Age | Commit message (Collapse) | Author | Files | Lines |
|
|
|
|
|
|
|
BUG #60
|
|
|
|
Signed-off-by: Andreas Schneider <asn@cryptomilk.org>
|
|
|
|
Reviewed-by: Andreas Schneider <asn@cryptomilk.org>
|
|
|
|
Reviewed-by: Andreas Schneider <asn@cryptomilk.org>
|
|
|
|
Signed-off-by: Andreas Schneider <asn@cryptomilk.org>
|
|
|
|
CID 1365415
Signed-off-by: Andreas Schneider <asn@cryptomilk.org>
|
|
Reviewed-by: Andreas Schneider <asn@cryptomilk.org>
|
|
In that system while the maximum word size is 64-bits, the
'unsigned long int' type is only 32-bits.
Reviewed-by: Andreas Schneider <asn@cryptomilk.org>
|
|
|
|
|
|
This is for multiple groups in one test to make sure they do not create
invalid xml.
Signed-off-by: Andreas Schneider <asn@cryptomilk.org>
|
|
Signed-off-by: Andreas Schneider <asn@cryptomilk.org>
|
|
Signed-off-by: Andreas Schneider <asn@cryptomilk.org>
|
|
On some platforms PRIu64 is not necessarily the width of the
LargestIntegralType. A new decimal format specifier for
LargestIntegralType was added and replaces the invocations of PRIu64.
|
|
Signed-off-by: Andreas Schneider <asn@cryptomilk.org>
|
|
Reviewed-by: Andreas Schneider <asn@cryptomilk.org>
|
|
Signed-off-by: Andreas Schneider <asn@cryptomilk.org>
|
|
CMake assigns "lib" as the prefix to shared libraries on MinGW
GCC, but MinGW does not expect such a prefix when linking to
DLLs. By setting, CMAKE_SHARED_LIBRARY_PREFIX to "" the
example tests were able to run with the exception of the
calculator_test which segfaults (SIGSEGV) on the setjmp in
test_divide_by_zero. This fault is specific to MinGW and does not
occur within runs on Cygwin or Visual Studio.
calculator_test was enabled only for static linking on MinGW
Reviewed-by: Andreas Schneider <asn@cryptomilk.org>
|
|
Visual Studio now includes support for vsnprintf and snprintf but does
so as an inline functions. check_symbol_exists is used to check for
its existence within stdio. [3] All tests are passing when run from
within the IDE or via NMake. Command line builds via msbuild seem to
error for the first few example tests.
Reviewed-by: Andreas Schneider <asn@cryptomilk.org>
|
|
As both parameter and function call order checking allow for ignoring
cases where they are never invoked, the mock return values are at
somewhat of a mismatch in that they must always be returned at least
once (even in the case of will_return_always()). Therefore, the ability
to set the count to -2 on will_return_count was added with a new macro
(will_return_maybe) that indicates that that the value field may never
be returned and still allow a successful test.
Reviewed-by: Andreas Schneider <asn@cryptomilk.org>
|
|
Reviewed-by: Andreas Schneider <asn@cryptomilk.org>
|
|
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Andreas Schneider <asn@cryptomilk.org>
|
|
Signed-off-by: Andreas Schneider <asn@cryptomilk.org>
|
|
Signed-off-by: Andreas Schneider <asn@cryptomilk.org>
|
|
Reviewed-by: Andreas Schneider <asn@cryptomilk.org>
|
|
|
|
Signed-off-by: Andreas Schneider <asn@cryptomilk.org>
|
|
Signed-off-by: Andreas Schneider <asn@cryptomilk.org>
|
|
Looks like it isn't clear for some people reading that file that we
support Windows and MSVC ...
|
|
BUG: https://open.cryptomilk.org/issues/43
|
|
|
|
|
|
Sometimes multiple test cases share the same test function, running it
on different data. To pass this data to test functions we must define
setup function for each data set. It's not very convienient when there
are many states to test.
This commit introduce more elegant way to pass data to tests. The
initial_state field of CMUnitTest structure can be defined by user.
It will be either passed to setup function, which makes any preparation
needed, or it will be passed directly to test function when setup func is
NULL.
Signed-off-by: Pawel Szewczyk <p.szewczyk@samsung.com>
|
|
|
|
Reviewed-by: Andreas Schneider <asn@cryptomilk.org>
|
|
Without using siglongjmp, the signal mask is not restored when
longjmp-ing from the signal handler, and whichever signal was
being handled remains blocked for the remainder of the tests. As a
result, the same signal cannot be caught twice, and will cause process
termination when being raised the second time.
This does not change 'jmp_buf global_expect_assert_env', because it is
part of the public API, and isn't used from a signal handler.
Reviewed-by: Andreas Schneider <asn@cryptomilk.org>
|
|
Reviewed-by: Jakub Hrozek <jakub.hrozek@posteo.se>
|
|
|
|
Thanks to Gregory Pakosz.
Signed-off-by: Andreas Schneider <asn@cryptomilk.org>
|
|
Signed-off-by: Andreas Schneider <asn@cryptomilk.org>
|
|
|
|
|