Teach the Makefile and Ninja generators to substitute for an <INCLUDES>
placeholder instead of putting -I in <FLAGS>. Update our values for
CMAKE_<LANG>_COMPILE_OBJECT,
CMAKE_<LANG>_CREATE_ASSEMBLY_SOURCE, and
CMAKE_<LANG>_CREATE_PREPROCESSED_SOURCE
to place <INCLUDES> just before <FLAGS>.
Restore call to AppendFortranFormatFlags accidentally dropped by
commit 0837538e (cmCommonTargetGenerator: Adopt GetFlags method,
2015-07-09). It was added originally by commit 6a56740e
(cmNinjaTargetGenerator: Add Fortran flag generation, 2015-07-09).
Simplify the per-source ComputeDefines implementation by getting
target-wide defines from GetDefines. Technically this changes behavior
by no-longer de-duplicating/sorting defines from both the target and the
source, but this makes it consistent with the Makefile generator. It
may also later help move target-wide defines into per-target compilation
rules.
Create a <LANG>_COMPILER_LAUNCHER target property (initialized by a
CMAKE_<LANG>_COMPILER_LAUNCHER variable) to specify a compiler launcher
tool. This will supersede the CMAKE_<LANG>_COMPILER_ARG1 approach to
using such tools. The old approach set CMAKE_<LANG>_COMPILER to the
launcher tool while the new approach leaves this variable set to the
actual compiler.
Implement this property for Makefile and Ninja generators. It cannot be
implemented for VS or Xcode generators as the IDE build tools offer no
such hooks.
Create a <LANG>_INCLUDE_WHAT_YOU_USE target property (initialized by a
CMAKE_<LANG>_INCLUDE_WHAT_YOU_USE variable) to specify an IWYU command
line to be run along with the compiler.
Our <LANG>_COMPILER and <LANG>_<TARGET_TYPE>_LINKER rule generation has
access to a specific cmTarget so the results may depend on it. Instead
generate separate rules for each target using an encoded target name.
In particular, this makes CTEST_USE_LAUNCHERS report proper target
information.
Check for CMAKE_COMPILER_IS_MINGW only after enabling a language when it
might actually be set. Previously this worked by accident because the
check for working compiler or a second language enabled would cause the
code path to be taken.
Store UsingMinGW as an instance member of cmGlobalNinjaGenerator so that
it is reset on each reconfigure. Otherwise cmake-gui cannot switch
between build trees for MinGW or non-MinGW tools.
The "/showIncludes" flag is only available with MS C and C++ compilers,
and on compilers that "simulate" them (like Intel for Windows). Fix our
logic to choose this type only for MS tools with these languages. All
other cases need to use "deps = gcc" and define DEP_FILE in the build
rule.
Custom command path normalization added in commit v3.1.0-rc1~471^2
(add_custom_command: Normalize OUTPUT and DEPENDS paths, 2014-05-28)
broke use of OBJECT_DEPENDS to bring in custom commands because the
latter paths were not normalized too. Normalize them and add a test
case.
Reported-by: Daniel v. Gerpen
A common idiom in CMake-based build systems is to have custom commands
that generate files not listed explicitly as outputs so that these
files do not have to be newer than the inputs. The file modification
times of such "byproducts" are updated only when their content changes.
Then other build rules can depend on the byproducts explicitly so that
their dependents rebuild when the content of the original byproducts
really does change.
This "undeclared byproduct" approach is necessary for Makefile, VS, and
Xcode build tools because if a byproduct were listed as an output of a
rule then the rule would always rerun when the input is newer than the
byproduct but the byproduct may never be updated.
Ninja solves this problem by offering a 'restat' feature to check
whether an output was really modified after running a rule and tracking
the fact that it is up to date separately from its timestamp. However,
Ninja also stats all dependencies up front and will only restat files
that are listed as outputs of rules with the 'restat' option enabled.
Therefore an undeclared byproduct that does not exist at the start of
the build will be considered missing and the build will fail even if
other dependencies would cause the byproduct to be available before its
dependents build.
CMake works around this limitation by adding 'phony' build rules for
custom command dependencies in the build tree that do not have any
explicit specification of what produces them. This is not optimal
because it prevents Ninja from reporting an error when an input to a
rule really is missing. A better approach is to allow projects to
explicitly specify the byproducts of their custom commands so that no
phony rules are needed for them. In order to work with the non-Ninja
generators, the byproducts must be known separately from the outputs.
Add a new "BYPRODUCTS" option to the add_custom_command and
add_custom_target commands to specify byproducts explicitly. Teach the
Ninja generator to specify byproducts as outputs of the custom commands.
In the case of POST_BUILD, PRE_LINK, and PRE_BUILD events on targets
that link, the byproducts must be specified as outputs of the link rule
that runs the commands. Activate 'restat' for such rules so that Ninja
knows it needs to check the byproducts, but not for link rules that have
no byproducts.
Ninja generator ensures that all custom commands being target
dependencies are run before other source compilations. However in case
there are no such dependencies it currently generates empty phony rules
which clutter the build graph.
Teach the Ninja generator to produce such rules only when necessary.
Some compilers do not offer an option to specify the path to the object
file, but rather only to the directory in which to place the object
file. See issue 14876 for some examples. Add a new OBJECT_FILE_DIR
placeholder to specify the directory containing the object file for the
current compilation. This may differ from the main target OBJECT_DIR
when the object corresponds to a source in a subdirectory.
Since commit v3.0.0-rc1~305^2 (Ninja: deptype msvc for Intel's compiler
on Windows, 2013-11-25) we used "deps = msvc" for the Intel compiler.
This is correct only on Windows. On Linux we still want "deps = gcc".
Fix the logic to use "deps = msvc" when the compiler id or the "simulate
id" is "MSVC". This will preserve the behavior on Intel for Windows and
fix the behavior on Intel for Linux. In the future this should be
converted to a platform information module variable.
This reduces ninja file output even more for projects with lots of
libraries with entangled transitive dependencies. ParaView goes from the
previous 58M to about 45M.
Disallow the use of config-specific source files with
the Visual Studio and Xcode generators. They don't have
any way to represent the condition currently.
Use the same common-config API in cmQtAutoGenerators. While
it accepts config-specific files, it doesn't have to support
multiple configurations yet.
Loop over the configs in cmTargetTraceDependencies
and cmGlobalGenerator::WriteSummary and consume all source
files.
Loop over the configs in cmComputeTargetDepends and compute the
object library dependencies for each config.
Remove use of UseObjectLibraries from Makefile and Ninja generators. It
is not needed now because those generators use GetExternalObjects
which already contains the objects from object libraries.
The VS10 generator calls both the UseObjectLibraries and the GetExternalObjects
methods. Ensure that duplicates are not created by skipping objects
from object libraries in handling of GetExternalObjects.
Similarly, fix VS6, VS7 and Xcode object handling by skipping
external objects from OBJECT_LIBRARY usage as appropriate.
The error message in the BadSourceExpression1 test is now reported
by the generator expression evaluator, so it has different text.
Not everything that isn't MSVC is GCC. I have support for LDC's depfile
format on its way upstream[1], but its future is uncertain. CMake should
at least support this for future depfile formats.
[1]https://github.com/martine/ninja/pull/721
Brad King
Stephen Kelly
Bill Hoffman
Thomas Herz
Nils Gladitz
Adam Strzelecki
Ben Boeckel
Ben Boeckel