| Commit message (Collapse) | Author | Age | Files | Lines |
|\
| |
| |
| | |
Merge-request: !3591
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
In commit fec441ec17 (Teach CROSSCOMPILING_EMULATOR to support
arguments, 2019-05-30, v3.15.0-rc1~6^2) the new member
`cmCustomCommandGenerator::EmulatorsWithArguments` was not initialized
to the proper size. Fix this and add a test case covering the crash
that could occur with multiple commands where an emulator appears only
in a later command.
Fixes: #19500
Co-Author: Brad King <brad.king@kitware.com>
|
|\ \
| | |
| | |
| | | |
Merge-request: !3563
|
| |/
| |
| |
| | |
Fixes: #19489
|
|\ \
| | |
| | |
| | | |
Merge-request: !3569
|
| |/
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The `AssemblerListingLocation` setting in VS project files is meant for
intermediate files created during the build much like object files.
When the VS 7 generator was first under development, commit 49aebe6c99
(new arch, 2002-09-04) placed both object files and the ASM list
location in the same directory. Later commit f9aef0e422 (Generator now
creates a separate intermediate files directory for each target,
2005-07-27) moved the object files to a per-target directory but the
ASM list location was not moved with them. Move it now.
Fixes: #19480
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
When building Swift executables and libraries which import a module, an
implicit link will be added by the driver. Because this links by name
rather than path, the library search path needs to be provided to
indicate where to find the library. For all local dependencies, add the
library paths for the targets when linking. This ensures that you can
link against local libraries without explicitly setting a library path.
Fixes: #19304
|
|/
|
|
|
|
|
|
|
|
| |
This changes the behaviour of the generators to use a per-language
library search path flag. This is needed for multi-language projects
with different compilers (e.g. cl + gfortran). Since the adjusted
variable has been part of the user settings, we control this based on a
policy.
Fixes: #19307
|
| |
|
|\
| |
| |
| | |
Merge-request: !3557
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The module mode fallback added by commit 22e65d10c1 (find_package: Fixed
CMAKE_FIND_PACKAGE_PREFER_CONFIG Module fallback, 2019-06-13,
v3.15.0-rc2~6^2) should not be used unless the `find_package` call
allows module mode. Doing so can lead to infinite recursion if a find
module tries to call config mode with `find_package(... NO_MODULE)`.
Fix the logic and add a test case.
Fixes: #19478
|
|\ \
| | |
| | |
| | | |
Merge-request: !3537
|
| |/
| |
| |
| |
| |
| |
| |
| | |
In commit 3a53005f7d (Build object library targets in VS, 2012-03-12,
v2.8.8~29^2~13) we updated the condition for unicode but accidentally
left out the SBCS case support for object libraries.
Fixes: #19469
|
|\ \
| |/
|/|
| | |
Merge-request: !3504
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Define `CMAKE_Fortran_SUBMODULE_SEP` with an empty string to mean that
the compiler uses no module prefix on its submodule files.
Also add a default fallback to use the `.mod` extension when
`CMAKE_Fortran_SUBMODULE_EXT` is not set. This is a better guess than
no extension at all.
|
| | |
|
|\ \
| | |
| | |
| | | |
Merge-request: !3526
|
| |/
| |
| |
| |
| |
| |
| | |
This ability was lost in commit 81b4d10d8f (CUDA: More exhaustive checks
to determine when to do device linking, 2019-05-09, v3.15.0-rc1~82^2)
and needs to be restored to use imported static libraries that have CUDA
symbols.
|
|\ \
| | |
| | |
| | | |
Merge-request: !3529
|
| |/
| |
| |
| |
| |
| |
| |
| |
| | |
In `ccmake` a single global generator instance may be used for multiple
configure step runs. The `cmGlobalGenerator::ClearGeneratorMembers`
method is supposed to clear global state that is specific to each
configure run but forgot to clear alias targets.
Fixes: #19457
|
|\ \
| | |
| | |
| | | |
Merge-request: !3521
|
| |/
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The message() command requires uppercase log levels. Even
though the cmake --loglevel option is not case sensitive, show
the supported values as uppercase to match the message()
docs as closely as possible, since they are related to the same
feature.
