you can use in your tests:
<p><table><tr align="left"><td valign="top"><strong>Macro</strong> </td><td valign="top"><strong>Description</strong> </td><td valign="top"><strong>Example</strong>
<br></td></tr><tr align="left"><td valign="top"><a href="#fail"><code>TS_FAIL(</code><em>message</em><code>)</code></a>
</td><td valign="top">Fail unconditionally
</td><td valign="top"><code>TS_FAIL("Test not implemented");</code>
<br></td></tr><tr align="left"><td valign="top"><a href="#assert"><code>TS_ASSERT(</code><em>expr</em><code>)</code></a>
</td><td valign="top">Verify <code></code><em>(expr)</em><code></code> is true
</td><td valign="top"><code>TS_ASSERT(messageReceived());</code>
<br></td></tr><tr align="left"><td valign="top"><a href="#equals"><code>TS_ASSERT_EQUALS(</code><em>x, y</em><code>)</code></a>
</td><td valign="top">Verify <code></code><em>(x==y)</em><code></code>
</td><td valign="top"><code>TS_ASSERT_EQUALS(nodeCount(), 14);</code>
<br></td></tr><tr align="left"><td valign="top"><a href="#sameData"><code>TS_ASSERT_SAME_DATA(</code><em>x, y, size</em><code>)</code></a>
</td><td valign="top">Verify two buffers are equal
</td><td valign="top"><code>TS_ASSERT_SAME_DATA(input, output,, size);</code>
<br></td></tr><tr align="left"><td valign="top"><a href="#delta"><code>TS_ASSERT_DELTA(</code><em>x, y, d</em><code>)</code></a>
</td><td valign="top">Verify <code></code><em>(x==y)</em><code></code> up to <code></code><em>d</em><code></code>
</td><td valign="top"><code>TS_ASSERT_DELTA(sqrt(4.0), 2.0, 0.0001);</code>
<br></td></tr><tr align="left"><td valign="top"><a href="#differs"><code>TS_ASSERT_DIFFERS(</code><em>x, y</em><code>)</code></a>
</td><td valign="top">Verify <code></code><em>!(x==y)</em><code></code>
<br></td></tr><tr align="left"><td valign="top"><a href="#lessthan"><code>TS_ASSERT_LESS_THAN(</code><em>x, y</em><code>)</code></a>
</td><td valign="top">Verify <code></code><em>(x<y)</em><code></code>
<br></td></tr><tr align="left"><td valign="top"><a href="#lte"><code>TS_ASSERT_LESS_THAN_EQUALS(</code><em>x, y</em><code>)</code></a>
</td><td valign="top">Verify <code></code><em>(x<=y)</em><code></code>
</td><td valign="top"><code>TS_ASSERT_LESS_THAN_EQUALS(requests, items);</code>
<br></td></tr><tr align="left"><td valign="top"><a href="#predicate"><code>TS_ASSERT_PREDICATE(</code><em>R, x</em><code>)</code></a>
</td><td valign="top">Verify <code></code><em>P(x)</em><code></code>
</td><td valign="top"><code>TS_ASSERT_PREDICATE(SeemsReasonable, salary);</code>
<br></td></tr><tr align="left"><td valign="top"><a href="#relation"><code>TS_ASSERT_RELATION(</code><em>R, x, y</em><code>)</code></a>
</td><td valign="top">Verify <code></code><em>x R y</em><code></code>
</td><td valign="top"><code>TS_ASSERT_RELATION(std::greater, salary, average);</code>
<br></td></tr><tr align="left"><td valign="top"><a href="#throwsx"><code>TS_ASSERT_THROWS(</code><em>expr, type</em><code>)</code></a>
</td><td valign="top">Verify that <code></code><em>(expr)</em><code></code> throws a specific type of exception
</td><td valign="top"><code>TS_ASSERT_THROWS(parse(file), Parser::ReadError);</code>
<br></td></tr><tr align="left"><td valign="top"><a href="#throwsx"><code>TS_ASSERT_THROWS_EQUALS(</code><em>expr, arg, x, y</em><code>)</code></a>
</td><td valign="top">Verify type and value of what <code></code><em>(expr)</em><code></code> throws
</td><td valign="top">(See text)
<br></td></tr><tr align="left"><td valign="top"><a href="#throwsx"><code>TS_ASSERT_THROWS_ASSERT(</code><em>expr, arg, assertion</em><code>)</code></a>
</td><td valign="top">Verify type and value of what <code></code><em>(expr)</em><code></code> throws
</td><td valign="top">(See text)
<br></td></tr><tr align="left"><td valign="top"><a href="#throwsx"><code>TS_ASSERT_THROWS_ANYTHING(</code><em>expr</em><code>)</code></a>
</td><td valign="top">Verify that <code></code><em>(expr)</em><code></code> throws an exception
</td><td valign="top"><code>TS_ASSERT_THROWS_ANYTHING(buggy());</code>
<br></td></tr><tr align="left"><td valign="top"><a href="#throwsx"><code>TS_ASSERT_THROWS_NOTHING(</code><em>expr</em><code>)</code></a>
</td><td valign="top">Verify that <code></code><em>(expr)</em><code></code> doesn't throw anything
</td><td valign="top"><code>TS_ASSERT_THROWS_NOTHING(robust());</code>
<br></td></tr><tr align="left"><td valign="top"><a href="#warn"><code>TS_WARN(</code><em>message</em><code>)</code></a>
</td><td valign="top">Print <code></code><em>message</em><code></code> as a warning
</td><td valign="top"><code>TS_WARN("TODO: Check invalid parameters");</code>
<br></td></tr><tr align="left"><td valign="top"><a href="#trace"><code>TS_TRACE(</code><em>message</em><code>)</code></a>
</td><td valign="top">Print <code></code><em>message</em><code></code> as an informational message
</td><td valign="top"><code>TS_TRACE(errno);</code>
<br></td></tr></table>
<h4 class="subsection"><a name="TOC9"></a>3.1.1 <code>TS_FAIL</code></h4>
<p><a name="fail"></a>
<p><code>TS_FAIL</code> just fails the test.
It is like an <code>assert(false)</code> with an error message.
For example:
<pre class="smallexample">
<pre class="verbatim">
void testSomething( void )
{
TS_FAIL( "I don't know how to test this!" );
}
</pre>
</pre>
<h4 class="subsection"><a name="TOC10"></a>3.1.2 <code>TS_ASSERT</code></h4>
<p><a name="assert"></a>
<p><code>TS_ASSERT</code> is the basic all-around tester. It works just like the
well-respected <code>assert()</code> macro (which I sincerely hope you know and
use!) An example:
<pre class="smallexample">
<pre class="verbatim">
void testSquare( void )
{
MyFileLibrary::createEmptyFile( "test.bin" );
TS_ASSERT( access( "test.bin", 0 ) == 0 );
}
</pre>
</pre>
<h4 class="subsection"><a name="TOC11"></a>3.1.3 <code>TS_ASSERT_EQUALS</code></h4>
<p><a name="equals"></a>
<p>This is the second most useful tester.
As the name hints, it is used to test if two values are equal.
<pre class="smallexample">
<pre class="verbatim">
void testSquare( void )
{
TS_ASSERT_EQUALS( square(-5), 25 );
}
</pre>
</pre>
<h4 class="subsection"><a name="TOC12"></a>3.1.4 <code>TS_ASSERT_SAME_DATA</code></h4>
<p><a name="sameData"></a>
<font size=-2>(v3.5.1)</font>
This assertion is similar to <a href="#equals"><code>TS_ASSERT_EQUALS(</code><em></em><code>)</code></a>, except that it
compares the contents of two buffers in memory. If the comparison
fails, the standard runner dumps the contents of both buffers as hex
values.
<pre class="smallexample">
<pre class="verbatim">
void testCopyMemory( void )
{
char input[77], output[77];
myCopyMemory( output, input, 77 );
TS_ASSERT_SAME_DATA( input, output, 77 );
}
</pre>
</pre>
<h4 class="subsection"><a name="TOC13"></a>3.1.5 <code>TS_ASSERT_DELTA</code></h4>
<p><a name="delta"></a>
<p>Similar to <a href="#equals"><code>TS_ASSERT_EQUALS(</code><em></em><code>)</code></a>, this macro
verifies two values are equal up to a delta.
This is basically used for floating-point values.
<pre class="smallexample">
<pre class="verbatim">
void testSquareRoot( void )
{
TS_ASSERT_DELTA( squareRoot(4.0), 2.0, 0.00001 );
}
</pre>
</pre>
<h4 class="subsection"><a name="TOC14"></a>3.1.6 <code>TS_ASSERT_DIFFERS</code></h4>
<p><a name="differs"></a>
<p>The opposite of <a href="#equals"><code>TS_ASSERT_EQUALS(</code><em></em><code>)</code></a>, this macro is used to assert
that two values are not equal.
