Since commit 328c0f65 (Simplify cmVisualStudio10TargetGenerator source
classification, 2012-03-19) the VS 10 generator uses the
cmGeneratorTarget source classification instead of directly getting the
list of source files from the target. This accidentally dropped the
CMakeLists.txt files from generated projects because they are added too
late for cmGeneratorTarget.
All generator-specific source files must be added to targets prior to
cmGeneratorTarget construction. Refactor addition of the CMakeLists.txt
files with CMake re-run custom commands to take place before normal
generation begins, and therefore early enough to be included in the
cmGeneratorTarget classification.
Commit 328c0f65 (Simplify cmVisualStudio10TargetGenerator source
classification, 2012-03-19) introduced the first use of source
classification from cmGeneratorTarget (which originated as Makefile
generator logic) in a Visual Studio generator for handling of header
files. Fix classification of header files to match known header
extensions instead of only the HEADER_FILE_ONLY property. Make it
consistent with the "Header Files" source group.
Refactoring by commit 11d9b211 (Add cmGeneratorTarget to represent a
target during generation, 2012-03-07) and commit 45c2f932 (Simplify
cmMakefileTargetGenerator using cmGeneratorTarget, 2012-03-07) preserved
behavior introduced by commit 7740ccd1 (some cleanup of the makefile
generator, 2006-02-14) that favored the IgnoreFile extension test over
the availability of a known compilation language associated with a
source file. If a source is not marked as HEADER_FILE_ONLY and has a
known language extension or an explicit LANGUAGE property it should be
treated as that language. The LANGUAGE source file property
documentation says so.
9b32475 automoc: add define to test which caused bug #130182066511 automoc: fix#13018, proper cmake escaping to avoid false rebuilds
c652812 make cmLocalGenerator::EscapeForCMake() static
This does not really test that the bug is fixed, but at least it makes
it easy to check manually whether the bug is there or not.
I have to see whether I can build a test which does test that
a target is not rebuilt everytime.
Alex
The variables stored in the AutomocInfo.cmake file were not properly
escaped, so when reading them back they could turn into lists, if they
contained double quotes initially.
This patch fixes this by using cmLocalGenerator::EscapeForCMake() to
escape the variables properly.
Alex
Since commit 4693cf84 (Xcode: Detect new default locations of Xcode 4.3
bits and pieces) Darwin.cmake detects the developer application
directory instead of hard-coding /Developer. Replace the hard-coded
path in CMAKE_SYSTEM_APPBUNDLE_PATH using the computed result.
Patch by Amine Khaldi!
Also, start using the -MT flag to set a target name for depfiles.
This works around a bug observed in distcc, as explained in the
comment. Based on a patch by Alexander Usov.
Use "file:///..." instead of "file:////..." when the file system path
starts in a slash. Commit 0916cc88 (CTest.UpdateHG: Fix repo URL for
local filesystem, 2012-03-02) added a third slash after "file://"
unconditionally. This worked for many file systems but not on Cygwin
where "file:////cygdrive/..." looks like "file://" followed by a network
file path "//cygdrive/...". Add the slash only if the file system path
does not already start with one.
This patch uses get_filename_component(REALPATH) so symlinks in the path
to Eclipse are resolved, which makes the version detection work in such cases.
Alex
Our only expectation of version number variables should be that
they begin with a decimal digit for VERSION_LESS, VERSION_EQUAL
and VERSION_GREATER comparison purposes. If people put extra
blah blah after a version number like "1.2.3 (this is some super
special extra information about our funky proprietary build of
the official 1.2.3 release)" then we should be ok with that.
So: now we have the following expectations for version number
variable content for the purposes of the AllFindModules test:
- it should start with a decimal digit (match "^[0-9]")
- it should not be empty
- it should not be VERSION_EQUAL 0
- it should not be NOT VERSION_GREATER 0
Clang warns
test/test_INT_format.h:28:26: warning: identifier after literal will
be treated as a reserved user-defined literal suffix in C++11
[-Wc++11-compat-reserved-user-defined-literal]
" expression [%"KWIML_INT_PRI##PRI"],"
because the KWIML_... part is an identifier lexically immediately
following a string literal. Add a space between the string literal
and the identifier to avoid the C++11 user-defined literal syntax.
Use the CMAKE_<LANG>_IMPLICIT_LINK_DIRECTORIES to look for libraries
that are in the -showme output from mpi<lang> programs. This is because
some libraries reported by -showme are found there, and FindMPI will
fail if they are not found.
The Watcom compiler interprets "-DB" as option
-db generate browsing information
so define "A_DEF" and "B_DEF" instead of just "A" and "B".
Skip CMAKE_SHARED_LIBRARY_C_FLAGS for Watcom because it is set to
-bd build Dynamic link library
which adds a DLL entry point to each object.
Commit 9a6ff950 (Fix for bug where VS2010 did not use .obj files,
2011-04-01) assumed that if an external object is GENERATED that it is
the output of a custom command in the current target. If it is
generated by another target then VS will not automatically include the
external object in the current target. This bug was preserved by the
refactoring in the parent commit.
Instead use <None> for external objects generated by a custom command in
the current target and <Object> for all other external objects. Update
the ExternalOBJ test to cover this case.
Alex Neundorf
KWSys Robot
Rolf Eike Beer
Ben Boeckel
Sean McBride
Brad King
David Cole
Tom Hughes
Eric NOULARD
Kashif Rasul
Peter Collingbourne
Bill Hoffman