Also fixes the wrong string being shown for the warning level
by cmake --help.
|
|\ \
| | |
| | |
| | | |
Merge-request: !3500
|
| |/
| |
| |
| |
| |
| |
| |
| |
| |
| | |
`QPalette::Foreground` is deprecated and was replaced by
`QPalette::WindowText` since at least Qt 4.8.
https://doc.qt.io/archives/qt-4.8/qpalette.html#ColorRole-enum
This replaces the `QPalette::Foreground` color role with `QPalette::WindowText`
in `QtDialog/RegexExplorer.cxx` to avoid compiler warnings.
|
|\ \
| | |
| | |
| | | |
Merge-request: !3491
|
| |/
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Checks added in commit 81b4d10d8f (CUDA: More exhaustive checks to
determine when to do device linking, 2019-05-09, v3.15.0-rc1~82^2)
assumed that CUDA properties would be set only if CUDA is enabled.
We cannot do a device link step if we do not have the CUDA language
enabled. This was discovered as some projects unconditionally set CUDA
properties such as `CUDA_RESOLVE_DEVICE_SYMBOLS` even when the CUDA
language has not been enabled.
Fixes: #19432
|
|/
|
|
| |
Fixes: #19436
|
| |
|
|\
| |
| |
| | |
Merge-request: !3485
|
| |
| |
| |
| |
| |
| |
| |
| | |
Apply the refactoring from commit 707283981f (VS: Use AddLanguageFlags
to de-duplicate CMAKE_<LANG>_FLAGS* lookup, 2019-04-01,
v3.15.0-rc1~300^2) to the CUDA, ASM_MASM, and ASM_NASM languages too.
In particular, CUDA needs the MSVC runtime library selection logic
in the `AddLanguageFlags` method that is used by other generators.
|
|\ \
| | |
| | |
| | | |
Merge-request: !3484
|
| |/
| |
| |
| |
| |
| |
| |
| |
| | |
Refactoring in commit f5acecaa6f (cmExportCommand: Port to
cmArgumentParser, 2019-03-23, v3.15.0-rc1~270^2~3) broke the `export`
command's support for specifying `TARGETS` with no entries. Fix it and
add a test case.
Fixes: #19415
|
|/
|
|
|
|
| |
Fix the logic added by commit d91b5a72cd (Ninja: Add support for CUDA
nvcc response files, 2019-05-30, v3.15.0-rc1~8^2) to always use the CUDA
compiler response file flag for response files during device linking.
|
| |
|
|\
| |
| |
| | |
Merge-request: !3446
|
| |
| |
| |
| | |
Fixes: #19361
|
|\ \
| | |
| | |
| | | |
Merge-request: !3442
|
| | | |
|
|\ \ \
| | | |
| | | | |
Merge-request: !3441
|
| | |/
| |/|
| | |
| | | |
Qt is now copyright "The Qt Company Ltd.".
|
|\ \ \
| | |/
| |/|
| | | |
Merge-request: !3433
|
| | |
| | |
| | |
| | | |
Fixes: #19339
|
| | |
| | |
| | |
| | |
| | |
| | | |
The `cmake -E capabilities` command-line tool is implemented only in a
full-featured (non-bootstrap) CMake, and server mode is now always
available in this case.
|
| |/ |
|
|/
|
|
|
|
|
|
|
|
|
|
|
|
| |
When Autogen searches the header for a source files in a target,
the expanded real path is used and not the (possibly symbolic) original path
of the source file. If the source file path contains symbolic links, then the
correct `cmSourceFile` instance of the header won't be found, but a new one
will be generated. This way all header source file properties get lost, which
is problematic especially for the `SKIP_AUTOMOC/UIC/GEN` properties.
This patch changes the header detection in Autogen to use the original source
file path instead of the expanded real source file path.
Fixes: #19346
|
| |
|
|\
| |
| |
| |
| |
| |
| | |
fec441ec17 Teach CROSSCOMPILING_EMULATOR to support arguments
Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !3402
|
| |
| |
| |
| | |
Fixes: #19321
|
|\ \
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
07a80c7002 cmake: Teach -E remove_directory to remove multiple directories
013bee698e Tests: Add RunCMake.CommandLine make_directory test checks
3eb16de327 Tests: Fix RunCMake.CommandLine make_directory test name
Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !3411
|
| | |
| | |
| | |
| | |
| | | |
The `make_directory` command can make multiple directories in a single
invocation. Make `remove_directory` mirror that behavior.
|