<pre class="smallexample">
<pre class="verbatim">
void testNumberGenerator( void )
{
int first = generateNumber();
int second = generateNumber();
TS_ASSERT_DIFFERS( first, second );
}
</pre>
</pre>
<h4 class="subsection"><a name="TOC15"></a>3.1.7 <code>TS_ASSERT_LESS_THAN</code></h4>
<p><a name="lessthan"></a>
<p>This macro asserts that the first operand is less than the second.
<pre class="smallexample">
<pre class="verbatim">
void testFindLargerNumber( void )
{
TS_ASSERT_LESS_THAN( 23, findLargerNumber(23) );
}
</pre>
</pre>
<h4 class="subsection"><a name="TOC16"></a>3.1.8 <code>TS_ASSERT_LESS_THAN_EQUALS</code></h4>
<p><a name="lte"></a>
<font size=-2>(v3.7.0)</font>
Not surprisingly, this macro asserts that the first operand is less than or equals the second.
<pre class="smallexample">
<pre class="verbatim">
void testBufferSize( void )
{
TS_ASSERT_LESS_THAN_EQUALS( bufferSize(), MAX_BUFFER_SIZE );
}
</pre>
</pre>
<h4 class="subsection"><a name="TOC17"></a>3.1.9 <code>TS_ASSERT_PREDICATE</code></h4>
<p><a name="predicate"></a>
<font size=-2>(v3.8.2)</font>
This macro can be seen as a generalization of
<a href="#assert"><code>TS_ASSERT(</code><em></em><code>)</code></a>. It takes as an argument the name of a class,
similar to an STL <code>unary_function</code>, and evaluates
<code>operator()</code>. The advantage this has over
<a href="#assert"><code>TS_ASSERT(</code><em></em><code>)</code></a> is that you can see the failed value.
<pre class="smallexample">
<pre class="verbatim">
class IsPrime
{
public:
bool operator()( unsigned ) const;
};
// ...
void testPrimeGenerator( void )
{
TS_ASSERT_PREDICATE( IsPrime, generatePrime() );
}
</pre>
</pre>
<h4 class="subsection"><a name="TOC18"></a>3.1.10 <code>TS_ASSERT_RELATION</code></h4>
<p><a name="relation"></a>
<font size=-2>(v3.8.0)</font>
Closely related to
<a href="#predicate"><code>TS_ASSERT_PREDICATE(</code><em></em><code>)</code></a>, this macro can be seen as a
generalization of <a href="#equals"><code>TS_ASSERT_EQUALS(</code><em></em><code>)</code></a>,
<a href="#differs"><code>TS_ASSERT_DIFFERS(</code><em></em><code>)</code></a>,
<a href="#lessthan"><code>TS_ASSERT_LESS_THAN(</code><em></em><code>)</code></a> and
<a href="#lte"><code>TS_ASSERT_LESS_THAN_EQUALS(</code><em></em><code>)</code></a>. It takes as an argument the
name of a class, similar to an STL <code>binary_function</code>, and evaluates
<code>operator()</code>. This can be used to very simply assert comparisons
which are not covered by the builtin macros.
<pre class="smallexample">
<pre class="verbatim">
void testGreater( void )
{
TS_ASSERT_RELATION( std::greater<int>, ticketsSold(), 1000 );
}
</pre>
</pre>
<h4 class="subsection"><a name="TOC19"></a>3.1.11 <code>TS_ASSERT_THROWS</code> and friends</h4>
<p><a name="throwsx"></a>
<p>These assertions are used to test whether an expression throws an exception.
<code>TS_ASSERT_THROWS</code> is used when you want to verify the type of exception
thrown, and <code>TS_ASSERT_THROWS_ANYTHING</code> is used to just make sure something
is thrown. As you might have guessed, <code>TS_ASSERT_THROWS_NOTHING</code> asserts
that nothing is thrown.
<font size=-2>(v3.10.0)</font>
<code>TS_ASSERT_THROWS_EQUALS</code> checks the type of the
exception as in <code>TS_ASSERT_THROWS</code> then allows you to compare two
value (one of which will presumably be the caught object).
<code>TS_ASSERT_THROWS_ASSERT</code> is the general case, and allows you to
make any assertion about the thrown value. These macros may seem a
little complicated, but they can be very useful; see below for an
example.
<pre class="smallexample">
<pre class="verbatim">
void testFunctionsWhichThrowExceptions( void )
{
TS_ASSERT_THROWS_NOTHING( checkInput(1) );
TS_ASSERT_THROWS( checkInput(-11), std::runtime_error );
TS_ASSERT_THROWS_ANYTHING( thirdPartyFunction() );
TS_ASSERT_THROWS_EQUALS( validate(), const std::exception &e,
e.what(), "Invalid value" );
TS_ASSERT_THROWS_ASSERT( validate(), const Error &e,
TS_ASSERT_DIFFERS( e.code(), SUCCESS ) );
}
</pre>
</pre>
<h4 class="subsection"><a name="TOC20"></a>3.1.12 <code>TS_TRACE</code> and <code>TS_WARN</code></h4>
<p><a name="warn"></a>
<a name="trace"></a>
<font size=-2>(v3.0.1)</font>
<code>TS_WARN</code> just prints out a message, like the
<code>#warning</code> preprocessor directive. I find it very useful for "to
do" items. For example:
<pre class="smallexample">
<pre class="verbatim">
void testToDoList( void )
{
TS_WARN( "TODO: Write some tests!" );
TS_WARN( "TODO: Make $$$ fast!" );
}
</pre>
</pre>
<p>In the GUI, <code>TS_WARN</code> sets the bar color to yellow (unless it was
already red).
<font size=-2>(v3.9.0)</font>
<code>TS_TRACE</code> is the same, except that it
doesn't change the color of the progress bar.
<h4 class="subsection"><a name="TOC21"></a>3.1.13 The <code>ETS_</code> macros</h4>
<p>The <code>TS_</code> macros mentioned above will catch exceptions thrown from tested code
and fail the test, as if you called <a href="#fail"><code>TS_FAIL(</code><em></em><code>)</code></a>.
Sometimes, however, you may want to catch the exception yourself; when you do, you can
use the <code>ETS_</code> versions of the macros.
<pre class="smallexample">
<pre class="verbatim">
void testInterestingThrower()
{
// Normal way: if an exception is caught we can't examine it
TS_ASSERT_EQUALS( foo(2), 4 );
// More elaborate way:
try { ETS_ASSERT_EQUALS( foo(2), 4 ); }
catch( const BadFoo &e ) { TS_FAIL( e.bar() ); }
}
</pre>
</pre>
<h4 class="subsection"><a name="TOC22"></a>3.1.14 The <code>TSM_</code> macros</h4>
<p>Sometimes the default output generated by the <code>ErrorPrinter</code> doesn't give you enough
information. This often happens when you move common test functionality to helper functions
inside the test suite; when an assertion fails, you do not know its origin.
<p>In the example below (which is the file <code>sample/MessageTest.h</code> from the CxxTest distribution),
we need the message feature to know which invocation of <code>checkValue()</code> failed:
<pre class="smallexample">
<pre class="verbatim">
class MessageTest : public CxxTest::TestSuite
{
public:
void testValues()
{
checkValue( 0, "My hovercraft" );
checkValue( 1, "is full" );
checkValue( 2, "of eels" );
}
void checkValue( unsigned value, const char *message )
{
TSM_ASSERT( message, value );
TSM_ASSERT_EQUALS( message, value, value * value );
}
};
</pre>
</pre>
<h5 class="subsubsection"><a name="TOC23"></a>3.1.14.1 The <code>ETSM_</code> macros</h5>
<p><strong>Note:</strong> As with normal asserts, all <code>TSM_</code> macros have their
non-exception-safe counterparts, the <code>ETSM_</code> macros.
<h3 class="section"><a name="TOC24"></a>3.2 Running the samples</h3>
<p><a name="samples"></a>
<p>CxxTest comes with some samples in the <code>sample/</code> subdirectory of
the distribution. If you look in that directory, you will see three
Borland C++, repectively. These files are provided as a starting point,
and some options may need to be tweaked in them for your system.
<p>If you are running under Windows, a good guess would be to run
-fMakefile.unix run_x11</code> should probably work.
<h3 class="section"><a name="TOC25"></a>3.3 Test fixtures</h3>
<p>When you have several test cases for the same module,
you often find that all of them start with more or less
the same code--creating objects, files, inputs, etc.
They may all have a common ending, too--cleaning up
the mess you left.
<p>You can (and should) put all this code in a common place by overriding
the virtual functions <code>TestSuite::setUp()</code> and
<code>TestSuite::tearDown()</code>. <code>setUp()</code> will
then be called before each test, and <code>tearDown()</code>
after each test.
