What's new in opmock 0.9.1
Aug 16, 2012
- Fixed a code generation bug when a function has anonymous const parameters.
New in opmock 0.9 (Mar 22, 2012)
- I grew jealous of Google Chrome and their never ending version bumps. So I decided it was time to go to 0.9 :-) 2 bugs fixes in this release:
- Yet another bug on consts with C++ functions of the kind const int foo(float j) const;
- A bug when a function returns a templated parameter like std::vector boo(); There were spurious parenthesis in the generated code.
New in opmock 0.8.9 (Mar 7, 2012)
- Fixed many bugs in this release thank's to the fellow users!
- one bug in cmp_cstring : the error message was not printing the 2 string properly in case they're not equal.
- bugs in the ASSERT macros : if macros were nested in the ASSERT macro itself, "bad things" tm could happen. Now macros are protected against this.
New in opmock 0.8.8 (Mar 6, 2012)
- Fixed a bug with the --skip-funct option : constructors and destructors were not filtered out. Fixed a bug when a parameter name was the same than the end of the a const type. For example : void foo(const M_MyData MyData); was resulting in a truncation of the type in the generated code (bad regular expression in the code). Note that however naming a parameter after a type is not a good idea.
New in opmock 0.8.7 (Mar 2, 2012)
- Added negative versions of all assert macros. Example OP_ASSERT_NOT_EQUAL_INT.
- Added an example for mixed use of mocks and matchers. have a look to the c_matchers folder! Updated documentation.
New in opmock 0.8.6 (Feb 29, 2012)
- Fixed a code generation bug with methods returning a const return value.
New in opmock 0.8.5 (Feb 28, 2012)
- Fixed a code generation bug with functions returning a const return value.
New in opmock 0.8.4 (Feb 28, 2012)
- Fix a bug with pointers comparisons using cmp_ptr. In the generated code, pointers were handled as a special case, because the value of the pointer was stored, while for all other parameters a pointer to the parameter is stored. This is because matchers can only accept void * parameters as "generic" type (design choice : the framework has to work both in C and C++). When a pointer was typedef'ed, opmock could not get the "real" type, and was generating code that was not compatible with the way cmp_ptr works. the code generation as well as cmp_ptr and cmp_string have been updated. Make sure you use as well the latest version of opmock.c and opmock.h in your builds as they are touched by the changes.
New in opmock 0.8.3 (Jan 11, 2012)
- Solves a compilation error when using very old compilers (gcc 2.95 to name it).
New in opmock 0.8.2 (Dec 5, 2011)
- Modified the refresh_tests.sh script so that it works whith bash 3. But you should really update to bash 4 if you can
- Fixed the macro OP_ASSERT_EQUAL_BUFFER. If it was used more than one time in the same test this was causing compilation error (multiple definitions of the same variable)
New in opmock 0.8.1 (Nov 24, 2011)
- Fixed another code generation error when an operation has a prototype like const int operation(...) const; The const at the beginning was causing problems.
New in opmock 0.8 (Nov 15, 2011)
- Fixed a code generation error when an operation has a prototype like int operation(...) const;
- The const at the end was causing problems.
- Added a few macros in opmock.h for unsigned and signed types (short, ushort, long, ulong, uchar).
New in opmock 0.7.9 (Oct 24, 2011)
- Fixed a nasty code generation problem : functions/methods with default values for arguments were resulting in incorrect generated code.
- This unveiled additional (now fixed) bugs with bad generation of namespaces.
New in opmock 0.7.8 (Oct 20, 2011)
- Fixed a problem with the OP_VERIFY* macros. They must fail the test if a mock has been configured but not called often enough in the scope of a test. This is now fixed.
New in opmock 0.7.7 (Oct 20, 2011)
- Fixed a code generation bug when using twice the const keyword in a function argument, for example 'const unsigned long long ** const param'.
- Cleaned up a little bit the samples.
New in opmock 0.7.6 (Oct 12, 2011)
- Fixed a bug in cmp_buffer : his was returning 0 (no error) if the first byte of 2 buffers was different.
New in opmock 0.7.5 (Oct 1, 2011)
- Modified the opmock.sh script and added a new example for generation of mocks with complex header includes. This fixes the situation where: -types declarations in the same name space are split in multiple headers Opmock was not generating proper code in this case. have a look to the sample recursive_include, don't forget to update your opmock.sh, and have a look to the documentation, section 9.
