Encourage placing preset includes near the beginning of a preset
file and ensure the example shows that usage. Move the prose
discussing includes to its own section to improve discoverability
and break up paragraphs to make each main point harder to miss.
Also clarify ${sourceDir} to remove any ambiguity with regard to
its meaning in included files.
Issue: #23214
Changes in commit 8f64df0a7c (CUDA: Generic all and all-major support,
2021-12-19, v3.23.0-rc1~23^2) broke our architecture verification checks
when using `-arch={all,all-major}` with NVCC 11.5+. If we test the
compiler with `-arch={all,all-major}`, we have no expected list of
architectures, so skip the check.
Fixes: #23278
Since commit 8f64df0a7c (CUDA: Generic all and all-major support,
2021-12-19, v3.23.0-rc1~23^2), these special values do not require any
specific compiler or version.
While at it, re-organize CUDA_ARCHITECTURES special value documentation.
Move the `versionadded` markup inside each special value so we can add
more in future versions.
Refactor the logic checking `CMAKE_CUDA_ARCHITECTURES` special values.
Switch on the value first, and then make other decisions for each case.
This makes room for other special values to be added later.
There's no point printing a blank line if FindHeaderInFramework()
returns an empty string. Pass in the debug object to that function
instead so it can record the actual paths it tests in the debug output
directly.
Revert commit 5efb6fb516 (FindThreads: Honor THREADS_PREFER_PTHREAD_FLAG
when pthread is found in libc, 2021-11-03, v3.21.5~4^2). The check for
the `-pthread` flag can pass on compilers like XL, that interprets it as
`-p -t hread` and returns zero. Prior to that commit, we did not use
the check in the `CMAKE_HAVE_LIBC_PTHREAD` code path. Now we do, it
succeeds, and we incorrectly add the `-pthread` flag for XL.
This change was backported to the 3.21 and 3.22 release series long
after they initially came out. Since there may be more cases where we
now add `-pthread` incorrectly, it is simplest to revert the change in
all release series pending further investigation.
Fixes: #23270
The list capability was added in CMake 3.15 by commit fec441ec17 (Teach
CROSSCOMPILING_EMULATOR to support arguments, 2019-05-30,
v3.15.0-rc1~6^2), but the documentation did not indicate the version
that added it.
Fixes build project error because the OS directory was a relative
path from the binary directory.
Fixes issue during project regeneration where the `-os_dir` project option
was missing because GHS_OS_DIR was not properly stored in the cache
file.
This error was introduced by commit aa59eaac4f (GHS: Allow setting GHS
variables from toolchain file, 2021-11-15, v3.23.0-rc1~337^2).
Since commit 8f64df0a7c (CUDA: Generic all and all-major support,
2021-12-19, v3.23.0-rc1~23^2), setting `CUDA_ARCHITECTURES` to a value
other than `all` or `all-major` is ignored with NVCC 11.5+. Fix the
logic to return early only when actually using an "all" value.
Fixes: #23243
When given two source paths via `-S` or just directory paths prefer
the last one. When the paths are mixed always prefer the last `-S`
entry.
Fixes: #23238
Add missing `CMake_TEST_Qt{4,5}` conditions on Qt4 an Qt5 tests.
Configuring CMake with `-DCMake_TEST_Qt5=OFF` should prevent any
attempt to search for Qt5.
Fixes: #23239
f73457ca2e cmake: Ignore any empty "" command line arguments
67f97f5478 Tests: Add RunCMake helper to run cmake with raw execute_process args
Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !6980
libarchive's crypto library checks use its `config.h` inside the
`try_compile` project. Since commit ade3b16e63 (libarchive: Use KWIML
to get fixed-size integer types, 2020-06-01, v3.18.0-rc1~33^2),
that header depends on KWIML inside CMake. Add the include directory
for KWIML to the crypto library checks. Otherwise, they always fail
due to not finding the KWIML headers, and libarchive decides not
to link the crypto library.
libarchive has other code besides the hash algorithms that depends on
the crypto library if its ENABLE_OPENSSL option is enabled (which in
CMake is controlled by CMAKE_USE_OPENSSL). It seems to be missing some
conditions to link the crypto library in those cases, and instead relies
on at least one of the above-mentioned checks to pass. If they all
fail, and we are using system curl, we might not link the crypto
library.
Fixes: #23234