<pre class="smallexample">
<pre class="verbatim">
class TestFileOps : public CxxTest::TestSuite
{
public:
void setUp() { mkdir( "playground" ); }
void tearDown() { system( "rm -Rf playground"); }
void testCreateFile()
{
FileCreator fc( "playground" );
fc.createFile( "test.bin" );
TS_ASSERT_EQUALS( access( "playground/test.bin", 0 ), 0 );
}
};
</pre>
</pre>
<p><strong>Note new users:</strong> This is probably the single most important
feature to use when your tests become non-trivial.
<h4 class="subsection"><a name="TOC26"></a>3.3.1 Test suite level fixtures</h4>
<p><code>setUp()</code>/<code>tearDown()</code> are executed around each test case. If
you need a fixture on the test suite level, i.e. something that gets
constructed once before all the tests in the test suite are run, see
<a href="#dynamic">Dynamically creating test suites</a> below.
<h3 class="section"><a name="TOC27"></a>3.4 Integrating with your build environment</h3>
<p>It's very hard to maintain your tests if you have to generate, compile and run the test runner
manually all the time.
Fortunately, that's why we have build tools!
<h4 class="subsection"><a name="TOC28"></a>3.4.1 Overview</h4>
<p>Let's assume you're developing an application.
What I usually do is the following:
<ul>
<li>Split the application into a library and a main module that just calls
the library classes.
This way, the test runner will be able to access all your classes through
the library.
<li>Create another application (or target, or project, or whatever) for the test runner.
Make the build tool generate it automatically.
<li>For extra points, make the build tool run the tests automatically.
</ul>
<h4 class="subsection"><a name="TOC29"></a>3.4.2 Actually doing it</h4>
<p>Unfortunately, there are way too many different build tools and IDE's for me
to give ways to use CxxTest with all of them.
<p>I will try to outline the usage for some cases.
<h5 class="subsubsection"><a name="TOC30"></a>3.4.2.1 Using Makefiles</h5>
<p>Generating the tests with a makefile is pretty straightforward.
Simply add rules to generate, compile and run the test runner.
<pre class="smallexample">
<pre class="verbatim">
all: lib run_tests app
# Rules to build your targets
lib: ...
app: ...
# A rule that runs the unit tests
run_tests: runner
# How to build the test runner
runner: runner.cpp lib
g++ -o $@ $^
# How to generate the test runner
cxxtestgen.pl -o $@ --error-printer $^
</pre>
</pre>
<h5 class="subsubsection"><a name="TOC31"></a>3.4.2.2 Using Cons</h5>
versatile make replacement which uses Perl scripts instead of Makefiles.
<p>See <code>sample/Construct</code> in the CxxTest distribution for an example of building CxxTest test runners
with Cons.
<h5 class="subsubsection"><a name="TOC32"></a>3.4.2.3 Using Microsoft Visual Studio</h5>
<p>I have tried several ways to integrate CxxTest with visual studio, none of
to see the best solution I'm aware of. Basically, the workspace has three
projects:
<ul>
<li>The project <code>CxxTest_3_Generate</code> runs <code>cxxtestgen</code>.
<li>The project <code>CxxTest_2_Build</code> compiles the generated file.
<li>The project <code>CxxTest_1_Run</code> runs the tests.
</ul>
<p>This method certainly works, and the test results are conveniently
displayed as compilation errors and warnings (for
<a href="#warn"><code>TS_WARN(</code><em></em><code>)</code></a>). However, there are still a few things missing;
to integrate this approach with your own project, you usually need to
work a little bit and tweak some makefiles and project options. I have
some of the process.
<h5 class="subsubsection"><a name="TOC33"></a>3.4.2.4 Using Microsoft Windows DDK</h5>
<p>Unit testing for device drivers?! Why not?
And besides, the <code>build</code> utility can also be used to build
user-mode application.
<p>To use CxxTest with the <code>build</code> utility,
you add the generated tests file as an extra dependency
file.
<p>You can see an example of how to do this in the CxxTest distribution
<h3 class="section"><a name="TOC34"></a>3.5 Graphical user interface</h3>
<p><a name="GUI"></a>
<p>There are currently three GUIs implemented: native Win32, native X11 and
Qt. To use this feature, just specify <code>--gui=X11Gui</code>,
<code>--gui=Win32Gui</code> or <code>--gui=QtGui</code> as a parameter for
progress bar is displayed, but the results are still written to standard
output, where they can be processed by your IDE (e.g. Emacs or Visual
Studio). The default behavior of the GUI is to close the window after
the last test.
<p>Note that whatevr GUI you use, you can combine it with the
<code>--runner</code> option to control the formatting of the text output,
e.g. Visual Studio likes it better if you use
<code>--runner=ParenPrinter</code>.
<h4 class="subsection"><a name="TOC35"></a>3.5.1 Starting the GUI minimized</h4>
<p>If you run the generated Win32 or Qt GUIs with the command line
<code>-minimized</code>, the test window will start minimized (iconified)
and only pop up if there is an error (the bar turns red). This is useful
if you find the progress bar distracting and only want to check it if
something happens.
<h4 class="subsection"><a name="TOC36"></a>3.5.2 Leaving the GUI open</h4>
<p>The Win32 GUI accepts the <code>-keep</code> which instructs it to leave the
window open after the tests are done. This allows you to see how many
tests failed and how much time it took.
<h4 class="subsection"><a name="TOC37"></a>3.5.3 Screenshots!</h4>
<p>As with any self-respecting GUI application, here are some screenshots for
you to enjoy:
<ul>
<li>Using the Qt GUI on Linux (with the WindowMaker window manager):
<br><p>
<li>Using the Win32 GUI on Windows 98:
<br><p>
<li>Using the X11 GUI (with the venerable TWM):
<br><p>
<li>And of course, no GUI is complete without the ability to mess around with
its appearance:
<br><p>
<p><em>Ahhh.</em> Nothing like a beautiful user interface.
</ul>
<h2 class="chapter"><a name="TOC38"></a>4 Advanced topics</h2>
<p>Topics in this section are more technical, and you probably won't find them
interesting unless you need them.
<h3 class="section"><a name="TOC39"></a>4.1 Aborting tests after failures</h3>
<p>Usually, when a <code>TS_ASSERT_*</code> macro fails, CxxTest moves on to the
next line. In many cases, however, this is not the desired behavior.
Consider the following code:
<pre class="smallexample">
<pre class="verbatim">
void test_memset()
{
char *buffer = new char[1024];
TS_ASSERT( buffer );
memset( buffer, 0, 1024 ); // But what if buffer == 0?
}
</pre>
</pre>
<p>If you have exception handling enabled, you can make CxxTest exit each
test as soon as a failure occurs. To do this, you need to define
<code>CXXTEST_ABORT_TEST_ON_FAIL</code> before including the CxxTest
headers. This can be done using the <code>--abort-on-fail</code>
command-line option or in a template file; see
doesn't find evidence of exception handling when scanning your files,
this feature will not work. To overcome this, use the
<code>--have-eh</code> command-line option.
<h4 class="subsection"><a name="TOC40"></a>4.1.1 Controlling this behavior at runtime</h4>
<font size=-2>(v3.8.5)</font>
In some scenarios, you may want some tests to abort on
failed assertions and others to continue. To do this you use the
<code>--abort-on-fail</code> option and call the function
<code>CxxTest::setAbortTestOnFail( bool )</code> to change the runtime
behavior. This flag is reset (normally, to <code>true</code>) after each
test, but you can set it in your test suite's <code>setUp()</code> function to
modify the behavior for all tests in a suite.
<font size=-2>(v3.9.0)</font>
Note that this behavior is available whenever you have
exception handling (<code>--have-eh</code> or <code>CXXTEST_HAVE_EH</code>); all
<code>--abort-on-fail</code> does is set the default to <code>true</code>.
<h3 class="section"><a name="TOC41"></a>4.2 Commenting out tests</h3>
<p>CxxTest does a very simple analysis of the input files, which is sufficient in most cases.
This means, for example, that you can't indent you test code in "weird" ways.
<p>A slight inconvenience arises, however, when you want to comment out
tests. Commenting out the tests using C-style comments or the
preprocessor will not work:
<pre class="smallexample">
<pre class="verbatim">
class MyTest : public CxxTest::TestSuite
{
public:
/*
void testCommentedOutStillGetsCalled()
{
}
*/
#if 0
void testMarkedOutStillGetsCalled()
{
}
#endif
};
</pre>
</pre>
<font size=-2>(v3.10.0)</font>
If you need to comment out tests, use C++-style
comments. Also, if you just don't want CxxTest to run a specific test
function, you can temporarily change its name, e.g. by prefixing it with
<code>x</code>:
<pre class="smallexample">
<pre class="verbatim">
class MyTest : public CxxTest::TestSuite
{
public:
// void testFutureStuff()
// {
// }
void xtestFutureStuff()
{
}
};
</pre>
</pre>
<h3 class="section"><a name="TOC42"></a>4.3 Comparing equality for your own types</h3>
<p>You may have noticed that <a href="#equals"><code>TS_ASSERT_EQUALS(</code><em></em><code>)</code></a> only works for built-in
types.
