5ece12b7e4 gitlab-ci: add ISPC to the Fedora CI image
8976817d6d ISPC: Update help documentation to include ISPC
2368f46ba4 ISPC: Support building with the MSVC toolchain
e783bf8aa6 ISPC: Support ISPC header generation byproducts and parallel builds
34cc6acc81 Add ISPC compiler support to CMake
419d70d490 Refactor some swift only logic to be re-used by other languages
Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !5065
Add an `OPTIMIZE_DEPENDENCIES` target property and supporting
`CMAKE_OPTIMIZE_DEPENDENCIES` variable to optionally enable pruning and
flattening of outgoing dependencies from static libraries. Since they
do not actually link, they only depend on side effects of their
dependencies. Therefore we can drop dependencies that contribute no
side effects.
Since commit f593b354da (PCH: Add support for multi architecture iOS
projects, 2020-04-02, v3.18.0-rc1~414^2) the `OSX_ARCHITECTURES` target
property (and corresponding `CMAKE_OSX_ARCHITECTURES` variable) affects
flags on non-Apple platforms by accident. Add a missing condition to
avoid this.
Fixes: #21072
INTERFACE libraries were created with the intention of collecting usage
requirements for use by other targets via `target_link_libraries`.
Therefore they were not allowed to have SOURCES and were not included in
the generated buildsystem. In practice, this has become limiting:
* Header-only libraries do have sources, they just do not compile.
Developers should be able to edit those sources (the header files)
in their IDE.
* Header-only libraries may need to generate some of their header
files via custom commands.
Some projects work around these limitations by pairing each interface
library with an `add_custom_target` that makes the header files and
custom commands appear in the generated buildsystem and in IDEs.
Lift such limitations by allowing INTERFACE libraries to have SOURCES.
For those with sources, add a corresponding build target to the
generated buildsystem.
Fixes: #19145
Previously we disallowed use of arbitrary properties on INTERFACE
libraries. The goal was to future-proof projects using them by not
allowing properties to be set that may affect their future inclusion in
the generated buildsystem. In order to prepare to actually include
INTERFACE libraries in the generated buildsystem, drop the filter and
allow arbitrary properties to be set.
Issue: #19145
Since commit 2600e923a6 (Disallow INTERFACE libraries with
add_custom_command(TARGET)., 2014-03-19, v3.0.0-rc3~1^2), we do not
trace dependencies of INTERFACE libraries that do not participate in the
build system. Therefore we can remove a check of the target type from
the implementation that had been added by commit 9db9c1fc8b (cmTarget:
Don't try to get sources of an INTERFACE_LIBRARY., 2014-02-14,
v3.0.0-rc1~20^2).
Since commit 2026915f8f (Swift: Propagate Swift_MODULE_DIRECTORY as include
directory, 2020-02-03, v3.18.0-rc1~547^2) we internally call
`GetAllConfigCompileLanguages` on all directly linked targets without
checking if they are interface libraries that don't compile at all.
That violates an internal assumption and assertion.
Fixes: #20977
The ability to disable adding architectures completely for packaging purposes
and cases requiring passing the architectures flags explicitly has been
requested.
Support a false value for CUDA_ARCHITECTURES and CMAKE_CUDA_ARCHITECTURES
for this purpose.
Implements #20821.
Clang isn't very good at finding the installed CUDA toolkit.
The upstream recommendation is that we should pass the toolkit explicitly.
Additionally:
* Avoids Clang having to search for the toolkit on every invocation.
* Allows the user to use a toolkit from a non-standard location by simply
setting CUDAToolkit_ROOT. The same way as with FindCUDAToolkit.
Clang wants the directory containing the device library and version.txt as the
toolkit path.
We thus pass the newly introduced CUDAToolkit_LIBRARY_ROOT as the toolkit path.
We save CUDAToolkit_ROOT_DIR and CUDAToolkit_LIBRARY_ROOT on Clang to have them
available in try_compile() and avoid unnecessary re-searching or a possibly
different installation being found in FindCUDAToolkit.
This however means that the selected toolkit can't be changed after the initial
language enablement.
We now determine CUDA compiler ID before doing actual detection, as we don't
want to spend time finding the CUDA toolkit for NVIDIA.
Implements #20754.
Since commit 729d997f10 (Precompile Headers: Add REUSE_FROM signature,
2019-08-30, v3.16.0-rc1~101^2), `GetPchFileObject` handles the case that
it is called first for another target's `REUSE_FROM` by calling
`AddSource` to make sure `GetObjectName` can produce the correct object
name. However, `AddSource` causes `ClearSourcesCache` to be called,
which since commit a9f4f58f0c (cmGeneratorTarget: Clear AllConfigSources
in ClearSourcesCache, 2020-05-15, v3.16.7~2^2) now correctly erases the
`AllConfigSources` structure. This is okay during `AddPchDependencies`,
but there is another code path in which it is problematic.
When the Visual Studio generator's `WriteAllSources` method is looping
over the sources, the `cmake_pch.cxx` source is encountered first. This
causes `OutputSourceSpecificFlags` to call `GetPchCreateCompileOptions`,
which calls `GetPchFile`, which under MSVC with `CMAKE_LINK_PCH` calls
`GetPchFileObject`. That leads to `ClearSourcesCache` erasing the
structure over which `WriteAllSources` is iterating!
This bug is caught by our `RunCMake.PrecompileHeaders` test when run
with the VS generator as of the commit that exposed it by fixing
`ClearSourcesCache`. However, that change was backported to the CMake
3.16 series after testing only with later versions versions that contain
commit a55df20499 (Multi-Ninja: Add precompile headers support,
2020-01-10, v3.17.0-rc1~136^2). By adding proper multi-config support
for PCH, that commit taught `cmLocalGenerator::AddPchDependencies` to
call `GetPchFile` with the real set of configurations instead of just
the empty string. This allows the `GetPchFile` cache of PCH sources to
be populated up front so that the later calls to it in the
`WriteAllSources` loop as described above do not actually call
`GetPchFileObject` or `ClearSourcesCache`. That hid the problem.
Fix this by re-ordering calls to `AddPchDependencies` to handle
`REUSE_FROM` targets only after the targets whose PCH they re-use.
Remove the now-unnecessary call to `AddSource` from `GetPchFileObject`
so that `ClearSourcesCache` is never called during `WriteAllSources`.
Update the PchReuseFrom test case to cover an ordering of targets that
causes generators to encounter a `REUSE_FROM` target before the target
whose PCH it re-uses.
Fixes: #20770
