Commit 4be67837 (read less from version headers into variables,
2012-08-19) switched from file(READ) and string(REGEX MATCHALL) to just
file(STRINGS) to extract the list of resource <file> entries. However,
the latter extracts entire lines that match the regex, not just the part
that matches the regex, so the subsequent string(REGEX REPLACE) fails to
match and replace anything. Return to the original parsing logic but
replace file(READ) with file(STRINGS) to load a minimal part of the file
before using string(REGEX MATCHALL) as before.
Added a new CUDA variable for specifying the CUDA_HOST_COMPILER. This will allow users to
be able to specify which host compiler to use for invoking NVCC with. By default it will
use the compiler used for host compilation. This is convenient for when you want to
specify a different compiler than the default compiler. You end up using the same
compiler for both the NVCC compilation and the host compilation instead of using the
default compiler in the path for NVCC.
a3815e6 -fix line length
9110d0e Eclipse on OSX: improve handling of framework include dirs (#13367)
d97b385 Eclipse on OSX: fix handling of framework include dirs (#13464)
Instead of reading the whole file using file(READ) and later matching on the
whole file use file(STRINGS ... REGEX) to get only those lines we are
interested in at all. This will make the list much smaller (good for debugging)
and also the regular expressions will need to match on much smaller strings.
Also unset the content variables once they are not used anymore.
The Spanish language MFC localization dll changed names from
VS 9 to 10. Use the correct file name ending with "esn.dll"
instead of the now non-existent one ending with "esp.dll"
Also, add the existing, but missing from our rules until now,
Russian language module.
Alphabetize the list while we're at it for easier reading in
the future.
We may want to consider adding some file(GLOB code here to
minimize the risk of missing files added in future versions
of VS.
On OSX, the output from gcc looks like this:
/usr/include/c++/4.2.1
/usr/include/c++/4.2.1/i686-apple-darwin10/x86_64
/usr/include/c++/4.2.1/backward
/usr/lib/gcc/i686-apple-darwin10/4.2.1/include
/usr/include
/System/Library/Frameworks (framework directory)
/Library/Frameworks (framework directory)
End of search list.
The "(framework directory)" part needs to be removed so that Eclipse handles it properly
Alex
Remove old search paths that aren't needed.
Keep using PATHS instead of HINTS because a Windows machine may have
a different Qt in its PATH and putting QTDIR and the registry entry
ahead of PATH could cause apps to fail when run.
Add options HG_REPOSITORY and HG_TAG to specify an external project
hosted in a Mercurial repository. Teach ExternalProject to clone the
repository and update from it. Extend the ExternalProject test to try a
Mercurial repository when hg is available.
Previously, it was inconsistent in that some platforms/compilers
had this flag for the RelWithDebInfo configuration and some didn't.
This fixes issue #11366.
Since commit 571dc748 (Recognize Clang C and C++ compilers, 2010-05-17)
we recognize Clang C and C++ support. Add Compiler/Clang-ASM.cmake to
enable use of Clang for ASM too. Also teach Assembler test to try Clang
as an assembler.
Suggested-by: Tobias Pape <tobiaspape@gmail.com>
Especially remove "lib64" when the given paths are all Unix ones and "lib" is
also explicitely given. In that case CMake will search "lib64" anyway for
platforms where it is known to make sense.
Instead of directly passing $ENV{SOMEVAR} to a find_* call pass in ENV SOMEVAR.
This will make sure the paths will get correctly handled through different
platforms, especially on Windows.
Also fixes one place where paths with windows delimiters (\) were hardcoded to
use forward slashes.
This was missed by commit 7bbaa428 (Remove trailing whitespace from most
CMake and C/C++ code, 2012-08-13) which only removed trailing spaces,
not TABs.
Ancient versions of CMake required else(), endif(), and similar block
termination commands to have arguments matching the command starting the
block. This is no longer the preferred style.