This is because CxxTest needs a way to compare object and to convert them to strings,
in order to print them should the test fail.
<p>If you do want to use <a href="#equals"><code>TS_ASSERT_EQUALS(</code><em></em><code>)</code></a> on your own data types,
this is how you do it.
<h4 class="subsection"><a name="TOC43"></a>4.3.1 The equality operator</h4>
<p>First of all, don't forget to implement the equality operator (<code>operator==()</code>)
on your data types!
<h4 class="subsection"><a name="TOC44"></a>4.3.2 Value traits</h4>
<p>Since CxxTest tries not to rely on any external library (including the standard library,
which is not always available), conversion from arbitrary data types to strings
is done using value traits.
<p>For example, to convert an integer to a string, CxxTest does the following actions:
<ul>
<li><code>int i = </code><em>value to convert</em><code>;</code>
<li><code>CxxTest::ValueTraits<int> converter(i);</code>
</ul>
<p>CxxTest comes with predefined <code>ValueTrait</code>s for <code>int</code>,
<code>cxxtest-selftest</code> archive.
<h4 class="subsection"><a name="TOC45"></a>4.3.3 Unknown types</h4>
<p>Obviously, CxxTest doesn't "know" about all possible types.
The default ValueTraits class for unknown types dumps up to 8 bytes of the value in hex format.
<p>For example, the following code
<pre class="smallexample">
<pre class="verbatim">
#include <cxxtest/TestSuite.h>
class TestMyData : public CxxTest::TestSuite
{
public:
struct Data
{
char data[3];
};
void testCompareData()
{
Data x, y;
TS_ASSERT_EQUALS( x, y );
}
};
</pre>
</pre>
would output
<pre class="smallexample">
<pre class="verbatim">
Running 1 test.
TestMyData.h:16: Expected (x == y), found ({ 12 12 12 } != { F6 F6 F6 })
Failed 1 of 1 test
Success rate: 0%
</pre>
</pre>
<h4 class="subsection"><a name="TOC46"></a>4.3.4 Enumeration traits</h4>
<font size=-2>(v3.10.0)</font>
CxxTest provides a simple way to define value traits for
your enumeration types, which is very handy for things like status
codes. To do this, simply use <code>CXXTEST_VALUE_TRAITS</code> as in the
following example:
<pre class="smallexample">
<pre class="verbatim">
enum Status { STATUS_IDLE, STATUS_BUSY, STATUS_ERROR };
CXXTEST_ENUM_TRAITS( Status,
CXXTEST_ENUM_MEMBER( STATUS_IDLE )
CXXTEST_ENUM_MEMBER( STATUS_BUSY )
CXXTEST_ENUM_MEMBER( STATUS_ERROR ) );
</pre>
</pre>
<h4 class="subsection"><a name="TOC47"></a>4.3.5 Defining new value traits</h4>
<p>Defining value traits for new (non-enumeration) types is easy. All you
need is to define a way to convert an object of your class to a
string. You can use this example as a possible skeleton:
<pre class="smallexample">
<pre class="verbatim">
class MyClass
{
int _value;
public:
MyClass( int value ) : _value( value ) {}
int value() const { return _value; }
// CxxTest requires a copy constructor
MyClass( const MyClass &other ) : _value( other._value ) {}
// If you want to use TS_ASSERT_EQUALS
bool operator== ( const MyClass &other ) const { return _value == other._value; }
// If you want to use TS_ASSERT_LESS_THAN
bool operator== ( const MyClass &other ) const { return _value < other._value; }
};
#ifdef CXXTEST_RUNNING
#include <cxxtest/ValueTraits.h>
#include <stdio.h>
namespace CxxTest
{
CXXTEST_TEMPLATE_INSTANTIATION
class ValueTraits<MyClass>
{
char _s[256];
public:
ValueTraits( const MyClass &m ) { sprintf( _s, "MyClass( %i )", m.value() ); }
const char *asString() const { return _s; }
};
};
#endif // CXXTEST_RUNNING
</pre>
</pre>
<h5 class="subsubsection"><a name="TOC48"></a>4.3.5.1 Defining value traits for template classes</h5>
<p>A simple modification to the above scheme allows you to define value
traits for your template classes. Unfortunately, this syntax (partial
template specialization) is not supported by some popular C++ compilers.
Here is an example:
<pre class="smallexample">
<pre class="verbatim">
template<class T>
class TMyClass
{
T _value;
public:
TMyClass( const T &value ) : _value( value );
const T &value() const { return _value; }
// CxxTest requires a copy constructor
TMyClass( const TMyClass<T> &other ) : _value( other._value ) {}
// If you want to use TS_ASSERT_EQUALS
bool operator== ( const TMyClass<T> &other ) const { return _value == other._value; }
};
#ifdef CXXTEST_RUNNING
#include <cxxtest/ValueTraits.h>
#include <typeinfo>
#include <sstream>
namespace CxxTest
{
template<class T>
class ValueTraits< TMyClass<T> >
{
std::ostringstream _s;
public:
ValueTraits( const TMyClass<T> &t )
{ _s << typeid(t).name() << "( " << t.value() << " )"; }
const char *asString() const { return _s.str().c_str(); }
};
};
#endif // CXXTEST_RUNNING
</pre>
</pre>
<h4 class="subsection"><a name="TOC49"></a>4.3.6 Overriding the default value traits</h4>
<font size=-2>(v2.8.2)</font>
If you don't like the way CxxTest defines the default <code>ValueTrait</code>s,
you can override them by <code>#define</code>-ing <code>CXXTEST_USER_VALUE_TRAITS</code>;
this causes CxxTest to omit the default definitions, and from there on you are
free to implement them as you like.
the <code>cxxtest-selftest</code> archive.
<h3 class="section"><a name="TOC50"></a>4.4 Global Fixtures</h3>
<font size=-2>(v3.5.1)</font>
The <code>setUp()</code> and <code>tearDown()</code> functions allow
to to have code executed before and after each test. What if you want
some code to be executed before all tests in <em>all</em> test suites?
Rather than duplicate that code, you can use <dfn>global fixtures</dfn>.
These are basically classes that inherit from
<code>CxxTest::GlobalFixture</code>. All objects of such classes are
automatically notified before and after each test case. It is best to
create them as static objects so they get called right from the start.
archive.
<p><em>Note:</em> Unlike <code>setUp()</code> and <code>tearDown()</code> in
<code>TestSuite</code>, global fixtures should return a <code>bool</code> value to
<h4 class="subsection"><a name="TOC51"></a>4.4.1 World fixtures</h4>
<font size=-2>(v3.8.1)</font>
CxxTest also allows you to specify code which is executed
once at the start of the testing process (and the corresponding cleanup
code). To do this, create (one or more) global fixture objects and
implement <code>setUpWorld()</code>/<code>tearDownWorld()</code>. For an example,
<h3 class="section"><a name="TOC52"></a>4.5 Mock Objects</h3>
<font size=-2>(v3.10.0)</font>
Mock Objects are a very useful testing tool, which
consists (in a nutshell) of passing special objects to tested code. For
instance, to test a class that implements some protocol over TCP, you
might have it use an abstract <code>ISocket</code> interface and in the tests
pass it a <code>MockSocket</code> object. This <code>MockSocket</code> object can
then do anything your tests find useful, e.g. keep a log of all data
"sent" to verify later.
<p>So far, so good. But the problem when developing in C/C++ is that your
code probably needs to call <em>global</em> functions which you cannot
override. Just consider any code which uses <code>fopen()</code>,
<code>fwrite()</code> and <code>fclose()</code>. It is not very elegant to have
this code actually create files while being tested. Even more
importantly, you (should) want to test how the code behaves when "bad"
things happen, say when <code>fopen()</code> fails. Although for some cases
you can cause the effects to happen in the test code, this quickly
becomes "hairy" and unmaintainable.
<p>CxxTest solves this problem by allowing you to override any global
function while testing. Here is an outline of how it works, before we
see an actual example:
<ul>
<li>For each function you want to override, you use the macro
<code>CXXTEST_MOCK_GLOBAL</code> to "prepare" the function (all is explained
below in excruciating detail).
<li>In the tested code you do not call the global functions directly;
rather, you access them in the <code>T</code> (for <em>Test</em>) namespace. For
instance, your code needs to call <code>T::fopen()</code> instead of
<code>fopen()</code>. This is the equivalent of using abstract interfaces
instead of concrete classes.
<li>You link the "real" binary with a source file that implements
<code>T::fopen()</code> by simply calling the original <code>fopen()</code>.
