Add a TIMEOUT option to gtest_discover_tests. This provides a
work-around in case a test goes out to lunch, rather than causing the
build to simply hang. (Although this is still a problem with the user's
project, hanging the build is not cool, especially in the case of
automatically running CI builds. It is much preferred that the build
should actively fail in this case, and it is trivially easy for us to
implement that.)
Due to a erroneous logical dependency on a temporary variable, FindMPI
would depend its FOUND status on the set of flags needed to support
compiler-implicit MPI support after the first configure run.
This behavior is especially causing a misreported NOT FOUND status after
the first run on MPI implementations without MPI compiler like MS-MPI or
MPICH2 on Windows.
Due to a minor logic error, MPI_LIBRARY, MPI_EXTRA_LIBRARY and
MPI_<LANG>_LIBRARIES hints are currently not correctly handled and
discarded.
Additionally, if either C or CXX aren't enabled, there subsequent calls
to FindMPI will cause cache variables of the sort "MPI_MPI_EXTRA_LIBRARY-NOTFOUND"
to appear due to an error in the compatibility handling. (Doesn't affect
functionality)
The check added by commit v3.10.0-rc2~2^2 (Clang: Diagnose unsupported
GNU-like clang targeting MSVC ABI, 2017-10-10) is incorrectly detecting
clang-cl 6.0 as GNU-like. Currently cmake is testing if the clang
compiler accepts `--version` to see if it accepts GNU style flags.
However, with the latest llvm snapshot this also works for clang-cl:
> clang-cl --version
clang version 6.0.0 (trunk)
Target: x86_64-pc-windows-msvc
Thread model: posix
InstalledDir: C:\Program Files\LLVM\bin
So instead we should use the `/?` flag which fails with clang but
works with clang-cl:
> clang-cl /? &> /dev/null; echo $?
0
> clang /? &> /dev/null; echo $?
1
Fixes: #17518
According to the documentation, tests can be discovered for a target
multiple times by using a different prefix and/or suffix to ensure name
uniqueness. However, while this worked for gtest_add_tests, it did not
work with gtest_discover_tests because the generated file that sets up
the tests was named based only on the target name, and so subsequent
discovery from the same target would clobber earlier discovery.
Fix this by introducing a counter that records how many times discovery
has been used on a target, and use this to generate unique names of the
generated test list files.
Since commit v3.10.0-rc5~3^2 (FindOpenGL: Default to non-GLVND libraries
for legacy GL, 2017-11-08) users may set `OPENGL_gl_LIBRARY` to empty to
use GLVND components for the legacy GL interfaces. This is useful only
when one knows in advance that the GLVND components will be found.
Add a `OpenGL_GL_PREFERENCE` variable to specify a preference for legacy
GL or GLVND. The latter can suppress `OPENGL_gl_LIBRARY` only when the
needed GLVND components are found. If no preference is explicitly
specified, choose a default based on whether GLVND components were
requested (because this indicates the project has been updated for
CMake 3.10).
Issue: #17437
Issue: #17449
`DOXYGEN_DOT_FOUND` is only set if `_Doxygen_keep_backward_compat` is
used (when no components are requested), so use `Doxygen_dot_FOUND`
directly. Preserve the "YES" or "NO" value used previously.
Projects using `OPENGL_LIBRARIES` or `OpenGL::GL` expect legacy GL.
Although GLVND OpenGL+GLX provides legacy GL interfaces, using those
library files may conflict with legacy GL library files used by
dependencies (or dependents) of such projects. Therefore we should
not yet use OpenGL+GLX when a legacy GL library is available.
If `OPENGL_gl_LIBRARY` is set then use it as the legacy GL library.
If it is *not* set then fall back to using GLVND OpenGL+GLX to provide
legacy GL interfaces. This will allow users to build projects using
GLVND even if they have not been ported.
Fixes: #17437
Fix a typo from commit v3.6.0-rc1~85^2 (HDF5: Refactor the use of
compiler wrappers, 2016-04-04) that accidentally used the HDF5 C++
version (which could be empty) while detecting the C patch version.
The detection failed for patch releases, such as `1.10.0-patch1`
becoming `.1` instead of `1.10.0.1`.
Since commit v3.9.0-rc1~148^2 (Do not assume GCC libs are linked by all
compilers, 2017-05-05) we no longer filter out all `gcc*` implicit link
libraries. This allows mixing of gcc and non-gcc compilers across
languages. However, this caused a subtle problem with how GCC makes
exception handling symbols available to linked binaries.
GCC (at least on MinGW) provides two different libraries with exception
handling symbols:
* gcc_s: A shared library with -fvisibility=default, used by -shared-libgcc.
* gcc_eh: A static library with -fvisibility=hidden, used by -static-libgcc.
The C compiler (on MinGW) defaults to -static-libgcc and uses gcc_eh.
The C++ compiler defaults to -shared-libgcc and uses gcc_s when linking
shared libraries and executables so that exceptions can propagate across
shared libraries [1]. When linking a mixed-language binary, the C++
compiler should be used along with its choice of gcc_s. In this case
gcc_eh should not be added even though the C compiler implies it because
gcc_s supersedes it.
Since the above-mentioned change, CMake is adding gcc_eh to C++ link
lines that also contain C code on MinGW. This causes both gcc_s and
gcc_eh to be used, which is incorrect. We can fix this simply by
excluding gcc_eh from the C compiler's implicit link libraries.
[1] https://gcc.gnu.org/onlinedocs/gcc-7.2.0/gcc/Link-Options.html#Link-OptionsFixes: #17436
When we run `csc.exe /help` we look for "Version" in the output.
Explicitly ask for the output in English.
Reported-by: guttally@users.noreply.github.com
This was done for C++ by commit v3.5.0-rc1~69^2 (Record compile features
for MinGW Clang on Windows, 2016-01-11). Make the same change for C.
The `UNIX` condition on Clang C compiler features was already dropped by
refactoring in commit v3.9.0-rc1~17^2~4 (Compilers: Port to use default
cmake_record_lang_compile_features macros, 2017-05-10).
Our documentation already claims support for this combination. This
was simply an oversight when support was added for MinGW Clang C++.
Issue: #15897
Issue: #15943
Since commit v3.9.0-rc1~224^2~1 (FindMPI: MPIEXEC handling improvements,
2017-04-21) the `ProcessorCount` module is being used to initialize
`MPIEXEC_MAX_NUMPROCS`. However, this leads to the logical cores being
counted rather than the physical ones, and some MPI implementations like
OpenMPI will error if mpiexec is called with that number. Switch it to
the number of physical cores using `cmake_host_system_information`.
This ensures that if `MPIEXEC_MAX_NUMPROCS` is being used to set up MPI
tests with CTest or similar that the tests won't spuriously fail due to
OpenMPI refusing to start the application.
`xmlrpc-c-config $modules --cflags` may report no `-I` flags if the
headers are in a standard location like `/usr/include`. In this case
it is okay for `XMLRPC_INCLUDE_DIRS` to be empty.
Ideally the `--cflags` output should be used as hints for a `find_path`
call to really find the headers as we do for other find modules, but
simply assuming no explicit include directories are needed is good
enough for now.
Fixes: #17347