Run the following shell code:
for c in else endif endforeach endfunction endmacro endwhile; do
echo 's/\b'"$c"'\(\s*\)(.\+)/'"$c"'\1()/'
done >convert.sed &&
git ls-files -z -- bootstrap '*.cmake' '*.cmake.in' '*CMakeLists.txt' |
egrep -z -v '^(Utilities/cm|Source/kwsys/)' |
egrep -z -v 'Tests/CMakeTests/While-Endwhile-' |
xargs -0 sed -i -f convert.sed &&
rm convert.sed
7e58e5b Prefer generic system compilers by default for C, C++, and Fortran
796e337 Factor common code out of CMakeDetermine(ASM|C|CXX|Fortran)Compiler
b708f1a CMakeDetermine(C|CXX)Compiler: Consider Clang compilers
Since commit c198730b (Detect Watcom compiler version with its id,
2011-12-07) the CMAKE_(C|CXX)_COMPILER_VERSION variables are set for the
Watcom compiler. Use these in Windows-wcl386.cmake to set the old
WATCOM1* version variables. This avoids using the old EXECUTE_PROCESS
command which failed due to extra quotes anyway.
This fix bug #0013451. The bug prevents theorerically relocatable RPM package
to be installed properly.
Signed-off-by: Eric NOULARD <eric.noulard@gmail.com>
Teach CMake to prefer the system default compiler automatically when no
compiler is specified. By default use "cc" for C, "CC" for C++, and
"f95" for Fortran. Load a new Platform/<os>-<lang>.cmake module to
allow each platform to specify for each language its system compiler
name(s) and/or exclude certain names.
Create Platform/(CYGWIN|Darwin|Linux|Windows)-CXX.cmake modules to
specify "c++" as the system C++ compiler name for these platforms. On
systems that use case-insensitive filesystems exclude C++ compiler names
that are distinguished from C compiler names only by case.
This will change the default compiler selection for existing build
scripts that do not specify a compiler when run on machines with
separate system and GNU compilers both installed in the PATH. We do not
make this change in default behavior lightly. However:
(1) If a given build really needs specific compilers one should specify
them explicitly e.g. by setting CC, CXX, and FC in the environment.
(2) The motivating case is to prefer the system Clang on newer OS X
systems over the older GNU compilers typically also installed. On
such systems the names "cc" and "c++" link to Clang. This is the
first platform known to CMake on which "c++" is not a GNU compiler.
The old behavior selected "gcc" for C and "c++" C++ and therefore
chooses GNU for C and Clang for C++ by default. The new behavior
selects GNU or Clang consistently for both languages on older or
newer OS X systems, respectively.
(3) Other than the motivating OS X case the conditions under which the
behavior changes do not tend to exist in default OS installations.
They typically occur only on non-GNU systems with manually-installed
GNU compilers.
(4) The consequences of the new behavior are not dire. At worst the
project fails to compile with the system compiler when it previously
worked with the non-system GNU compiler. Such failure is easy to
work around (see #1).
In short this change creates a more sensible default behavior everywhere
and fixes poor default behavior on a widely-used platform at the cost of
a modest change in behavior in less-common conditions.
The compiler candidate list selection and search code for C, C++, ASM,
and Fortran languages was duplicated across four modules. To look for
compilers adjacent to already-enabled languages the C and CXX modules
each used _CMAKE_USER_(C|CXX)_COMPILER_PATH and the ASM module used
_CMAKE_TOOLCHAIN_LOCATION. Since commit 4debb7ac (Bias Fortran compiler
search with C/C++ compilers, 2009-09-09) CMake prefers Fortran compilers
matching the vendor and directory of an enabled C or C++ compiler.
Factor out the common functionality among the four languages into a new
CMakeDetermineCompiler module. Generalize the Fortran implementation so
that all languages may each use the vendor and directory of the other
languages that have already been enabled. For now do not list any
vendor-specific names for C, C++, or ASM so that only the directory
preference is used for these languages (existing behavior).
Look for "clang" or "clang++" compiler executables so Clang will be used
when it is the only compiler available. Prefer them last to avoid
changing compiler default preferences for existing scripts.
In case that any of the input variables that hold the library names contains
more than just a single library the "debug" or "optimized" keywords were only
prepended to the first item, making all other libs appear in all
configurations. Just treat both input variables as lists.
Thanks to Philipp Berger <newsletters@philippberger.de> for pointing me at
this.
4bb94c9 Ninja: sysconf() is declared in unistd.h
bb36759 Ninja: enable response file support on Mac (length 262144)
3a2c8e8 Ninja: disable work around when linking with mingw
3856e66 Ninja: error on missing rspfile_content
8c1e35c Ninja: remove some unused default arguments
7f647cf Ninja: also write link libraries to rsp file
The work around is only needed by older GCCs (only testet 4.4/4.7)
Ninja is very new so chances are high that there is also a new mingw.
Use slashes in link rsp file, because ar.exe can't handle \.
