correct Xcode generator Swift definitions
original code was defining GCC_PREPROCESSOR_DEFINITIONS which is valid only for C languages
add definitions to SWIFT_ACTIVE_COMPILATION_CONDITIONS when Swift language is used in the target
add test in SwiftOnly
for old Xcode (<8.0), append defines to cflags so it ends up in OTHER_SWIFT_FLAGS
Fixes: #23637
C++ modules have two variants which are of importance to CMake:
- `CXX_MODULES`: interface modules (those using `export module M;`,
`export module M:part;`, or `module M:internal_part;`)
- `CXX_MODULE_HEADER_UNITS`: importable header units
Creating C++ modules or partitions are *not* supported in any other
source listing. This is because the source files must be installed (so
their scope matters), but not part of usage requirements (what it means
for a module source to be injected into a consumer is not clear at this
moment). Due to the way `FILE_SET` works with scopes, they are a perfect
fit as long as `INTERFACE` is not allowed (which it is not).
Rename the booleans 's_ErrorOccured' and 's_FatalErrorOccured' to
's_ErrorOccurred' and 's_FatalErrorOccurred', respectively.
Rename the getters and setters to 'Get[Fatal]ErrorOccurred' and
'Set[Fatal]ErrorOccurred', and fix all uses across the codebase.
When determining a given target's object directory, also check for its
`OSX_ARCHITECTURES` before resorting to global defaults. This fixes inconsistent
object library references when:
- `CMAKE_OSX_ARCHITECTURES` is unset or singular
- but the object library's `OSX_ARCHITECTURES` property is set to multiple archs
Allow `cmGlobalGenerator`s to decide `HasKnownObjectFileLocation()` per given
`cmTarget`
- `cmGlobalGenerator::HasKnownObjectFileLocation()` now takes an optional `cmGeneratorTarget`
- `cmTarget::HasKnownObjectFileLocation()` added as a shorthand
We evaluate `LINK_LIBRARIES` and `INTERFACE_LINK_LIBRARIES` for two purposes:
* Constructing the link line.
* Collecting usage requirements.
We evaluate `INTERFACE_LINK_LIBRARIES` separately for each purpose in
order to support the `$<LINK_ONLY:...>` generator expression used to
express private link dependencies of a static library. Previously we
only evaluated `LINK_LIBRARIES` for linking, and used that result for
collecting usage requirements too. Therefore `$<LINK_ONLY:...>` does
not work in `LINK_LIBRARIES`.
With the introduction of `INTERFACE_LINK_LIBRARIES_DIRECT`, evaluation
of `LINK_LIBRARIES` now needs to distinguish these two cases in order to
honor link dependencies encountered through `$<LINK_ONLY:...>` without
also exposing other usage requirements through private dependencies of a
static library. Revise internal infrastructure to distinguish the two
cases when evaluating `LINK_LIBRARIES`. Make the information available
in code paths for `INTERFACE_LINK_LIBRARIES_DIRECT` and `LINK_ONLY`.
Defer actually using the information to later commits.
Issue: #22496
Since commit 61495cdaae (Fix Xcode project references to the source
tree, 2009-09-22, v2.8.0~43) we force source file references to use
relative paths from the source tree. If the source tree path is a
symbolic link inside the build tree, the relative `../` sequence
goes to the wrong place. The problem with debug breakpoints motivating
that change does not seem to occur in modern Xcode versions, so update
the logic to use a relative path only when it does not need to start
in any `../` sequence.
This generator expression offers the capability, for the link step, to
decorate libraries with prefix/suffix flags and/or adding any specific flag for each
library.
Fixes: #22812, #18751, #20078, #22703
Check the complete include path for being a system include, not
the derived framework search path. The code for Ninja and Makefile
generators does exactly the same.
Fixes: #23011
Make `cmCustomCommand` have just only default constructor.
Use each setter instead. This follows the builder pattern.
Introduce `cc::SetOutputs(std::string output)`.
This will be used later, as substitution for `cc::SetOutputs({output})`.
In commit 8d5f4c4db9 (Xcode: Switch to the "new build system" for Xcode
12 and above, 2020-09-14, v3.19.0-rc1~143^2~7) we accidentally added
code in an `else` block that under the opposite condition by which
the block can be entered. Remove it.
Fixes: #22681
Most calls to `MaybeConvertToRelativePath` use one of our common work
directories (e.g. top of the build tree) as the local path. Add helpers
for each of the common cases to simplify and clarify call sites.
These have been added to:
GCC_PREPROCESSOR_DEFINITIONS
OTHER_CFLAGS
OTHER_LDFLAGS
This is to allow Cocoapods to work correctly as it uses xcconfig files to alter build settings in Xcode, and requires these build settings to inherit from their parent, not overwrite.
In commit ce2dee9e5b (Xcode: Don't add framework as -framework argument
in linker info list, 2020-09-28, v3.19.0-rc1~47^2) we split up the path
to a framework into the directory and framework name parts, but only
retained the quoting on the directory part. Restore quoting of the
framework name.
Fixes: #21910
Make the `config` argument non-optional so all callers must be explicit.
Convert the path style argument to an enumeration to make its role clear
at call sites.
The path style argument is implemented by `ConvertToIncludeReference`,
which was introduced with the Ninja generator by commit 5b114c9bee
(Introduce a cmLocalGenerator::ConvertToIncludeReference function,
2011-09-07, v2.8.7~187^2~4). Its only purpose is to allow the Ninja
generator to use relative paths in `-I` flags. Add a comment explaining
this role.
5389bb4274 Xcode: Don't hard-code SDK-provided implicit framework search paths
df08f8df30 cmComputeLinkInformation: Fix misspelt private variable name
375b307bae Apple: Fix linking to frameworks that do not exist until build time
Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !5760
When a framework is linked to a target by its full path and that
framework is located in one of the implicit framework search directories,
CMake 3.18.5 and earlier discarded that path.
ce2dee9e5ba (Xcode: Don't add framework as -framework argument in
linker info list, 2020-09-28) introduced a regression which resulted in
the framework path always being added to the search path even if it
matched one of the implicit search paths. This broke the ability to do
device and simulator builds from the same configured project.
Fixes: #21678
The Xcode "new build system" only considers a script build phase up to
date if it has run before, even if outputs are newer than inputs. Use a
deterministic object id for script build phases associated with custom
commands so that they do not need to re-run after CMake re-generates the
project.
Fixes: #21669