<li>You link the test binary with a source file that implements
<code>T::fopen()</code> by calling a mock object.
<li>To test, you should create a class that inherits <code>T::Base_fopen</code>
and implement its <code>fopen()</code> function. Simply by creating an object
of this class, calls made to <code>T::fopen()</code> will be redirected to it.
</ul>
<p>This may seem daunting at first, so let us work our way through a simple
example. Say we want to override the well known standard library
function <code>time()</code>.
<ul>
<li>Prepare a header file to be used by both the real and test code.
<pre class="smallexample">
<pre class="verbatim">
#include <time.h>
CXXTEST_MOCK_GLOBAL( time_t, /* Return type */
time, /* Name of the function */
( time_t *t ), /* Prototype */
( t ) /* Argument list */ );
</pre>
</pre>
<li>In our tested code, we now include the special header instead of the
system-supplied one, and call <code>T::time()</code> instead of <code>time()</code>.
<pre class="smallexample">
<pre class="verbatim">
// code.cpp
int generateRandomNumber()
{
return T::time( NULL ) * 3;
}
</pre>
</pre>
<li>We also need to create a source file that implements <code>T::time()</code> by
calling the real function. This is extremely easy: just define
<code>CXXTEST_MOCK_REAL_SOURCE_FILE</code> before you include the header file:
<pre class="smallexample">
<pre class="verbatim">
#define CXXTEST_MOCK_REAL_SOURCE_FILE
</pre>
</pre>
<li>Before we can start testing, we need a different implementation of
<code>T::time()</code> for our tests. This is just as easy as the previous
one:
<pre class="smallexample">
<pre class="verbatim">
#define CXXTEST_MOCK_TEST_SOURCE_FILE
</pre>
</pre>
<li>Now comes the fun part. In our test code, all we need to do is create a
mock, and the tested code will magically call it:
<pre class="smallexample">
<pre class="verbatim">
// TestRandom.h
#include <cxxtest/TestSuite.h>
class TheTimeIsOne : public T::Base_time
{
public:
time_t time( time_t * ) { return 1; }
};
class TestRandom : public CxxTest::TestSuite
{
public:
void test_Random()
{
TheTimeIsOne t;
TS_ASSERT_EQUALS( generateRandomNumber(), 3 );
}
};
</pre>
</pre>
</ul>
<h4 class="subsection"><a name="TOC53"></a>4.5.1 Actually doing it</h4>
<p>I know that this might seem a bit heavy at first glance, but once you
start using mock objects you will never go back. The hardest part may
be getting this to work with your build system, which is why I have
uses GNU Make and G++.
<h4 class="subsection"><a name="TOC54"></a>4.5.2 Advanced topic with mock functions</h4>
<h5 class="subsubsection"><a name="TOC55"></a>4.5.2.1 Void functions</h5>
<p>Void function are a little different, and you use
<code>CXXTEST_MOCK_VOID_GLOBAL</code> to override them. This is identical to
<code>CXXTEST_MOCK_GLOBAL</code> except that it doesn't specify the return
<h5 class="subsubsection"><a name="TOC56"></a>4.5.2.2 Calling the real functions while testing</h5>
<p>From time to time, you might want to let the tested code call the real
functions (while being tested). To do this, you create a special mock
is present, calls to <code>T::time()</code> will be redirected to the real
function.
<h5 class="subsubsection"><a name="TOC57"></a>4.5.2.3 When there is no real function</h5>
<p>Sometimes your code needs to call functions which are not available when
testing. This happens for example when you test driver code using a
user-mode test runner, and you need to call kernel functions. You can
use CxxTest's mock framework to provide testable implementations for the
test code, while maintaing the original functions for the real code.
This you do with <code>CXXTEST_SUPPLY_GLOBAL</code> (and
<code>CXXTEST_SUPPLY_VOID_GLOBAL</code>). For example, say you want to supply
your code with the Win32 kernel function <code>IoCallDriver</code>:
<pre class="smallexample">
<pre class="verbatim">
CXXTEST_SUPPLY_GLOBAL( NTSTATUS, /* Return type */
IoCallDriver, /* Name */
( PDEVICE_OBJECT Device, /* Prototype */
PIRP Irp ),
( Device, Irp ) /* How to call */ );
</pre>
</pre>
The tested code (your driver) can now call <code>IoCallDriver()</code>
normally (no need for <code>T::</code>), and the test code uses
<code>T::Base_IoCallDriver</code> as with normal mock objects.
<p><strong>Note:</strong> Since these macros can also be used to actually declare
the function prototypes (e.g. in the above example you might not be able
<code>extern "C"</code> version which declares the functions with <code>C</code>
linkage. These are <code>CXXTEST_SUPPLY_GLOBAL_C</code> and
<code>CXXTEST_SUPPLY_GLOBAL_VOID_C</code>.
<h5 class="subsubsection"><a name="TOC58"></a>4.5.2.4 Functions in namespaces</h5>
<p>Sometimes the functions you want to override are not in the global
namespace like <code>time()</code>: they may be global functions in other
namespaces or even static class member functions. The default mock
implementation isn't suitable for these. For them, you can use the
generic <code>CXXTEST_MOCK</code>, which is best explained by example. Say you
have a namespace <code>Files</code>, and you want to override the function
<code>bool Files::FileExists( const String &name )</code>, so that the mock
class will be called <code>T::Base_Files_FileExists</code> and the function to
implement would be <code>fileExists</code>. You would define it thus (of
course, you would normally want the mock class name and member function
to be the same as the real function):
<pre class="smallexample">
<pre class="verbatim">
CXXTEST_MOCK( Files_FileExists, /* Suffix of mock class */
bool, /* Return type */
fileExists, /* Name of mock member */
( const String &name ), /* Prototype */
Files::FileExists, /* Name of real function */
( name ) /* Parameter list */ );
</pre>
</pre>
Needless to say, there is also <code>CXXTEST_MOCK_VOID</code> for void functions.
<p>There is also an equivalent version for <code>CXXTEST_SUPPLY_GLOBAL</code>, as
demonstrated by another function from the Win32 DDK:
<pre class="smallexample">
<pre class="verbatim">
CXXTEST_SUPPLY( AllocateIrp, /* => T::Base_AllocateIrp */
PIRP, /* Return type */
allocateIrp, /* Name of mock member */
( CCHAR StackSize ), /* Prototype */
IoAllocateIrp, /* Name of real function */
( StackSize ) /* Parameter list */ );
</pre>
</pre>
And, with this macro you have <code>CXXTEST_SUPPLY_VOID</code> and of course
<code>CXXTEST_SUPPLY_C</code> and <code>CXXTEST_SUPPLY_VOID_C</code>.
<h5 class="subsubsection"><a name="TOC59"></a>4.5.2.5 Overloaded functions</h5>
<p>If you have two or more global functions which have the same name, you
cannot create two mock classes with the same name. The solution is to
use the general <code>CXXTEST_MOCK</code>/<code>CXXTEST_MOCK_VOID</code> as above:
just give the two mock classes different names.
<h5 class="subsubsection"><a name="TOC60"></a>4.5.2.6 Changing the mock namespace</h5>
<p>Finally, if you don't like or for some reason can't use the <code>T::</code>
namespace for mock functions, you can change it by defining
<code>CXXTEST_MOCK_NAMESPACE</code>. Have fun.
<h3 class="section"><a name="TOC61"></a>4.6 Test Listeners and Test Runners</h3>
<p>A <code>TestListener</code> is a class that receives notifications about
the testing process, notably which assertions failed. CxxTest defines
a standard test listener class, <code>ErrorPrinter</code>, which is
responsible for printing the dots and messages seen above. When the
test runners generated in the examples run, they create an
<code>ErrorPrinter</code> and pass it to
<code>TestRunner::runAllTests()</code>. As you might have guessed, this
functions runs all the test you've defined and reports to the
<code>TestListener</code> it was passed.
<h4 class="subsection"><a name="TOC62"></a>4.6.1 Other test listeners</h4>
<p>If you don't like or can't use the <code>ErrorPrinter</code>, you can use
any other test listener.
To do this you have to omit the <code>--error-printer</code>, <code>--runner=</code>
or <code>--gui=</code> switch when generating the tests file.
It is then up to you to write the <code>main()</code> function, using the
test listener of your fancy.
<h5 class="subsubsection"><a name="TOC63"></a>4.6.1.1 The <code>stdio</code> printer</h5>
<p>If the <code>ErrorPrinter</code>'s usage of <code>std::cout</code> clashes
with your environment or is unsupported by your compiler, don't dispair!
You may still be able to use the <code>StdioPrinter</code>, which does the
exact same thing but uses good old <code>printf()</code>.