Among other flags this sets RPATH flags correctly so that CMake knows
how to treat CMAKE_PLATFORM_REQUIRED_RUNTIME_PATH for the ASM language.
This is the GNU compiler equivalent to commit a0bab7ae (Add ASM platform
information for XL compiler on AIX, 2011-03-02), made for XL.
Fixes a problem where when specifying a version number without the REQUIRED
parameter, Boost_FOUND would be true on the first configure but false on
subsequent configures.
0331a5a Qt4Macros: add some quotes to prevent damage from spaces in the paths
f46903b Qt4Macros: improve basename extraction in QT4_ADD_DBUS_INTERFACES
aa841ae FindQt4: extend documentation
-convert the filename to lowercase before the extraction, this allows this to
work even if the filename as uppercase ".XML" extension
-use get_filename_component(... NAME) to strip the path
eb410e8 Ninja: disable cldeps for bcc32, it's too old, and ninja would also not build
5ead31d Ninja: try work around for bcc32 bug
1333b57 Ninja: build server fixes
9081e3a remove warning about unused parameter
f430bea Ninja: maybe this fixes the bcc32 build
f2c1288 Ninja: msvc6 for-scoping
44b9bbc Ninja: build with old msvc versions
57156a5 Ninja: build server fixes
f1abdce Ninja: some bytes of the rc files couldn't be piped correctly
2de963d Ninja: don't remove space between command and parameters
50b6f33 Ninja: build cmcldeps with mingw
c05653e Ninja: try to make GetProcessId visible
ab245ff Ninja: but cl supports /nologo ...
bf58e9a Ninja: no /nologo option in old rc.exe
2fb07fc Ninja: Eclipse and KDevelop fixes for ninja
518c065 Ninja: don't pollute build dir with preprocessed rc files
...
Add "LinearMath_Debug" to the list of names to search for
BULLET_MATH_LIBRARY_DEBUG. The _Debug name is present in all other
lines already. Now the script manages to find all debug&release
libraries on my system.
bd34963 Refactor generation of shared library flags
55d7aa4 Add platform variable for flags specific to shared libraries
31d7a0f Add platform variables for position independent code flags
Store in new platform variables
CMAKE_${lang}_COMPILE_OPTIONS_PIC
CMAKE_${lang}_COMPILE_OPTIONS_PIE
flags for position independent code generation.
In almost all cases, this means duplication of the
CMAKE_SHARED_LIBRARY_${lang}_FLAGS for the _PIC case and using the
assumed pie equivalent for the _PIE case. Note that the GNU compiler
has supported -fPIE since 3.4 and that there is no -fPIC on GNU for
Windows or Cygwin.
There is a possibility that the _PIE variables are not correct.
However, as there is no backwards compatibility to be concerned about
(as the POSITION_INDEPENDENT_CODE property is not used anywhere yet),
the current state suffices.
cmcldeps wraps cl and adds /showInclude before calling cl.
It parses the output of cl for used headers, drops system
headers and writes them to a GCC like dependency file.
cmcldeps uses ATM ninja code for process handling,
but could be ported later to SystemTools.
TODO: Why needs ninja multiple calls in the BuildDepends test?
The GnuWin32 "bzip2" installer stores in the registry key
HKEY_LOCAL_MACHINE\SOFTWARE\GnuWin32\Bzip2
an "InstallPath" value. Use this entry as a search location.
3545645 Exclude the CompileCommandOutput test on WIN32.
fbaddf4 Escape the source file to be compiled if required.
db839be Make the CMAKE_EXPORT_COMPILE_COMMANDS option work with Ninja.
8778357 Add newline to the output.
2c04bc0 Move the EscapeJSON method to a sharable location.
The default for `CMAKE_FIND_FRAMEWORK`, defined in `Darwin.cmake` and
`Darwin-icc.cmake`, is now guarded so that it will not override command line
arguments passed by users.
Similarly for `CMAKE_FIND_APPBUNDLE`
If on APPLE, the phonon backend plugin is set to phonon_qt7. If on WIN32, the phonon backend plugin is set to phonon_ds9. I did not add any for generic UNIXes as they could have a whole host of things. A more comprehensive script would actually attempt to detect which phonon plugins were installed. However, this is a simple fix for now that will work for most people.
