The EXPORT-OldIFace test case uses install(TARGETS) and so generates a
warning:
CMake Warning in CMakeLists.txt:
WARNING: Target "foo" has runtime paths which cannot be changed during
install. To change runtime paths, OS X version 10.6 or newer is required.
Therefore, runtime paths will not be changed when installing.
CMAKE_BUILD_WITH_INSTALL_RPATH may be used to work around this limitation.
Set CMAKE_BUILD_WITH_INSTALL_RPATH to avoid the warning since we do not
need to run the binaries from the build tree anyway.
Previously we did not clearly document that `--target` is only supported
to be specified once. Even worse, specifying it multiple times would
silently ignore any previously specified targets and only build the last
target.
Update the documentation to specify this. Update the implementation to
reject multiple `--target` options to prevent user errors.
Updates to Tests/Fortran by commit v3.2.0-rc1~501^2 (Avoid if() quoted
auto-dereference, 2014-10-14) changed our check
"${CMAKE_Fortran_COMPILER_ID}" MATCHES "${CMAKE_C_COMPILER_ID}"
to
CMAKE_Fortran_COMPILER_ID MATCHES CMAKE_C_COMPILER_ID
because CMP0054 warned about the LHS compiler id "MSVC" being expanded.
However, the RHS of if(MATCHES) does not auto-dereference so this check
has returned FALSE since then and the FortranCInterface part of the test
has not been running!
Fix this by using STREQUAL with quoted arguments and setting CMP0054 to
NEW (by requiring 3.1).
Refactoring merged by commit v3.5.0-rc1~299 (Merge topic
'use-generator-target', 2015-10-20) in and around
commit v3.5.0-rc1~299^2~13 (cmExportSet: Store a cmGeneratorTarget,
2015-10-17) changed export sets to delay looking up actual targets and
stores only their names. However, in InstallCommand::HandleExportMode
we need to lookup targets immediately to check them for
EXPORT_LINK_INTERFACE_LIBRARIES. The check was accidentally made local
to the current directory, so if an export set contains a target from
another directory the lookup fails and CMake crashes. Fix the check to
look up the target name globally, and tolerate when no target is found
just in case.
Reported-by: Kelly Thompson <kgt@lanl.gov>
Refactoring in commit v3.5.0-rc1~347^2~2 (Set the current dirs on the
snapshot before creating the cmMakefile) accidentally changed the
source and binary directories configured in `cmake -E cmake_depends`
for use during dependency scanning. This can cause the wrong directory
information to be loaded. It also breaks Fortran module dependency
scanning for modules provided by targets in subdirectories that do
not have Fortran_MODULE_DIRECTORY set.
Fix the dependency scanning directory configuration and add a test to
cover the Fortran module case in which the breakage was observed.
Reported-by: Kelly Thompson <kgt@lanl.gov>
While evaluating `if(MATCHES)` we get a `const char*` pointer to the
string to be matched. On code like
if(CMAKE_MATCH_COUNT MATCHES "Y")
the string to be matched may be owned by our own result variables.
We must move the value to our own buffer before clearing them.
Otherwise we risk reading freed storage.
The add_custom_command(TARGET) signature only works for targets defined
in the current directory. Clarify this in the error message when the
target exists but was defined elsewhere.
Inspired-by: Bartosz Kosiorek <gang65@poczta.onet.pl>
Imported targets are now the recommended way of dealing with external
library dependencies. Add one for FindPNG and update documentation
accordingly. Also add a test case activated by CMake_TEST_FindPNG.
CMake 3.4 may crash on this case. The problem seems to have been
fixed since then, but keep it working by adding a test case now.
Reported-by: Gonzalo Garramuño <ggarra13@gmail.com>
Since this command was introduced in 2002 it has incorrectly constructed
the child process command line by concatenating arguments separated by
spaces with no quoting. Fix this by passing the command argument vector
directly to RunSingleCommand without an intermediate quoting and
re-parsing step.