<font size=-2>(v3.8.5)</font>
reference to <code>stdout</code> as the default output stream. In some
will cause compiler errors. To overcome this problem, use
<code>--runner=StdioFilePrinter</code>, which is exactly the same as
<code>--runner=StdioPrinter</code>, but with no default output stream.
<p>As an example, CxxTest also provides the simplest possible test listener,
one that just reports if there were any failures.
<h5 class="subsubsection"><a name="TOC65"></a>4.6.1.3 Template files</h5>
<p>To use you own test runner, or to use the supplied ones in different ways, you can use
CxxTest <em>template files</em>. These are ordinary source files with the embedded "command"
<code><CxxTest world></code> which tells <code>cxxtestgen.pl</code> to insert the world definition
at that point. You then specify the template file using the <code>--template</code> option.
<p><strong>Note:</strong> CxxTest needs to insert certain definitions and
<code>#include</code> directives in the runner file. It normally does that
before the first <code>#include <cxxtest/*.h></code> found in the template
file. If this behvaior is not what you need, use the "command"
<code>cxxtest-selftest</code> archive for an example of this.
<h3 class="section"><a name="TOC66"></a>4.7 Dynamically creating test suites</h3>
<p><a name="dynamic"></a>
Usually, your test suites are instantiated statically in the tests file, i.e. say you
defined <code>class MyTest : public CxxTest::TestSuite</code>, the generated file will
contain something like <code>static MyTest g_MyTest;</code>.
<p>If, however, your test suite must be created dynamically (it may need a constructor,
for instance), CxxTest doesn't know how to create it unless you tell it how.
You do this by writing two static functions, <code>createSuite()</code> and <code>destroySuite()</code>.
<h3 class="section"><a name="TOC67"></a>4.8 Static initialization</h3>
<font size=-2>(v3.9.0)</font>
The generated runner source file depends quite
heavily on static initialization of the various "description" object
approach, use the <code>--no-static-init</code> option.
<h2 class="appendix"><a name="TOC68"></a>Appendix A Command line options</h2>
<p>Here are the different command line options for <code>cxxtestgen</code>:
<h3 class="section"><a name="TOC69"></a>A.1 <code>--version</code></h3>
<font size=-2>(v3.7.1)</font>
Specify <code>--version</code> or <code>-v</code> to see the version of CxxTest you are using.
<h3 class="section"><a name="TOC70"></a>A.2 <code>--output</code></h3>
<p>Specify <code>--output=FILE</code> or <code>-o FILE</code> to determine the output file name.
<h3 class="section"><a name="TOC71"></a>A.3 <code>--error-printer</code></h3>
<p>This option creates a test runner which uses the standard error printer class.
<h3 class="section"><a name="TOC72"></a>A.4 <code>--runner</code></h3>
<p>Specify <code>--runner=CLASS</code> to generate a test
<code>CxxTest::CLASS</code> as the test runner.
<p>The currently available runners are:
<dl>
<dt><code>--runner=ErrorPrinter</code>
<dd>This is the standard error printer, which formats its output to <code>std::cout</code>.
<br><dt><code>--runner=ParenPrinter</code>
<dd>Identical to <code>ErrorPrinter</code> except that it prints line numbers in parantheses.
This is the way Visual Studio expects it.
<br><dt><code>--runner=StdioPrinter</code>
<dd>The same as <code>ErrorPrinter</code> except that it uses <code>printf</code>
instead of <code>cout</code>.
<br><dt><code>--runner=YesNoRunner</code>
</dl>
<h3 class="section"><a name="TOC73"></a>A.5 <code>--gui</code></h3>
<p>Specify <code>--gui=CLASS</code> to generate a test runner that
to display a graphical user interface. This option can be combined with
the <code>--runner</code> option to determine the text-mode output format.
The default is the standard error printer.
<p>There are three different GUIs:
<dl>
<dt><code>--gui=Win32Gui</code>
<dd>A native Win32 GUI. It has been tested on Windows 98, 2000 and XP and
should work unmodified on other 32-bit versions of Windows.
<br><dt><code>--gui=X11Gui</code>
<dd>A native XLib GUI. This GUI is very spartan and should work on any X server.
<br><dt><code>--gui=QtGui</code>
<dd>A GUI that uses the Qt library from Troll. It has been tested with Qt versiond 2.2.1 and 3.0.1.
</dl>
<h3 class="section"><a name="TOC74"></a>A.6 <code>--include</code></h3>
<font size=-2>(v3.5.1)</font>
If you specify <code>--include=FILE</code>, <code>cxxtestgen</code> will add
<code>#include "FILE"</code> to the runner before including any other header.
This allows you to define things that modify the behavior of CxxTest,
e.g. your own ValueTraits.
<p><strong>Note:</strong> If you want the runner to <code>#inculde <FILE></code>, specify
<code>--include=\<FILE\></code>.
<h3 class="section"><a name="TOC75"></a>A.7 <code>--template</code></h3>
<p>Specify <code>--template=FILE</code> to use <code>FILE</code> as a template file.
are not enough. One example is the Windows DDK; see
<h3 class="section"><a name="TOC76"></a>A.8 <code>--have-eh</code></h3>
<font size=-2>(v2.8.4)</font>
<code>cxxtestgen</code> will scan its input files for uses of exception
handling; if found, the <code>TS_</code> macros will catch exceptions,
allowing the testing to continue. Use <code>--have-eh</code> to tell
<code>cxxtestgen</code> to enable that functionality even if exceptions
are not used in the input files.
<h3 class="section"><a name="TOC77"></a>A.9 <code>--no-eh</code></h3>
<font size=-2>(v3.8.5)</font>
If you want <code>cxxtestgen</code> to ignore what may look as uses of
exception handling in your test files, specify <code>--no-eh</code>.
<h3 class="section"><a name="TOC78"></a>A.10 <code>--have-std</code></h3>
<font size=-2>(v3.10.0)</font>
Same as <code>--have-eh</code> but for the standard library;
basically, if you use this flag, CxxTest will print the values of
<code>std::string</code>.
<p><strong>Note:</strong> If you reference the standard library anywhere in your
test files, CxxTest will (usually) recognize it and automatically define
this.
<h3 class="section"><a name="TOC79"></a>A.11 <code>--no-std</code></h3>
<font size=-2>(v3.10.0)</font>
The counterpart to <code>--have-std</code>, this tells
CxxTest to ignore any evidence it finds for the <code>std::</code> namespace
in your code. Use it if your environment does not support <code>std::</code>
but <code>cxxtestgen</code> thinks it does.
<h3 class="section"><a name="TOC80"></a>A.12 <code>--longlong</code></h3>
<font size=-2>(v3.6.0)</font>
Specify <code>--longlong=TYPE</code> to have CxxTest recognize <code>TYPE</code>
just <code>--longlong=</code> (no type), CxxTest will use the default type
name of <code>long long</code>.
<h3 class="section"><a name="TOC81"></a>A.13 <code>--abort-on-fail</code></h3>
<font size=-2>(v2.8.2)</font>
This useful option tells CxxTest to abort the current test when any
<code>TS_ASSERT</code> macro has failed.
<h3 class="section"><a name="TOC82"></a>A.14 <code>--part</code></h3>
<font size=-2>(v3.5.1)</font>
This option tells CxxTest now to write the CxxTest globals in the output
file. Use this to link together more than one generated file.
<h3 class="section"><a name="TOC83"></a>A.15 <code>--root</code></h3>
<font size=-2>(v3.5.1)</font>
This is the counterpart of <code>--part</code>; it makes sure that the
Cxxtest globals are written to the output file. If you specify this
option, you can use <code>cxxtestgen</code> without any input files to
create a file that hold only the "root" runner.
<h3 class="section"><a name="TOC84"></a>A.16 <code>--no-static-init</code></h3>
<font size=-2>(v3.9.0)</font>
Use this option if you encounter problems with the static
initializations in the test runner.
<h2 class="appendix"><a name="TOC85"></a>Appendix B Controlling the behavior of CxxTest</h2>
<p>Here are various <code>#define</code>s you can use to modify how CxxTest
works. You will need to <code>#define</code> them <em>before</em> including any
of the CxxTest headers, so use them in a template file or with the
<code>--include</code> option.
<h3 class="section"><a name="TOC86"></a>B.1 <code>CXXTEST_HAVE_STD</code></h3>
<p>This is equivalent to the <code>--have-std</code> option.
<h3 class="section"><a name="TOC87"></a>B.2 <code>CXXTEST_HAVE_EH</code></h3>
<p>This is equivalent to the <code>--have-eh</code> option.
<h3 class="section"><a name="TOC88"></a>B.3 <code>CXXTEST_ABORT_TEST_ON_FAIL</code></h3>
<font size=-2>(v2.8.0)</font>
This is equivalent to the <code>--abort-on-fail</code> option.