Using the QT4_ADD_RESOURCES() macro required the resource file to exist
already for dependency scanning. This prevented the use of resource
files that are generated by cmake because it needs the file before
resolving the dependency. This patch adds support for generated
resource files by not scanning the resource file for dependencies if it
doesn't exist yet. The add_custom_command() to generate the resource
file should add the dependencies on the files used by the resource file.
14b213c add test for #13241: empty SIZEOF_VOIDP in write_basic_package_version_file
00ae36f write_basic_package_version_file() now works with unset CMAKE_SIZEOF_VOID_P
Fix the build command when Ninja is the CMake Generator, but the
external project is created for a project that does not use CMake but
does use "make".
Until now an unnamed component was always named "Unspecified".
Now this name is taken from the new cmake variable CMAKE_INSTALL_DEFAULT_COMPONENT_NAME,
which is initialized to "Unspecified". But it can now be set to something
project-specific, per directory
Alex
Add a boolean target property NO_SONAME which may be used to disable
soname for the specified shared library or module even if the platform
supports it. This property should be useful for private shared
libraries or various plugins which live in private directories and have
not been designed to be found or loaded globally.
Replace references to <CMAKE_SHARED_LIBRARY_SONAME_${LANG}_FLAG> and
hard-coded -install_name flags with a conditional <SONAME_FLAG> which is
expanded to the value of the CMAKE_SHARED_LIBRARY_SONAME_${LANG}_FLAG
definition as long as soname supports is enabled for the target in
question. Keep expanding CMAKE_SHARED_LIBRARY_SONAME_${LANG}_FLAG in
rules in case third party projects still use it. Such projects would
not yet use NO_SONAME so the adjacent <TARGET_SONAME> will always be
expanded. Make <TARGET_INSTALLNAME_DIR> NO_SONAME aware as well. Since
-install_name is soname on OS X, this should not be a problem if this
variable is expanded only if soname is enabled.
The Ninja generator performs rule variable substitution only once
globally per rule to put its own placeholders. Final substitution is
performed by ninja at build time. Therefore we cannot conditionally
replace the soname placeholders on a per-target basis. Rather than
omitting $SONAME from rules.ninja, simply do not write its contents for
targets which have NO_SONAME. Since 3 variables are affected by
NO_SONAME ($SONAME, $SONAME_FLAG, $INSTALLNAME_DIR), set them only if
soname is enabled.
By tracking a stamp file within the git clone script itself.
Avoids a 2nd git clone operation after switching from Debug
to Release builds in Visual Studio, or vice-versa.
Add "private/internal-use-only" function _ep_get_step_stampfile
to get the name of the stamp file for a given step.
The functionality provided by this commit should be identical
to its parent commit.
In the case of git, only track the repository in the
repository info dependency tracking file. Not the tag.
The download step should only re-run if the repository changes.
The download step should NOT re-run if the tag changes.
The update step is an 'always' re-running step, and so should
already re-run, unless it's been eliminated by use of
UPDATE_COMMAND ""
Use of the deprecated option with Intel 2011 produces
icl: command line remark #10010: option '/GX' is deprecated and will
be removed in a future release. See '/help deprecated'
so use its replacement option which has been supported for several
older versions anyway.
If CPACK_NSIS_ENABLE_UNINSTALL_BEFORE_INSTALL is set to ON the NSIS installer will look for a previous installed version and ask the user about uninstall.
Commit f67139ae added running a verify script in between running
the download and extract scripts. Since then, it has always been
missing the COMMAND keyword added in this commit.
It worked anyway (semi-accidentally) by running a command line like:
cmake -P script1.cmake cmake -P script2.cmake
CMake, when running -P scripts on the command line, runs them in order,
and apparently ignores spurious arguments in between (the middle "cmake"
in the above example) and so, all appeared to work as intended.
This commit adds the missing keyword and the commands that run are
now two separate sequential cmake invocations like:
cmake -P script1.cmake
cmake -P script2.cmake
...which was the original intent of commit f67139ae
Allows custom NSIS commands to run prior to any installation
actions. Projects that need to run an uninstaller first,
especially one from a non-NSIS previous revision of a project
that is NOW using CPack and NSIS, may do so by putting custom
NSIS commands into this variable.
Inspired-by: David Golub
As Dave Abrahams pointed out CMAKE_CURRENT_SOURCE_PATH is wrong, it's of
course CMAKE_CURRENT_SOURCE_DIR.
Also wrap the path in quotes so the example would even work if the source path
has spaces.