Reported-by: Andrey Pokrovskiy <wonder.mice@gmail.com>
The target_link_libraries command records the PRIVATE dependencies of a
STATIC library in INTERFACE_LINK_LIBRARIES as "$<LINK_ONLY:dep>". This
hides the target name from export namespacing logic inside a generator
expression. When user-written generator expressions reference a target
name they must put it inside a "$<TARGET_NAME:dep>" expression to allow
the export logic to rename the target. In the case that the private
dependency is not already a generator expression, target_link_libraries
must use "$<LINK_ONLY:$<TARGET_NAME:dep>>" to allow the export logic to
rename the target.
Reported-by: Tamás Kenéz <tamas.kenez@gmail.com>
Expand the -W set of cmake options to include support for the -Werror
and -Wno-error format, which is used to control upgrading and
downgrading warning and error messages. Implement support for these new
formats for the dev and deprecated message types.
Add tests and updated documentation for new options.
fae47798 Utilities/Release: Configure Windows binary to support Windows XP
083312a8 Utilities/Release: Switch to .msi builder for Windows binary
240b065f Utilities/Release: Optionally load environment on remote build server
a95b4715 Utilities/Release: Add optional remote launcher to ssh calls
d8bc26a0 Xcode: Parse variant and genex for CMAKE_XCODE_ATTRIBUTE (#14947)
dc0ddb9e Xcode: Store configuration name along with XcodeObject (#14947)
28f98cee Xcode: Make CMAKE_XCODE_ATTRIBUTE calculation last step (#14947)
28db2268 Xcode: Factor out XCODE_ATTRIBUTE_ variant filter (#14947)
Before this change backslashes in strings were escaped during compile
flags adds via AppendFlag(). But global flags like OTHER_CPLUSPLUSFLAGS
are not added as flags but as plain strings so they were not escaped
properly.
Now the escaping is performed within cmXCodeObject::PrintString() which
ensures that strings are always encoded.
The behaviour of double specified keywords is rather undefined or at
least not clearly documented. This change introduces a strict check and
emits a warning in case a keyword has been specified more than once.
Implement a native `cmake_parse_arguments` command that is fully
compatible with the documented behaviour of the previous implementation.
Leave the CMakeParseArguments module empty but existing for
compatibility.
This patch solves the problem of installing both: Device and Simulator
libraries on iOS. Before only one of them was installed.
If the IOS_INSTALL_COMBINED property is set on a target, a
special install hook will be activated which builds the corresponding
target and combines both at the install location.
The original patch was contributed by Ruslan Baratov, and polished by
Gregor Jasny.
Avoid constructing full paths to .git repositories in the test. Use
relative paths and let Git convert them to absolute paths internally.
This is simpler and also avoids trouble with various absolute path root
component conventions on Windows (`c:/`, `/c/`, `/cygdrive/c/`).
- Add TIFF::TIFF imported target
- Document imported target
- Add testcase to test the standard variables and the imported
target
Also:
- Add TIFF_INCLUDE_DIRS to match common practice
- Update documentation generally, including documenting
TIFF_INCLUDE_DIRS
If multiple input files are provided then the destination must be a
directory. If only one input file is provided then destination may be
either a file or directory.
7a327727 Embarcadero: Fix erroneous interpretation of __CODEGEARC_VERSION__.
25211d75 Compiler ID: Compiler versions must be a valid, numeric version string.
060442c2 Embarcadero: Check code using CMAKE_CXX_COMPILER_ID and CMAKE_C_COMPILER_ID.
f3b3219c Embarcadero/Watcom: Properly skip VSResource test for other generators.
ddbda722 Embarcadero: Fix bug where duplicate Ninja job pools would be created.
The CMAKE_CXX_COMPILER_ID and CMAKE_C_COMPILER_ID variables are set to
"Borland" for older versions of the compiler. Newer CodeGear/Embarcadero
compilers will have those variables set to "Embarcadero". Search for lines of
code referencing both the variable name and Borland to be sure that they also
refer to Embarcadero.
A typical iOS application bundle (also Framework Bundle) contains the
application executable and any resources used by the application (for
instance, the application icon, other images, and localized content) in
the top-level bundle directory. The same rule applies to Framework
Bundles.
