libdb/test/c

The C test cases are currently (loosly) based on the CuTest harness. Loosely
because the harness has been heavily modified from the original version.

There are still a few old test cases in the source tree. Those will be
converted to run within CuTest at some stage. New tests should all work within
the CuTest structure.

=================== BUILDING ================

To build CuTest on POSIX (*nix) run "make cutest"

To build CuTest on Windows load the test_cutest.vcproj project into the 
Berkeley DB solution file, and build the resulting project. It should depend
on the Berkeley DB library (db).


=================== RUNNING  ================

To run CuTest on POSIX (*nix) run:
./cutest
from the build_unix directory.

To run CuTest on Windows run:

./test_cutest.exe
from the build_windows/Win32/Debug directory.

=================== WRITING  ================

To write a new test case, open the relevant test suite source file under
test/c/suites/

Create a new function, with a name that begins with "Test", that returns an int
and takes a CuTest * argument. For example:

int TestExampleFunction(CuTest *ct)


To write a new test suite (to test a different type of functionality):
	* Create a source file in test/c/suites the name beginning with Test.
          Of course, as with any new source file, it needs to be
          mentioned in dist/Makefile.in and the corresponding Windows files.
	* Include "CuTest.h" in that source file
	* Probably include "test_util.h" in the new source file, to access
	  shared utility functions and macros.
	* Add any test cases, conforming to the test case guidelines above.
	* Optionally specify setup and teardown functions for the suite.
	  Named <SuiteName>SuiteSetup and <SuiteName>SuiteTeardown respectively.
	* Optionally specify setup and teardown functions for each test case
	  in the suite. Named <SuiteName>TestSetup and <SuiteName>TestTeardown
	  respectively.
	* Use the CuAssertXXX macros to validate behavior.
	* See TestEnvConfig.c for an example of good practice.


When adding a new test case or a new test suite, it's necessary to run the a
script to regenerate the test harness wrapper. You'll need a shell script to
do that. The script is:
test/c/cutest/gen_tester.sh

It should be run as:
$ cd test/c/cutest
$ ./gen_tester.sh > CuTests.c


NOTE: A comment on naming.

All tests and suites should be clearly named. Long names are fine, and more
descriptive is generally better.

A test suite should be named to reflect the component and functionality that
is being tested.
An individual test case should be named to identify the specific functionality
being tested.
It's best to avoid including SR numbers in test and suite names. Relevant SRs
should be referenced in comments.

=================== FUTURES  ================a

 * Add a configuration object to the setup and teardown functions, so they can
   pass and maintain state.
 * Fix the naming so that it's more Berkeley DB compatible (not camel case).
 * Implement the RunTest and RunSuite functionality in test/c/cutest/CuTest.c,
   so that it's possible to run parts of the test suite. The format will be:

./cutest -s TestSuiteName -s TestSuiteName -t TestSuiteName:TestCase
 * Implement more shared functionality in test/c/common, move the function
   implementations from the header file in there, so shared library builds will
   work.