5fa15ec9f3 Help: Document that target_sources defines [INTERFACE_]HEADER_SETS
c5d4812f20 cmTarget: make HEADER_SETS and INTERFACE_HEADER_SETS read-only
05783b168d cmFileSet: store visibility with the fileset
Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !7168
There is no reason to allow these properties to be manipulated by user
code. Instead, use the stored visibility on the fileset objects to
derive what these properties should contain.
When the Apple linker sees -headerpad_max_install_names and
bitcode is enabled with a flag like -fembed-bitcode, it issues a warning
and ignores the -headerpad_max_install_names flag. This causes
unrelated compiler and linker flag checks to fail for valid flags.
In f745e0497e (CheckCompilerFlags: Catch linker warning about ignored
flags, 2022-01-03), we started detecting linker warnings, which caused
a regression for projects that were setting -fembed-bitcode in their
CMAKE_CXX_FLAGS or similar. Prevent that regression by removing
the -headerpad_max_install_names linker flag when we know it will
warn and be ignored anyway.
Fixes: #23390
Issue: #23408
The feature needs a specialized implementation to place headers
in the right place inside frameworks. To avoid silently doing
the wrong thing, make this case an error for the 3.23 series.
Issue: #23386
In commit afcd9fe669 (AIX: Add an option to disable automatic exports
from shared libraries, 2020-01-30, v3.17.0-rc1~47^2) the population of
the `<AIX_EXPORTS>` rule variable placeholder was accidentally added to
the device linking rule rather than the main linking rule. This caused
our `ExportImportList` script on AIX, when called for executables with
`ENABLE_EXPORTS` set, to be given an `AIX_EXPORTS` file name that does
not exist, leading to a warning from the `dump` tool.
Move the population of the `<AIX_EXPORTS>` placeholder in the Makefile
generators to the main link rule.
Issue: #20290
Remove the requirement that the variable name have a prefix while
keeping the suffix requirement. Require that the property name
contains an underscore. Update docs and tests accordingly.
Fixes: #23340
Revert commit f3ad061858 (Add usage requirements to update direct link
dependencies, 2022-01-12, v3.23.0-rc1~44^2) and the property storage
updates in its predecessor commit 193a999cd5 (cmTarget: Add
INTERFACE_LINK_LIBRARIES_DIRECT{,_EXCLUDE} backtrace storage,
2022-01-06, v3.23.0-rc1~44^2~1) from the 3.23 release branch.
After initial experience using the feature in practice, additional
design considerations have been raised for discussion in the original
issue. To avoid rushing this for the 3.23 series, we've decided to
revert the feature for now so it can be revised for a future release.
Issue: #22496
There's no benefit to storing the CPACK_PRODUCTBUILD_DOMAINS
prefix in a variable and appending to it in the C++ code. It has the
disadvantage of making it harder to find usages of the variables with
a suffix appended to that string. Expand out the strings at the places
they are used so that they are easier to spot.
Since commit c1e812ad4f (target_link_libraries: Improve tolerance of
unquoted generator expressions, 2022-02-15, v3.23.0-rc2~11^2) we
accumulate consecutive non-keyword arguments to recover an unquoted
generator expression as a single entry. When given multiple consecutive
non-genex library names, the grouping breaks our logic that expects each
entry is either a raw target name or a genex. Revise the logic to only
accumulate multiple arguments when they end inside a partial genex.
This bug caused `target_link_libraries` to stop wrapping static library
private dependencies in `$<LINK_ONLY:...>` for `INTERFACE_LINK_LIBRARIES`
when multiple consecutive library names are given. Add a test case
covering that behavior.
Fixes: #23302
Since commit 880ca66b51 (Fix: `while()` can silently ignore incorrect
condition, 2021-08-09, v3.22.0-rc1~238^2~4) we correctly reject the
code
set(paren "(")
while(${paren})
endwhile()
However, rejecting it breaks compatibility with projects that used such
code accidentally. In CMake 3.21 and below, any error in the condition
was ignored because the `false` result exited the loop first. Restore
tolerance of the error for now. A policy will be needed to make it an
error later.
Note that the same condition with `if` was always correctly rejected.
Fixes: #22524
Issue: #23296
Co-authored-by: Brad King <brad.king@kitware.com>
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.