6190415 OS X: Mark find_program results as advanced
d9edf46 OS X: Use correct extra path when searching for applicaton bundles (#13066)
98b9a7f OS X: Use OSX_DEVELOPER_ROOT for app search path (#13066)
Since commit 44d007b6 (CheckIncludeFiles: fix status output, 2012-02-01)
check_include_files reports the list of files tested instead of the name
of the variable storing the result. Some projects incrementally test
and concatenate very long lists leading to long messages that do not
provide much information. Users report confusion especially when the
lines wrap.
For lists of more than two files produce messages of the format
Looing for N include files first.h, ..., last.h
where N is the list length and "..." is literal. Leave the log file
entries and cache entry description unchanged as they should have the
full detail of the check performed.
The parent commit added a search path relative to OSX_DEVELOPER_ROOT.
But with Xcode 4.3 the nested Applications folder is in a different
relative location compared to that root. This commit makes the intent
of the previous commit work with older and newer Xcode directory layouts.
Furthermore, it only adds paths that exist to the search path.
Starting with Python3, standard Python installs may have additional ABI
flags attached to include directories and library names. As of 3.2, the
following flags are in the configure file:
d -> --with-debug
m -> --with-pymalloc
u -> --with-wide-unicode
Python 3.3 seems to no longer have --with-wide-unicode. Hopefully Python
will ensure that the possible flags always show up in a stable order.
The 'd' flag is ignored since the debug library is considered separate.
There is still the problem where ABI flags cannot be specified in
find_package since the letters confuse the version comparator.
If PYTHON_INCLUDE_PATH is put into the cache, then it will always
override whatever might be found and PYTHON_INCLUDE_DIR is never given a
chance to find something different. It being marked as INTERNAL also
means that it cannot be changed without editing CMakeCache.txt directly.
Basically, the scenario is that if the Python version is changed, then
deleting PYTHON_INCLUDE_DIR doesn't work because any cached
PYTHON_INCLUDE_PATH variable is set before find_path is even called. Any
build tree using a previous version will still need either manual
removal of PYTHON_INCLUDE_PATH or a complete reconfigure, but in the
future changing the Python version can be accomplished by deleting
PYTHON_INCLUDE_DIR and reconfiguring with the new version.
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
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.
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
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.
0f4dfa6 CPack: Use real path to PackageMaker to find its version file (#12621)
4693cf8 Xcode: Detect new default locations of Xcode 4.3 bits and pieces (#12621)
As discussed on cmake-developers, this patch adds a macro
check_required_components() to the file configured via
configure_package_config_file(), so for proper handling
of components in Config.cmake files users can simply call
check_required_components(PackageName)
and this will do the right thing.
Alex
if the HANDLE_COMPONENTS is used, FPHSA() now also checks all required COMPONENTS,
i.e. all elements from <name>_FIND_COMPONENTS for which <name>_FIND_REQUIRED_<comp>
is true, and sets <name>_FOUND only to true if all have been found.
As discussed on cmake-developers.
Alex
This patch adds documentation for OPTIONAL_COMPONENTS to cmFindPackage.cxx,
and also extends Modules/readme.txt significantly with regard to how
components should be handled.
Alex
8485208 Ninja: shell escape $(CMAKE_SOURCE_DIR) and $(CMAKE_BINARY_DIR)
df84767 Ninja: add support for OBJECT_OUTPUTS, fix PrecompiledHeader test case
48eb7fc Ninja: Avoid using 'this' in member initializers
bba37dd Ninja: Fix for PDB files with spaces in the path.
ac800f4 Ninja: Constify use of cmCustomCommand
9a0d5a8 Ninja: add /DEF: flag to linker call
d40eebd Ninja: Add a cache option CMAKE_ENABLE_NINJA to enable the ninja generator.
8c63433 Ninja: Add friend struct so it can access the private ConvertToNinjaPath.
dbe3dce Ninja: add .def file support
f1bb08f Ninja: ensure the output dir exists at compile time
7a6b5f4 Ninja: Remove an unnecessary variable
80ff210 Ninja: Use cmSystemTools::ExpandListArgument to split compile/link commands
d2731a3 Ninja: Add a missed license header
eabc9b0 Ninja: CMake: Adapt Ninja generator for per-target include dirs
bada88e Merge branch 'target-include-directories' into ninja-generator
54bd175 Ninja: windows msvc: create for each target a .pdb file
...
Xcode 4.3 installs into "/Applications" by default, from the Mac App Store.
Also, the paths to the available SDKs changed: they are now within the
Xcode.app bundle.