Explicitly enable deprecated warnings by default, via the
cmake::GetSuppressDeprecatedWarnings method, which signals
suppression is turned off unless the CMake variables are set
as required.
Add tests and update the documentation for the new
functionality.
Change the '-Wdev' and '-Wno-dev' options to also enable and
suppress the deprecated warnings output, via the
'CMAKE_WARN_DEPRECATED' CMake variable, by default. This
action does not happen if the user specifies a deprecated
warning message option.
Add tests and update the documentation for the new
functionality.
Add 'deprecated' warning options type, to allow setting
CMAKE_WARN_DEPRECATED via the -W '-Wdeprecated' and
'-Wno-deprecated' options.
Add tests for new options and updated documentation.
Refactor the -Wdev and -Wno-dev options parser to use a generic -W
parser that follows the GCC pattern, excluding support for
-Werror=TYPE and -Wno-error=TYPE formats for now.
Explicitly enable author warnings by default, via the
cmake::GetSuppressDevWarnings method, which signals suppression
is turned off unless the CMake variables are set as required.
Add test cases for author and deprecated messages displayed by
default.
Make the message suppression more consistent, by adding a check
for the message related CMake variables in cmake::IssueMessage,
which allows callers of IssueMessage other than the message
command to behave as expected. Also added a check for
CMAKE_SUPPRESS_DEVELOPER_WARNINGS in the message command to
mirror the deprecated message type behaviour.
Added a 'force' flag to the cmake::IssueMessage method, to
make the message suppression consistent, when setting the
message related CMake variables directly in a CMake file.
Expand message command tests to cover the AUTHOR_WARNING message
type as well.
Change the Cobertura handler to look for an environment variable
called "COBERTURADIR" which contains the directory where the
coverage.xml file is found. If that variable doesn't exist,
continue to use the default of the binary directory.
Update the test to use an appropriate value in the environment
variables.
This fixes a bug where 64 bit builds with /bigobj incorrectly determined
that the object files were not 64 bit. This manifested itself with
printf type functions showing up as undefined because the leading
underscore was being removed and should not be removed.
If a custom command has a SYMBOLIC output (that is never actually
created) then do not mark the custom command build statement as
'restat'. Otherwise other custom commands that depend on the symbolic
output may not always re-run because after running the first custom
command Ninja 'restat' will detect that the output timestamp did not
change and skip its dependents.
This was observed with the ExternalProject BUILD_ALWAYS option where
Ninja would not re-run the 'install' step each time 'build' re-runs.
Refactoring in commit v3.4.0-rc1~74^2~1 (MSVC: Rewrite manifest file
handling with Makefile and Ninja, 2015-09-15) broke handling of this
option. Fix it and add a test case.
Commit c389f8bb (cmLocalGenerator: Port Find method away from
GetGeneratorTarget, 2015-10-25) ported the implementation of
FindGeneratorTargetToUse away from the FindTargetToUse method,
but neglected to handle alias targets.
The latter method has a parameter to determine whether to
include alias targets in the search, but as that is only
needed at configure time, this generate-time equivalent does
not need the condition.
Changes in commit v3.4.0-rc1~124^2~1 (cmFindProgramCommand: Re-implement
search using more flexible approach, 2015-09-01) did not preserve the
behavior of looking for the given name with no search path at all.
Fix this and add a test case covering finding an absolute path with
no search directories.
In some cases, CMake returned the following error:
-- Checking for module 'foo'
-- Package 'foo' not found
When the actual error returned by pkg-config was:
Package 'bar', required by 'foo', not found
Now, the actual error is forwarded to the user.
-- Checking for module 'foo'
-- Package 'bar', required by 'foo', not found
For the standard case (i.e. the package was indeed not found), the
CMake error was:
-- Checking for module 'foo'
-- Package 'foo' not found
But it now prints:
-- Checking for module 'foo'
-- No package 'foo' found
The associated test was also updated. ${last} refers to the last
CLI argument.
Refactoring in commit v3.4.0-rc1~390^2~1 (cmCTestMultiProcessHandler:
Refactor RUN_SERIAL implementation, 2015-06-01) forgot to update a code
path for cleaning up after a failed RUN_SERIAL test. This causes an
infinite loop after a RUN_SERIAL test fails. Fix it and add a test.
a4bbdc5e cmLocalGenerator: Remove cmGeneratorTargetsType from setter API.
