For the OLD CMP0022 behavior, we need to treat the implementation
as the interface when computing the interface libraries. Make it
possible to do that without computing the link languages by adding
a new GetLinkImplementationLibraries method. Extend the existing
GetLinkImplementation method to populate the languages if the
libraries have already been computed and cached.
Change GetTransitivePropertyTargets to invoke GetLinkInterfaceLibraries
instead of GetLinkInterface. This is key, as it is a method called
by cmGeneratorExpressionEvaluator.
Change the cmGeneratorExpressionEvaluator to invoke
GetLinkImplementationLibraries instead of GetLinkImplementation.
When evaluating the SOURCES property, we will need to be able to access
the link libraries without accessing the link languages, as the languages
depend on the SOURCES.
Arrange the logic so that the part which deals with libraries only is
at the top. In a follow-up commit, this will be split into two methods.
Ensure that the explanatory CMP0022 comment is only present in one
location.
The callers already skip non-targets, so unify the target search.
Change supporting functions to accept a container of targets instead
of strings where possible.
In a follow-up, the list of sources will become dependent on
the config, so check for existence in cmTarget::GetSourceFiles
instead of up-front with cmGlobalGenerator::CheckTargets().
6e89c8a5 install: Support generator expressions in FILES and PROGRAMS mode
f11f7b34 cmInstallFilesGenerator: Add reference to calling cmMakefile
e190236c Help: Format install() command documentation
If there is no ARGV1, that is fine; version will be made empty, and no
version will be passed to find_package().
This is relevant when find_dependency is invoked multiple times,
sometimes with a version specified and sometimes without.
find_dependency(dep1 3.4)
find_dependency(dep2) # version still set to 3.4.
Teach the install(FILES) and install(PROGRAMS) commands to evaluate
generator expressions in the list of files.
Extend the ExportImport test to cover installation cases involving
generator expressions.
Add a Makefile member to the cmInstallFilesGenerator class and
populate it on construction. This will be useful in a following
change to evaluate generator expressions with proper context.
The char-by-char parsing causes lots of reallocations which shouldn't be
necessary. To improve this, fast-path strings without a semicolon,
reserve space in the result, and insert into the result in chunks.
Optimize cmGeneratorExpressionLexer::Tokenize to use a switch statement.
The many dereferences of the input pointer were expensive. Also remove
excess pointer arithmetic.
There are some corner cases in variable expansion which would be nice to
capture before going and rewriting the variable expansion code. The
majority of these are related to configuring files and strings with '@'
in them in conjunction with @ONLY being specified. Another is testing
for '(' usage inside of ENV variable references based on whether it is
quoted or not.
Teach ExternalProject_Add a new BUILD_ALWAYS option to skip using
the build step stamp file and execute the step on every build.
Extend the BuildDepends test with a case to cover this option.
Historically CMake used three version components for the feature level.
We released new features while incrementing only the third version
component. Since commit v2.8.2~105^2~4 (New version scheme to support
branchy workflow, 2010-04-23) we used the fourth version component for
bug-fix releases and the development date:
<major>.<minor>.<patch>[.<tweak>][-rc<n>] = Release
<major>.<minor>.<patch>.<date>[-<id>] = Development
This solidified use of three components for the feature level, and was
necessary to continue releasing 2.x versions because:
* Some existing projects performed floating-point comparisons of
${CMAKE_MAJOR_VERSION}.${CMAKE_MINOR_VERSION} to 2.x numbers
so ``x`` could never be higher than 9.
* Version 2.9.<date> was used briefly in post-2.8.0 development in
CVS prior to the transition to Git, so using it in releases may
have caused confusion.
Now that we are moving to 3.x versions, these two restrictions go away.
Therefore we now change to use only two components for the feature
level and use the scheme:
<major>.<minor>.<patch>[-rc<n>] = Release
<major>.<minor>.<date>[-<id>] = Development
People will be tempted to put things there for convenience, thereby
causing conflicts similar to
http://thread.gmane.org/gmane.comp.compilers.clang.devel/35162/focus=35169
where it is conceivable that the LLVM developers could put a flag on
a target for convenience, which would cause conflicts for some downstreams.
When building boost with an alternate namespace the libraries generated
will have a different naming convention. This is often done to ensure
no symbol conflicts with external libraries built against a different
version of boost. If the namespace used is "myprivateboost::" instead
of "boost::" then the libraries built will be named myprivateboost_foo
instead of boost_foo. Add an option to specify a custom namespace used
to alter the library names that get searched for.
Stephen Kelly
Brad King
Julien Schueller
Kitware Robot
Alex Merry
Ben Boeckel
Ben Boeckel
Ho Cheung
Dinar Valeev
Chuck Atkins