07fc7b75 Tests: Test using objects from a language enabled in a subdirectory (#15325)
fdbfcfdf Ninja: Generate rules only for languages compiled in a target (#15325)
Refactoring in commit v3.1.0-rc1~688^2~2 (cmTarget: Compute languages
from object libraries on demand, 2014-03-18) taught cmTarget::GetLanguages
to (correctly) include the languages of object library sources. Previously
this was done only in cmTarget::ComputeLinkImplementationLanguages to
choose the linker language.
The Ninja generator writes out generic build rules for each language
compiled within a target using the rule variables defined in the
directory of the target. This only needs to be done for languages
actually compiled within the current target. Switch from using the
cmTarget::GetLanguages method to get the list of languages over to
using cmTarget::GetSourceFiles directly so we do not get the languages
in object libraries.
Strictly speaking this should make no difference because it is not safe
to use objects from a language not enabled in the directory containing
a target or else the link information for the language may not be
considered. However, in cases when no link information happens to be
needed for a language it was possible in CMake 3.0 and below to enable
a language only in a subdirectory providing an object library, and then
use the objects from a containing directory. The above change teaches
the Ninja generator to continue working in this case.
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.
Use an explicit "$TARGET_FILE" variable instead of "$out" so that
we can have multiple output files while still only referencing the
main one in command lines.
A library or executable target that consists only of a reference to an
object library may have no sources to compile or custom commands to run.
The command in the target is the link (or archive) command. Add missing
order-only dependencies to link commands so that target ordering
dependencies are satisfied for it without depending on an intermediate
compilation rule.
Extend the ObjectLibrary test to cover this case.
Co-Author: Brad King <brad.king@kitware.com>
97f2b7f5 Ninja: set correct LANGUAGE_COMPILE_FLAGS when linking
49fcffc6 Ninja: cmake formatting, make code more readable
b735c8cb MinGW: link like on Unix and use compile flags when linking
9407174b target_sources: New command to add sources to target.
81ad69e0 Make the SOURCES target property writable.
6e636f2e cmTarget: Make the SOURCES origin tracable.
3676fb49 cmTarget: Allow transitive evaluation of SOURCES property.
e6971df6 cmTarget: Make the source files depend on the config.
df753df9 cmGeneratorTarget: Don't add computed sources to the target.
869328aa cmComputeTargetDepends: Use valid config to compute target depends.
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.
Until now the cmCustomCommandGenerator was used only to compute the
command lines of a custom command. Generalize it to get the comment,
working directory, dependencies, and outputs of custom commands. Update
use in all generators to support this.
Use the clang RemoveCStrCalls tool to automatically migrate the
code. This was only run on linux, so does not have any positive or
negative effect on other platforms.
Casts from std::string -> cmStdString were high on the list of things
taking up time. Avoid such implicit casts across function calls by just
using std::string everywhere.
The comment that the symbol name is too long is no longer relevant since
modern debuggers alias the templates anyways and the size is a
non-issue since the underlying methods are generated since it's
inherited.
Just enough to reach the BuildMacContentDirectory method and the
NeedRelinkBeforeInstall methods.
In the future, those methods can be moved to cmGeneratorTarget.
Fix a regression created by commit 373faae5 (Refactor how bundles and
frameworks are supported, 2013-05-05).
Since the ninja file isn't aware of how framework symlinks work, we
suppress symlink creation and let cmOSXBundleGenerator handle it. Also,
use the real name of framework library in build rules as was done
before, instead of the symlink.
To properly track the usage of dependencies that are generated at
compile time as the side effect of other build steps we need
to make the WriteBuild and WritePhonyBuild commands non static
As CMAKE_ROOT_FIND_PATH can be a list, a new CMAKE_SYSROOT is
introduced, which is never a list.
The contents of this variable is passed to supporting compilers
as --sysroot. It is also accounted for when processing implicit
link directories reported by the compiler, and when generating
RPATH information.
This fixes bug #13797.
The kinds of changes applied in 373faae5 for frameworks are now
applied to CFBundle. The prefix and suffix for CFBundles are
now handled in cmTarget::GetFullNameInternal.
Make handling of directory separators consistent between
non-bundle and bundle code.
Remove xcode specific flag from cmTarget when getting install_name.
Add (more) consistent convenience functions in cmTarget to get
directories inside of bundles and frameworks to add files to.
This refactor also fixes bug #12263 where frameworks
had the wrong install name when SKIP_BUILD_RPATH.
Also make install_name for frameworks consistent between Makefile
and Xcode generator.
link.exe has problems with very very long lines in rsp files too.
Use $in_newline instead of $in variable for rspcontent which
separates the arguments with a newline instead of a simple space
and was specially made for this purpose.
In the response file also linker options could be passed,
and because <OBJECTS> is replaced by a response file, it
is necessary that no compiler option follows <OBJECTS>.
Stephen Kelly
Brad King
Nils Gladitz
Ben Boeckel
Peter Kümmel
Jiri Malak
Adrien Destugues
Clinton Stimpson
Robert Maynard
Andreas Mohr
Patrick Gansterer