04b6bb16 cmLocalGenerator: Simplify semantic of adding generator targets.
400e3d19 cmLocalGenerator: Don't store imported generator targets
726e461b CMP0063: Split unit test by target type.
Commit v3.4.0-rc1~494^2~4 (cmMakefile: Add API for elseif to create
backtrace., 2015-05-29) removed the use of cmMakefileCall to push/pop
execution context in favor of a new way to create backtraces.
However, a call to cmMakefile::GetExecutionContext is still invoked to
issue a contextual CMP0054 warning through cmConditionEvaluator. As
the elseif is not part of the call stack, this resulted in trying to
access an empty vector.
Avoid the attempt at getting execution context when evaluating elseif by
constructing a context and backtrace on behalf of the cmConditionEvaluator
in all cases.
The HTML file for the Delphi Code coverage was being found by the
Dashboard coverage run of CMake itself. Switch it to be a configured
file to eliminate this extra reading.
Add new API for the subdirs command to cmState.
This fixes a regression introduced in commit f716460e (cmMakefile: Move
invokation to initialize snapshot., 2015-10-06).
Revert commit v3.4.0-rc1~10^2~2 (Features: Disable support for Oracle
SolarisStudio on non-Linux, 2015-09-29) and two follow-up commits.
The support of compile features and language standards on Orcale
SolarisStudio needs more investigation so for CMake 3.4 we should
just act as 3.3 did.
2402bb8c Help: Document Windows 10 Universal Applications in cmake-toolchains(7)
1be2f12c VS: Add support for Windows 10 Universal (Store) Applications
2798dbda VS: Refactor indentation of LinkLibraryDependencies
8c426183 MSVC: Add system libs for WindowsStore on VS 2015
d1b87d72 VS: Select Windows 10 Store SDK and toolset for VS 2015
Teach the VS 2015 generator to support WindowsStore 10.0 applications.
Add target properties to customize them:
* VS_WINDOWS_TARGET_PLATFORM_MIN_VERSION: Specifies the minimum version
of the OS that the project can target.
* VS_DESKTOP_EXTENSIONS_VERSION, VS_MOBILE_EXTENSIONS_VERSIONS,
VS_IOT_EXTENSIONS_VERSION: Add a reference to the version of the SDK
specified to the target allowing to target the extended functionality in
a universal project.
* VS_IOT_STARTUP_TASK: Specifies that the target should be
built as an IOT continuous background task.
5fdf7594 Tests: Suppress WriteCompilerDetectionHeader failure on SunPro
c824b23d Features: Fix C++98 flags on Oracle SolarisStudio 12.4 on Linux
61bc0f73 Features: Disable support for Oracle SolarisStudio on non-Linux
On an Oracle 12.4 build the c_using_fortran executable cannot find the
"fsu" library at runtime. Since this is an implementation detail of the
"hello" library, link that library to it privately so that "-lfsu" does
not propagate to the executables consuming it.
Rather than using the CXXFLAGS environment variable in the make-only
build, copy the CMAKE_CXX_FLAGS used to build the files on the CMake
side. This will account for any changes made by CompileFlags.cmake
or cache-provided flags.
Use the run_cmake() function to generate the test build tree with
the proper CMake generator and also to verify that it succeeds.
Drop our PreTestError helper as it is no longer needed.
The DEPENDENCIES test case uses install(TARGETS) and so generates a warning:
CMake Warning in CMakeLists.txt:
WARNING: Target "test_prog" has runtime paths which cannot be changed
during install. To change runtime paths, OS X version 10.6 or newer is
required. Therefore, runtime paths will not be changed when installing.
CMAKE_BUILD_WITH_INSTALL_RPATH may be used to work around this limitation.
Set CMAKE_BUILD_WITH_INSTALL_RPATH to avoid the warning since we do not
need to run the binaries from the build tree anyway.
We do support SunPro 5.13 compiler features, but only on Linux.