New in opmock 0.7.4 (Oct 1, 2011)
- Improved some assertions, added UINT, BYTE, UBYTE types to print numerical values instead of char values. Added a OP_VERIFY_NO_ORDER macro to allow running the verify phase but avoid failing the test if operation calls were not made in the right order. Some documentation updates.
New in opmock 0.7.3 (Oct 1, 2011)
- Added call order verification when using a mock. For example, if your code is supposed to call lock() THEN unlock(), but it actually calls unlock() then lock(), the VERIFY macro will fail. This can be used as well with google test or anything else : just check if there are errors after a test is run.
New in opmock 0.7.2 (Sep 13, 2011)
- Added a cmp_short comparator.
New in opmock 0.7.1 (Sep 13, 2011)
- Added a Junit compatible xml output for tests suites, when the Opmock unit testing framework is used. Creates a TEST-test.xml file in the test folder. Of course, the junit xml format has many shortcomings with C/C++ as it insists on having classes or packages, but it should be compatible with common CI servers like Hudson.
New in opmock 0.7 (Sep 13, 2011)
- Version bump! There are some breaking changes in this release, so it's worth it. All code/options related to hybrid mocks (where you could have both the original implementation and a mocked implementation of a function/class in the same build) have been removed. They were not used and were not a very good idea to start with. Less crap in the code. The code generation for C++ has changed : you don't need any more to modify your existing header files to insert 'renaming' macros. If your headers have been modified and use macros like OPMOCK_CLASS, simply remove the macros. The side effect is that you will have to modify a bit existing tests.
- Example:
- Before: foo:boo:MyClass::myFunction_ExpectAndReturn(...);
- After: foo:boo:MyClass_Mock::myFunction_ExpectAndReturn(...);
- This is the same, except that you need to post fix the class name with _Mock. Another change is a simplification for overloaded operations or functions : all overloaded versions of an operation are mocked the same way, as the different parameters are enough to distinguish them. Before, there was a specific numbering for each overload. Last, the typedefs for call backs have been modified a bit. When implementing call backs for overloaded functions/operations, the typedefs will have different names, with a counter appended. Be sure to implement the correct signature by looking at the generated hpp files.
New in opmock 0.6.3 (Sep 13, 2011)
- bug fix : free functions in C++ with default parameter values were resulting in incorrect generated code. This is now handled as a function overload. Removed the *extern* keyword in generated function body code.
New in opmock 0.6.2 (Sep 13, 2011)
- bug fix : static operations in a class were resulting in incorrect generated code.
New in opmock 0.6.1 (Sep 13, 2011)
- Added a script to generate automatically your test suites and main.c files. Have a look to the documentation and to the test_gen example for detailed instructions.
New in opmock 0.6 (Sep 13, 2011)
- There are so many changes in this release, thank's to user feedback, that it deserves a version bump ! The internal structure of the generated code has been modified quiet a lot. The good news is that this makes the tests simpler: There's no need to call manually any mock reset function anymore. All mocks are reset globally before running a test. If you use Reset functions in your tests, have a quick look to the manual. You'll see that things are much simpler now.
- There's no need to call manually any mock verify function anymore. All mocks are verified globally after running a test. This means as well that there are no more side effects from one test to the other when a mock is not properly configured (for example, called the wrong number of times).
- The OP_VERIFY() macro does not take parameters anymore. Calling this macro at the end of your test means that you want to fail the test if a mock error has been detected (bad number of calls, bad parameters). Not calling this macro means that you're not interested in the verify phase.
- Lots of changes in the test suite
- Lots of updates in the documentation
- Miscellanous syntax changes. Have a look to the documentation for details. Many thanks go to Glenn Waters, who contributed a significant part of these patches.
New in opmock 0.5.7 (Sep 13, 2011)
- Fixed a regression: opmock was crashing when generating code with a C function taking a void parameter, like foo(void);
New in opmock 0.5.6 (Sep 13, 2011)
- Improved the way the verify macros and functions work.
- Now, if one calls OP_VERIFY(function) this will fail the test:
- if the number of calls to the mock was not correct
- or if at least one of the calls to the mock had wrong parameters
- or if the mock was called with no configured behavior (ie empty call stack)
- Using this macro/these functions is still optional, but can help when mocks are used rather than callback functions.
New in opmock 0.5.5 (Sep 13, 2011)
- Fixed 2 bugs:
- generate correct code when functions use anonymous parameters
- generate correct code when a function parameter is a function pointer AND has a void parameter.