PackageMaker is installed as a separate program, and may be installed
anywhere. It is not installed with Xcode 4.3 by default anymore.
Download the "Auxiliary Tools for Xcode" to get PackageMaker.
Put PackageMaker inside the Xcode.app bundle, in its nested Applications
folder, or put it alongside Xcode in "/Applications" and CMake will find
it.
Update references to "find" paths: add new possible locations for finding
Xcode.app and PackageMaker.app. Prefer the most recent version's locations
first, but keep the old locations as fallback search paths, too.
Thanks to all the contributors who provided and tested out various patches
for fixing this issue. Especially, but by no means limited to:
Francisco Requena Espí, Jamie Kirkpatrick and drfrogsplat.
The function configure_package_config_file() may be used instead of
configure_file() for generating the Config.cmake files for installation,
they help to make those files relocatable.
Alex
678c24d FindQt3: fix detection of Qt3 include directory
91a1670 FindQt3: let FPHSA handle the version selection
9dae4d6 FindQt3: fix version extraction for versions with letters
2cd898d FindQt3: fix warning when Qt3 is not found
4689eed reflect that the QtAutomoc depends on QtGui
ab9661c Remove QtGui dependency in Qt4Deploy test and verify QtSql existance.
52e8279 Fix for Qt4Deploy on some test machines.
672e3bb Add test for DeployQt4.cmake
4853e1e Fix plugin installation issues.
35cbf23 Ensure libs are passed to BundleUtilities.
0ac1535 Fix bad plugin paths.
a2123e8 Fix mismatched arguments.
fc6f340 Don't use QT_LIBRARIES_PLUGINS by default.
53d02ea FindPythonLibs: stop scanning when libraries are found
91d5a2a FindPythonLibs: put debug libraries into PYTHON_LIBRARIES
c9c1a17 FindPythonLibs: get the exact version of the found library (#3080)
f772378 FindPythonLibs: make the version selection work as for PythonInterp
bbddaee FindPython{Interp,Libs}: document Python_ADDITIONAL_VERSIONS as input
There are versions out there that neither understand --version nor -V. Try a
completely different approach: execute a small python script that prints the
version number (and only that) in an easily reusable way using
sys.version_info. This is documented to work since Python 2.0. Use sys.version
for older versions, which is documented to exist since 1.5. If even that
doesn't work then simply assume we are on 1.4.0.
This concerns all variables common to all CPack generators.
Variables mainly used and/or set in CPack.cmake are documented
therein. C++ built-in variables are documented in
cmCPackDocumentVariables.cxx.
Use CMAKE_<LANG>_COMPILER_VERSION instead of calling the compiler. This macro
predates those useful variables. This also fixes the issue that g++ version
detection was not working if C language was not enabled.
35c48e1 Check*.cmake: Expand imported targets in CMAKE_REQUIRED_LIBRARIES
61cb4ea bootstrap: move while() and endwhile() into the bootstrap build
c9f2886 -don't pull in CheckTypeSize.cmake from the cmake which is being built
628f365 -remove trailing whitespace
Add the function cmake_expand_imported_targets() to expand imported
targets in a list of libraries into their on-disk file names for a
particular configuration. Adapt the implementation from KDE's
HANDLE_IMPORTED_TARGETS_IN_CMAKE_REQUIRED_LIBRARIES which has been in
use for over 2 years. Call the function from all the Check*.cmake
macros to handle imported targets named in CMAKE_REQUIRED_LIBRARIES.
Alex
A lot of the libraries won't be found by default so will spit out
all sorts of errors. Don't kill QT_LIBRARIES_PLUGINS altogether as
it is sometimes useful.
The Borland compiler was re-branded as CodeGear during 2007-2009 and
since 2009 is the Embarcadero compiler. They offer predefined macros:
http://docwiki.embarcadero.com/RADStudio/en/Predefined_Macros
and distinguish themselves by __CODEGEARC__ and __CODEGEARC_VERSION__.
Version 6.30 (C++Builder XE) changed the meaning of some flags:
http://docwiki.embarcadero.com/RADStudio/en/C%2B%2B_Compiler_Option_Changes_for_XE
Teach Embarcadero compiler information files to generate build rules
with flags matching the compiler version. Leave the flags unchanged
for old Borland versions. Always set the BORLAND toolchain indicator
for compatibility with existing projects that test it. Also set the
EMBARCADERO indicator for newer toolchains.