<h3 class="section"><a name="TOC89"></a>B.4 <code>CXXTEST_USER_VALUE_TRAITS</code></h3>
<p>This tells CxxTest you wish to define you own ValueTraits. It will only
declare the default traits, which dump up to 8 bytes of the data as hex
values.
<h3 class="section"><a name="TOC90"></a>B.5 <code>CXXTEST_OLD_TEMPLATE_SYNTAX</code></h3>
<p>Some compilers (e.g. Borland C++ 5) don't support the standard way of
instantiating template classes. Use this define to overcome the problem.
<h3 class="section"><a name="TOC91"></a>B.6 <code>CXXTEST_OLD_STD</code></h3>
<p>Again, this is used to support pre-<code>std::</code> standard libraries.
<h3 class="section"><a name="TOC92"></a>B.7 <code>CXXTEST_MAX_DUMP_SIZE</code></h3>
<p>This sets the standard maximum number of bytes to dump if
<a href="#sameData"><code>TS_ASSERT_SAME_DATA(</code><em></em><code>)</code></a> fails. The default is 0, meaning
no limit.
<h3 class="section"><a name="TOC93"></a>B.8 <code>CXXTEST_DEFAULT_ABORT</code></h3>
<p>This sets the default value of the dynamic "abort on fail" flag. Of
course, this flag is only used when "abort on fail" is enabled.
<h3 class="section"><a name="TOC94"></a>B.9 <code>CXXTEST_LONGLONG</code></h3>
<p>This is equivalent to <code>--longlong</code>.
<h2 class="appendix"><a name="TOC95"></a>Appendix C Runtime options</h2>
<p>The following functions can be called during runtime (i.e. from your
tests) to control the behavior of CxxTest. They are reset to their
default values after each test is executed (more precisely, after
<code>tearDown()</code> is called). Consequently, if you set them in the
<code>setUp()</code> function, they will be valid for the entire test suite.
<h3 class="section"><a name="TOC96"></a>C.1 <code>setAbortTestOnFail( bool )</code></h3>
<p>This only works when you have exception handling. It can be used to
tell CxxTest to temporarily change its behavior. The default value of
the flag is <code>false</code>, <code>true</code> if you set <code>--abort-on-fail</code>,
or <code>CXXTEST_DEFAULT_ABORT</code> if you <code>#define</code> it.
<h3 class="section"><a name="TOC97"></a>C.2 <code>setMaxDumpSize( unsigned )</code></h3>
<p>This temporarily sets the maximum number of bytes to dump if
<a href="#sameData"><code>TS_ASSERT_SAME_DATA(</code><em></em><code>)</code></a> fails. The default is 0, meaning
no limit, or <code>CXXTEST_MAX_DUMP_SIZE</code> if you <code>#define</code> it.
<h2 class="appendix"><a name="TOC98"></a>Appendix D Version history</h2>
<ul>
<li><strong>Version 3.10.0 (2004-11-20)</strong>
<ul>
<li>Added mock framework for global functions
<li>Added <code>TS_ASSERT_THROWS_ASSERT</code> and <code>TS_ASSERT_THROWS_EQUALS</code>
<li>Added <code>CXXTEST_ENUM_TRAITS</code>
<li>Improved support for STL classes (vector, map etc.)
<li>Added support for Digital Mars compiler
<li>Support C++-style commenting of tests
</ul>
<li><strong>Version 3.9.1 (2004-01-19)</strong>
<ul>
<li>Fixed small bug with runner exit code
<li>Embedded test suites are now deprecated
</ul>
<li><strong>Version 3.9.0 (2004-01-17)</strong>
<ul>
<li>Added <code>TS_TRACE</code>
<li>Added <code>--no-static-init</code>
<li>CxxTest::<code>setAbortTestOnFail()</code> works even without <code>--abort-on-fail</code>
</ul>
<li><strong>Version 3.8.5 (2004-01-08)</strong>
<ul>
<li>Added <code>--no-eh</code>
<li>Added <code>CxxTest::setAbortTestOnFail()</code> and <code>CXXTEST_DEFAULT_ABORT</code>
<li>Added <code>CxxTest::setMaxDumpSize()</code>
<li>Added StdioFilePrinter
</ul>
<li><strong>Version 3.8.4 (2003-12-31)</strong>
<ul>
<li>Split distribution into cxxtest and cxxtest-selftest
</ul>
<li><strong>Version 3.8.3 (2003-12-24)</strong>
<ul>
<li>Added <code>TS_ASSERT_PREDICATE</code>
<li>Template files can now specify where to insert the preamble
<li>Can compile in MSVC with warning level 4
<li>Changed output format slightly
</ul>
<li><strong>Version 3.8.1 (2003-12-21)</strong>
<ul>
<li>Fixed small bug when using multiple <code>--part</code> files.
<li>Fixed X11 GUI crash when there's no X server.
<li>Added <code>GlobalFixture::setUpWorld()</code>/<code>tearDownWorld()</code>
<li>Added <code>leaveOnly()</code>, <code>activateAllTests()</code> and <code>sample/only.tpl</code>
<li>Should now run without warnings on Sun compiler.
</ul>
<li><strong>Version 3.8.0 (2003-12-13)</strong>
<ul>
<li>Added <code>TS_ASSERT_RELATION</code>
<li><code>TSM_</code> macros now also tell you what went wrong
<li>Renamed <code>Win32Gui::free()</code> to avoid clashes
<li>Now compatible with more versions of Borland compiler
<li>Improved the documentation
</ul>
<li><strong>Version 3.7.1 (2003-09-29)</strong>
<ul>
<li>Added <code>--version</code>
<li>Compiles with even more exotic g++ warnings
<li>Win32 Gui compiles with UNICODE
<li>Should compile on some more platforms (Sun Forte, HP aCC)
</ul>
<li><strong>Version 3.7.0 (2003-09-20)</strong>
<ul>
<li>Added <code>TS_ASSERT_LESS_THAN_EQUALS</code>
<li>Minor cleanups
</ul>
<li><strong>Version 3.6.1 (2003-09-15)</strong>
<ul>
<li>Improved QT GUI
<li>Improved portability some more
</ul>
<li><strong>Version 3.6.0 (2003-09-04)</strong>
<ul>
<li>Added <code>--longlong</code>
<li>Some portability improvements
</ul>
<li><strong>Version 3.5.1 (2003-09-03)</strong>
<ul>
<li>Major internal rewrite of macros
<li>Added <code>TS_ASSERT_SAME_DATA</code>
<li>Added <code>--include</code> option
<li>Added <code>--part</code> and <code>--root</code> to enable splitting the test runner
<li>Added global fixtures
<li>Enhanced Win32 GUI with timers, <code>-keep</code> and <code>-title</code>
<li>Now compiles with strict warnings
</ul>
<li><strong>Version 3.1.1 (2003-08-27)</strong>
<ul>
<li>Fixed small bug in <code>TS_ASSERT_THROWS_*()</code>
</ul>
<li><strong>Version 3.1.0 (2003-08-23)</strong>
<ul>
<li>Default ValueTraits now dumps value as hex bytes
<li><code>TS_ASSERT_THROWS*()</code> are now "abort on fail"-friendly
<li>Win32 GUI now supports Windows 98 and doesn't need comctl32.lib
</ul>
<li><strong>Version 3.0.1 (2003-08-07)</strong>
<ul>
<li>Added simple GUI for X11, Win32 and Qt
<li>Added <code>TS_</code><code>WARN()</code> macro
<li>Removed <code>--exit-code</code>
<li>Improved samples
<li>Improved support for older (pre-std::) compilers
<li>Made a PDF version of the User's Guide
</ul>
<li><strong>Version 2.8.4 (2003-07-21)</strong>
<ul>
<li>Now supports g++-3.3
<li>Added <code>--have-eh</code>
<li>Fixed bug in <code>numberToString()</code>
</ul>
<li><strong>Version 2.8.3 (2003-06-30)</strong>
<ul>
<li>Fixed bugs in cxxtestgen.pl
<li>Thanks Martin Jost for pointing out these problems!