Suppress the portion of the test that fails on Solaris until
the larger problem can be addressed.
9298f56d CPack: allow packaging of empty directories
ecd71c9c SystemTools: time operations on directories
19c6f4ee SystemTools: set time file permissions
749a140a CPackDeb: allow empty directories in component packages
The Ninja generator and Visual Studio generators are special-cased for the
QtAutogen feature. In order to reduce the number of custom targets, the Visual
Studio generators prefer to create custom commands instead, and in order to
create appropriate Ninja files, generated rcc files are listed as byproducts.
This requires the use of the GetConfigCommonSourceFiles API of the
cmGeneratorTarget for those generators when initializing the autogen target.
The initializer method is called from Compute() after the cmGeneratorTarget
objects are created, however the initialization of the object directory occurs
later in the InitGeneratorTargets method. That means that the resulting object
locations are computed incorrectly and cached before the object directory is
determined, so the generated buildsystem can not find the object files.
The initialization of the object directory was split from the creation of
cmGeneratorTarget instances in commit 0e0258c8 (cmGlobalGenerator: Split
creation of generator object from initialization., 2015-07-25). The motivation
for the split was to do only what is essential to do early in cases where
cmGeneratorTargets need to be created at configure-time. That is required for
the purpose of implementing policies CMP0024 and CMP0026, and for
try_compile(LINK_LIBRARIES). However, the split was not really necessary.
Compute the object directory in the cmGeneratorTarget constructor instead.
The QtAutogen unit test already tests the use of TARGET_OBJECTS with AUTOMOC,
and that test already passes on Ninja. The reason it already passes is that
the QtAutogen target also uses the AUTORCC feature, and specifies several qrc
files in its SOURCES. Later in the Compute algorithm (after the
InitGeneratorTargets call), the rcc files are determined and target->AddSource
is called. The AddSource call clears the previously mentioned cache of source
files, causing it to be regenerated when next queried, this time taking account
of the object directory.
Extend the test suite with a new target which does not make use of AUTORCC with
qrc files so that the test added alone would break without the fix in this
commit.
Some commands on Windows do not understand forward slash paths and
require backslashes. In order to help projects generate shell
invocations of such commands, provide a generator expression to convert
paths to the shell-preferred path format for the current generator.
This will allow custom commands to generate paths the same way CMake
does for compiler command invocations.
The latter is now the preferred URL for visiting cmake.org with a
browser. Convert using the shell code:
git ls-files -z | xargs -0 sed -i 's|http://www\.cmake|https://cmake|g'
In the Ninja generator we run all build rules from the top of the build
tree rather than changing into each subdirectory. Therefore we convert
all paths relative to the HOME_OUTPUT directory. However, the Convert
method on cmLocalGenerator restricts relative path conversions to avoid
leaving the build tree with a "../" sequence. Therefore conversions
performed for "subdirectories" that are outside the top of the build
tree always use full paths while conversions performed for
subdirectories that are inside the top of the build tree may use
relative paths to refer to the same files.
Since Ninja always runs rules from the top of the build tree we should
convert them using only the top-level cmLocalGenerator in order to
remain consistent. Also extend the test suite with a case that fails
without this fix.
The changes in commit 47b060ae (CPackDeb: allow empty directories in
component packages, 2015-09-21), commit b58de9fe (CPack: allow packaging
of empty directories, 2015-09-21), and commit b761e90d (CPack: remove
accidental changes, 2015-09-22) regressed packaging of CMake itself.
Revert the changes until they can be revised and rebased on other
changes that make additional fixes.
Brad King
Orion Poplawski
Gregor Jasny
Sebastian Schuberth
Ashley Whetter
Sam Thursfield
Ben Boeckel
Roger Leigh
Marc Chevrier
Yves Frederix
Michael Scott
James Johnston
Matthias Maennich
Ruslan Baratov
Bartosz Kosiorek
Domen Vrankar
Bartosz Kosiorek
Roger Leigh
Joseph Snyder
Bill Hoffman
Stephen Kelly
Raffi Enficiaud
Benjamin Chrétien
Tamas Kenez
Gilles Khouzam
Stefan Kislinskiy