The GenerateExportHeaders test was failing on one machine, the version
could not be determined there, so the _gcc_version was empty,
so the first argument to if() was empty, so it complained:
http://open.cdash.org/testDetails.php?test=135623436&build=2016288
Use double quotes to turn the non-existant first argument into an empty
string.
Alex
Since we know which compiler we have we can test those OpenMP flags first that
are likely to be correct. This doesn't make any difference for GNU compilers,
but it should avoid useless try_compiles and output cluttering for all others.
The Borland compiler is now the Embarcadero compiler. Rename the shared
platform information file to reflect this. This does not change the
interface, as old versions are still "Borland", but will allow new
versions released by Embarcadero to be supported cleanly.
Newer Ruby versions (from 1.9 onward) seem to warn if you query Config::CONFIG
and print a warning to use RbConfig instead. RbConfig seems to also work in
older versions, at least in 1.8. Use a macro to query RbConfig first and only
if that doesn't give anything fall back to Config.
This contains a change, which changes the behaviour a bit:
now X11_xf86vmode_FOUND is only set to TRUE and the include directory
is added to X11_INCLUDE_DIR, if additionally to X11_xf86vmode_INCLUDE_PATH
also X11_Xxf86vm_LIB has been found.
I hope this doesn't cause regressions somewhere.
Alex
Configure the build_mingw.cmake.in config_mingw.cmake.in files
into the binary directory of the directory being built, not the
top level binary directory for the project.
This also cleans up a bunch of things on the way:
-when perl was queried for paths they were not converted to CMake style on
Windows.
-the result when perl was queried for the perl library name was ignored since
it was expanded with the possible paths, which is not a valid input for
find_library(). If perl returns a library name we now will look only for this
name and not for the default names and use the default names only when the
executable does not give us a hint.
-get rid of 2 variables that were only used at one place and directly put the
values in the call to find_library() and find_path().
Inspired by Jeff Trull
1e16406 CMakeAddFortranSubdirectory: Add NO_EXTERNAL_INSTALL option
6f6891b CMakeAddFortranSubdirectory: Always parse arguments
48a09f8 CMakeAddFortranSubdirectory: Make IMPORTED targets GLOBAL
067c1f4 VSGNUFortran: Disable test in special cases
bd69e1c VSGNUFortran: Add special case for SunPro Fortran runtime library
414a780 CMakeAddFortranSubdirectory: Validate gfortran architecture
7e0d9f1 CMakeAddFortranSubdirectory: Find gfortran in PATH
d6b0312 CMakeAddFortranSubdirectory: Fix documentation format and typos
e4ae038 CMakeAddFortranSubdirectory: Allow full paths to directories
538c345 Add CMakeAddFortranSubdirectory to use MinGW gfortran in VS
3c6af5f Merge branch 'add-CheckLanguage-module' into CMakeAddFortranSubdirectory
d4b77eb Avoid discovering system infos for documentation. Adding some path is enough.
9002f73 Fix non existent std::string::clear on VS6
02ccb32 Create getDocumentedModulesListInDir which may be used in other context.
24fbc28 Add missing section markup for CPackComponent
bafd8a9 Example of builtin variable documentation (i.e. only used in C++ source code).
543f1ad Make the load of script documentation more efficient and dynamic.
cdbd1a9 Fix another compiler warning due to a typo
52c53de Really avoid compiler warning about unused vars
37f90ed Calm down compiler warning about unused var
7c82b7f Fix potential bad memory access, thanks to Eike
62b589b Suppress unused var, beautify code, avoid 1 extra newline.
751713f Update bash completion file in order to handle new CPack doc options.
1629615 CPack Documentation extraction from CMake script begins to work
83e34dd Implement simple CMake script comment markup language.
c6a0169 CPack begin the implementation of --help-command* and --help-variables*
CPack help will be searched in any CPack*.cmake file located
near to CPack.cmake file. The script files is parsed iff
the first line begin with ##section. Moreover the documentation
section name is specified on the remaining part of the line
minus the space immediately following ##section.
The functions in FindGettext create a custom target. If the functions
are called multiple times, multiple times the same target is created.
This works only if CMP0002 is set to OLD.
With this patch there is only one central target created, and each
invocation of the function creates a target with a unique name and
make the central target depend on this one.
Alex
Also, removed detection of header file from <PREFIX>/include.
Can't find any example in alsa source code where the library headers
were installed outside of <PREFIX>/include/alsa.
