Prerequirements should be searched for in
generator prerequirements function and only
checked for presence in configure file in
per test prerequirements function.
CPack tests can now define one or more
packaging types and that test will be
executed once per generator per packaging
type. This also enables us to set default
values per packaging type for each generator
which makes tests shorter.
Fakeroot executable is already searched
for in DEB packager tests prerequirements
so the test prerequirement just has to
check that it is present in config file
and not search for it once more.
Content list for a certain test
should be the same between different
generators so whenever possible use
a single expected files list per
file for all generators.
Each generator generates its own file
format with different extensions but
they all create those file names by
a certain pattern and all of them
require the same pieces for that pattern
to work. Now we can only provide pieces
once and and each generator will use those
pieces to generate an expected file name.
Package names are irrelevant for most
CPack tests so tests now set default
package name that is created from test
name and used if it is not explicitly
provided in the test.
This causes the `-Wno-deprecated` option to be honored even inside a
`try_compile` test project, which is needed to suppress all deprecation
warnings as the option documents.
Closes: #16446
80ebc55a cmake: Report if the <LANG>_CLANG_TIDY tool exits with non-zero
ce1abfa4 cmake: If ldd for LINK_WHAT_YOU_USE fails to run then report why
44de6157 cmake: Comment why we ignore the include-what-you-use return code
When using `<LANG>_CLANG_TIDY` our internal launcher for the tool must
capture its return code and stderr and report them on failure.
Otherwise incorrect command lines silently fail.
Closes: #16435
The include-what-you-use tool always returns non-zero to indicate that
it did not actually produce an object file as Clang would from the same
command line. Add a comment explaining that this is why we ignore its
return code. Also update our `pseudo_iwyu` test suite tool to always
exit with an error too.
Users or scripts commonly remove or replace `CMakeCache.txt` without
also removing `CMakeFiles/`. In this case the information saved in the
cache from platform information initialization is missing, so we need to
re-initialize it. In such a case, remove the platform information
directory so that re-initialization will occur and restore needed
information to the cache.
Closes: #14820
With VS generators for 2010 and above, passing a `.targets` file to
`target_link_libraries` is expected to generate content in the
`.vcxproj` file to import the targets file. Add a test to cover this.
f636d1e7 Help: Add notes for topic 'add-SHA-3'
cb5dba8e Tests: Add SHA-3 algorithm coverage to CPack checksum test
a9fa6099 ExternalProject: Add support for SHA-3 algorithms
b4ffd26f ExternalData: Add support for SHA-3 algorithms
60939702 file: Add support for SHA-3 algorithms
92f95aed string: Add support for SHA-3 algorithms
c326209f cmCryptoHash: Add support for SHA-3 algorithms
979e8ba6 Help: De-duplicate and spell out supported hash algorithms
7a79f7e3 librhash: Activate SHA-3 implementation
87584b5e Merge branch 'upstream-librhash' into add-SHA-3
7fcbd47e librhash 2016-11-06 (de79828d)
c50c32a3 librhash: Update import script to add SHA-3 implementation
Expose the binutils' machine name (typically used as a prefix on the
tool names) publicly. This is expected to match the `gcc -dumpmachine`
value.
Suggested-by: Ruslan Baratov <ruslan_baratov@yahoo.com>
The NDK provides prebuilt toolchain files in directories named for the
host architecture. The NDK build system calls this `HOST_TAG`.
Expose the value publicly for use by clients that need to pass it
to external tools.
Suggested-by: Ruslan Baratov <ruslan_baratov@yahoo.com>
When this variable is not set by the user or toolchain file, set it to
the default selected. This will be useful for client code that needs to
pass the value to an external tool that needs to find the same toolchain
in the NDK. Leave it empty for a standalone toolchain.
Suggested-by: Ruslan Baratov <ruslan_baratov@yahoo.com>
09cda9d5 Allow imported INTERFACE libraries to specify a link library name
1d1f1eeb cmTarget: Refactor GetMappedConfig to choose location property up front
479932fa cmTarget: Add comment clarifying interface library special case
925e4270 cmTarget: Clarify comments in GetMappedConfig
Add an `IMPORTED_LIBNAME[_<CONFIG>]` target property to specify a library
name to be placed on the link line in place of an interface library
since it has no library file of its own. Restrict use of the property
to imported `INTERFACE` libraries.
This will be particularly useful for find modules that need to provide
imported libraries from system SDKs where the full path to the library
file is not known. Now such find modules will be able to provide an
imported interface library and set `IMPORTED_LIBNAME` to refer to the
SDK library by name.
Issue: #15267
Suggests tag was not present in older versions
of rpmbuild so we test that the rpm package is
always generated either with Suggests tag present
or skipped if not supported.
The `cxx_static_assert` feature may be available in C++ 98 mode of some
compilers or not available at all in others. Intstead of using an
individual feature to test cyclic requirement of a feature requiring
C++ 11, use the `std_cxx_11` meta-feature that has exactly this meaning.
f648b9be Tests: Check that CROSSCOMPILING_EMULATOR is not used on imported targets
e7480d67 Fix custom command target substitution with CROSSCOMPILING_EMULATOR
In commit v3.6.0-rc1~88^2 (CustomCommandGenerator: Add support for
CROSSCOMPILING_EMULATOR, 2016-05-04) logic was introduced to substitute
a target's `CROSSCOMPILING_EMULATOR` for argv0 in a custom command.
However, it broke the case when the argv0 was a target name and now
fails to expand the target name to its location at the same time as
inserting the emulator. Fix the latter case.
Inspired-by: Brian Maher <brian@brimworks.com>
Closes: #16288
The targets added by commit v3.6.0-rc1~240^2~2 (Ninja: Add `$subdir/all`
targets, 2016-03-11) use as `$subdir` the relative path from the top of
the source tree to the current source directory. This is not correct
when using `add_subdirectory(test test_bin)`. Instead we need to use
the relative path from the top of the binary tree to the current binary
directory as was done for related targets by commit v3.7.0-rc1~268^2
(Ninja: Add `$subdir/{test,install,package}` targets, 2016-08-05).