</ul>
<li><strong>Version 2.8.2 (2003-06-10)</strong>
<ul>
<li>Fixed bug when using <code>CXXTEST_ABORT_TEST_ON_FAIL</code> without standard library
<li>Added <code>CXXTEST_USER_TRAITS</code>
<li>Added <code>--abort-on-fail</code>
</ul>
<li><strong>Version 2.8.1 (2003-01-16)</strong>
<ul>
<li>Fixed <code>charToString()</code> for negative chars
</ul>
<li><strong>Version 2.8.0 (2003-01-13)</strong>
<ul>
<li>Added <code>CXXTEST_ABORT_TEST_ON_FAIL</code> for xUnit-like behaviour
<li>Improved ValueTraits
<li>Improved output formatter
<li>Started version history
</ul>
<li><strong>Version 2.7.0 (2002-09-29)</strong>
<ul>
<li>Added embedded test suites
<li>Major internal improvements
</ul>
</ul>
<p><a name="contents"></a>
<div class="contents">
<h2>Table of Contents</h2>
<ul>
<li><a name="toc_TOC0" href="#TOC0">1 Introduction</a>
<ul>
<li><a href="#TOC1">1.1 About this guide</a>
</li></ul>
<li><a name="toc_TOC2" href="#TOC2">2 Getting started</a>
<ul>
<li><a href="#TOC3">2.1 Getting CxxTest</a>
<li><a href="#TOC4">2.2 Your first test!</a>
<li><a href="#TOC5">2.3 Your second test</a>
<li><a href="#TOC6">2.4 Graphical user interface</a>
</li></ul>
<li><a name="toc_TOC7" href="#TOC7">3 <em>Really</em> using CxxTest</a>
<ul>
<li><a href="#TOC8">3.1 What can you test</a>
<ul>
<li><a href="#TOC9">3.1.1 <code>TS_FAIL</code></a>
<li><a href="#TOC10">3.1.2 <code>TS_ASSERT</code></a>
<li><a href="#TOC11">3.1.3 <code>TS_ASSERT_EQUALS</code></a>
<li><a href="#TOC12">3.1.4 <code>TS_ASSERT_SAME_DATA</code></a>
<li><a href="#TOC13">3.1.5 <code>TS_ASSERT_DELTA</code></a>
<li><a href="#TOC14">3.1.6 <code>TS_ASSERT_DIFFERS</code></a>
<li><a href="#TOC15">3.1.7 <code>TS_ASSERT_LESS_THAN</code></a>
<li><a href="#TOC16">3.1.8 <code>TS_ASSERT_LESS_THAN_EQUALS</code></a>
<li><a href="#TOC17">3.1.9 <code>TS_ASSERT_PREDICATE</code></a>
<li><a href="#TOC18">3.1.10 <code>TS_ASSERT_RELATION</code></a>
<li><a href="#TOC19">3.1.11 <code>TS_ASSERT_THROWS</code> and friends</a>
<li><a href="#TOC20">3.1.12 <code>TS_TRACE</code> and <code>TS_WARN</code></a>
<li><a href="#TOC21">3.1.13 The <code>ETS_</code> macros</a>
<li><a href="#TOC22">3.1.14 The <code>TSM_</code> macros</a>
<ul>
<li><a href="#TOC23">3.1.14.1 The <code>ETSM_</code> macros</a>
</li></ul>
</li></ul>
<li><a href="#TOC24">3.2 Running the samples</a>
<li><a href="#TOC25">3.3 Test fixtures</a>
<ul>
<li><a href="#TOC26">3.3.1 Test suite level fixtures</a>
</li></ul>
<li><a href="#TOC27">3.4 Integrating with your build environment</a>
<ul>
<li><a href="#TOC28">3.4.1 Overview</a>
<li><a href="#TOC29">3.4.2 Actually doing it</a>
<ul>
<li><a href="#TOC30">3.4.2.1 Using Makefiles</a>
<li><a href="#TOC31">3.4.2.2 Using Cons</a>
<li><a href="#TOC32">3.4.2.3 Using Microsoft Visual Studio</a>
<li><a href="#TOC33">3.4.2.4 Using Microsoft Windows DDK</a>
</li></ul>
</li></ul>
<li><a href="#TOC34">3.5 Graphical user interface</a>
<ul>
<li><a href="#TOC35">3.5.1 Starting the GUI minimized</a>
<li><a href="#TOC36">3.5.2 Leaving the GUI open</a>
<li><a href="#TOC37">3.5.3 Screenshots!</a>
</li></ul>
</li></ul>
<li><a name="toc_TOC38" href="#TOC38">4 Advanced topics</a>
<ul>
<li><a href="#TOC39">4.1 Aborting tests after failures</a>
<ul>
<li><a href="#TOC40">4.1.1 Controlling this behavior at runtime</a>
</li></ul>
<li><a href="#TOC41">4.2 Commenting out tests</a>
<li><a href="#TOC42">4.3 Comparing equality for your own types</a>
<ul>
<li><a href="#TOC43">4.3.1 The equality operator</a>
<li><a href="#TOC44">4.3.2 Value traits</a>
<li><a href="#TOC45">4.3.3 Unknown types</a>
<li><a href="#TOC46">4.3.4 Enumeration traits</a>
<li><a href="#TOC47">4.3.5 Defining new value traits</a>
<ul>
<li><a href="#TOC48">4.3.5.1 Defining value traits for template classes</a>
</li></ul>
<li><a href="#TOC49">4.3.6 Overriding the default value traits</a>
</li></ul>
<li><a href="#TOC50">4.4 Global Fixtures</a>
<ul>
<li><a href="#TOC51">4.4.1 World fixtures</a>
</li></ul>
<li><a href="#TOC52">4.5 Mock Objects</a>
<ul>
<li><a href="#TOC53">4.5.1 Actually doing it</a>
<li><a href="#TOC54">4.5.2 Advanced topic with mock functions</a>
<ul>
<li><a href="#TOC55">4.5.2.1 Void functions</a>
<li><a href="#TOC56">4.5.2.2 Calling the real functions while testing</a>
<li><a href="#TOC57">4.5.2.3 When there is no real function</a>
<li><a href="#TOC58">4.5.2.4 Functions in namespaces</a>
<li><a href="#TOC59">4.5.2.5 Overloaded functions</a>
<li><a href="#TOC60">4.5.2.6 Changing the mock namespace</a>
</li></ul>
</li></ul>
<li><a href="#TOC61">4.6 Test Listeners and Test Runners</a>
<ul>
<li><a href="#TOC62">4.6.1 Other test listeners</a>
<ul>
<li><a href="#TOC63">4.6.1.1 The <code>stdio</code> printer</a>
<li><a href="#TOC65">4.6.1.3 Template files</a>
</li></ul>
</li></ul>
<li><a href="#TOC66">4.7 Dynamically creating test suites</a>
<li><a href="#TOC67">4.8 Static initialization</a>
</li></ul>
<li><a name="toc_TOC68" href="#TOC68">Appendix A Command line options</a>
<ul>
<li><a href="#TOC69">A.1 <code>--version</code></a>
<li><a href="#TOC70">A.2 <code>--output</code></a>
<li><a href="#TOC71">A.3 <code>--error-printer</code></a>
<li><a href="#TOC72">A.4 <code>--runner</code></a>
<li><a href="#TOC73">A.5 <code>--gui</code></a>
<li><a href="#TOC74">A.6 <code>--include</code></a>
<li><a href="#TOC75">A.7 <code>--template</code></a>
<li><a href="#TOC76">A.8 <code>--have-eh</code></a>
<li><a href="#TOC77">A.9 <code>--no-eh</code></a>
<li><a href="#TOC78">A.10 <code>--have-std</code></a>
<li><a href="#TOC79">A.11 <code>--no-std</code></a>
<li><a href="#TOC80">A.12 <code>--longlong</code></a>
<li><a href="#TOC81">A.13 <code>--abort-on-fail</code></a>
<li><a href="#TOC82">A.14 <code>--part</code></a>
<li><a href="#TOC83">A.15 <code>--root</code></a>
<li><a href="#TOC84">A.16 <code>--no-static-init</code></a>
</li></ul>
<li><a name="toc_TOC85" href="#TOC85">Appendix B Controlling the behavior of CxxTest</a>
<ul>
<li><a href="#TOC86">B.1 <code>CXXTEST_HAVE_STD</code></a>
<li><a href="#TOC87">B.2 <code>CXXTEST_HAVE_EH</code></a>
<li><a href="#TOC88">B.3 <code>CXXTEST_ABORT_TEST_ON_FAIL</code></a>
<li><a href="#TOC89">B.4 <code>CXXTEST_USER_VALUE_TRAITS</code></a>
<li><a href="#TOC90">B.5 <code>CXXTEST_OLD_TEMPLATE_SYNTAX</code></a>
<li><a href="#TOC91">B.6 <code>CXXTEST_OLD_STD</code></a>
<li><a href="#TOC92">B.7 <code>CXXTEST_MAX_DUMP_SIZE</code></a>
<li><a href="#TOC93">B.8 <code>CXXTEST_DEFAULT_ABORT</code></a>
<li><a href="#TOC94">B.9 <code>CXXTEST_LONGLONG</code></a>
</li></ul>
<li><a name="toc_TOC95" href="#TOC95">Appendix C Runtime options</a>
<ul>
<li><a href="#TOC96">C.1 <code>setAbortTestOnFail( bool )</code></a>
<li><a href="#TOC97">C.2 <code>setMaxDumpSize( unsigned )</code></a>
</li></ul>
<li><a name="toc_TOC98" href="#TOC98">Appendix D Version history</a>
</li></ul>
</div>
</body></html>