We do not yet support "make install" in the external project case.
Document this explicitly in the interface. Require the caller to use an
option to "disable" the unsupported behavior. This will allow us to add
the behavior by default in the future without clobbering existing
projects that handle the installation themselves.
cmake_add_fortran_directory uses imported targets when using the
mingw fortran compiler. This change makes those targets global
in scope so they act just like the real targets that exist when
a fortran compiler exists and regular add_subdirectory is used.
Verify that MINGW_GFORTRAN not only points to a MinGW gfortran but also
one that compiles for the target architecture. This prevents using a
32-bit gfortran in a 64-bit MSVC build.
In the find_program(MINGW_GFORTRAN) call use the PATHS option for
hard-coded guesses instead of HINTS. This allows the user environment
to override the guesses and corrects usage of the command options.
Fix the implementation to allow full paths with spaces. Change the
interpretation of relative paths to be with respect to the current binary
directory. This matches the convention used in ExternalProject. Test
both full and relative paths in the VSGNUFortran test.
This patch adds a new module that allows for easy integration of MinGW
gfortran and the Visual Studio compiler. It is done in a function called
cmake_add_fortran_subdirectory. The patch also includes a test for this
feature.
Define a "check_language(<lang>)" macro to test whether <lang> can be
enabled. Cache the result in CMAKE_<lang>_COMPILER. Add a test case
covering expected results.
If CXX or Fortran is enabled before C then the values of
CMAKE_SHARED_MODULE_C_FLAGS
CMAKE_SHARED_MODULE_CREATE_C_FLAGS
may not be available. On platforms where MODULE library (plugin) creation
is the same as SHARED library creation initialize the MODULE creation
flags from the SHARED creation flags of the matching language instead of
assuming that C has been enabled first.
Teach the COnly and CxxOnly tests to build MODULE libraries. The latter
covers this specific case.
Causes compiler modules (currently only GNU) to set a
CMAKE_DEPFILE_FLAGS_${lang} variable, which communicates to
the generator the flags required to cause the compiler to create
dependency files.
CHECK_INCLUDE_FILES("foo.h" HAVE_FOO_H) gave an output like:
Looking for include files HAVE_FOO_H
After this change it does now what CHECK_INCLUDE_FILE() also does:
Looking for include files foo.h
When using the NSIS generator from CPack the file NSIS.template.in is
used to generate a project.nsi file for NSIS to process. The file
consists code in the NSIS scripting language. Among other functions
there is an onInit function the initializes the installer. The function
(tries to) recognise admin and power users but fails since NSIS
scripting language relative includes the jump from the current command
so +3 means "run the third command after this one", so a failed check
for admin completely skips the check for a power user and goes directly
to "done:".
User permission lookup was added in initial NSIS support by commit
a11b9a4c (Merge from CPack branch, 2006-01-01). Later commit b1b052fd
(Several changes to for NSIS, 2006-03-01) added a line inside a block
that should be skipped by a jump without updating the jump length.
Update the jump length to correct the behavior.
If the debug and release libraries are the same (which usually means only one
of them was found) do not output the library as "optimized" and "debug", but
just as one plain library. At the end this means that the Find* output of the
avarage (Un*x) user will be much less cluttered.
RUBY_VERSION was always set, even if no RUBY_EXECUTABLE was found. While it
may make sense to assume a default version if we can't execute the binary, it
certainly doesn't make sense to report a version if there is no executable at
all.
98d2031 Fix BundleUtilities test failure with space in build path.
36d6641 Fix new BundleUtilities test failure on Mac 10.4.x
0d96dec GetPrerequisites: Add test for @rpath support.
880139a GetPrerequisites: Add support for @rpath on Mac OS X.
9a6b102 GetPrerequisites: Add support for @rpath on Mac OS X.
bb2b264 FindOpenSSL: also parse version number define with uppercase letters
7053a00 FindOpenSSL: only try to parse opensslv.h if it exists
44ba7a3 Merge branch 'master' of git://cmake.org/cmake into openssl-version
8e8672c FindOpenSSL: improve version number handling
- Enhance extract doc parser. Seems robust now. The legacy
module documentation parser works as before ignoring
the new markup.
- Proof of concept for CPack (generic), CPack RPM and CPack Deb
generator for macro and variables.
Try cpack --help-command and cpack --help-variables
The language is very simple. It use ##<keyword> special comment
which opens a structured documentation block and ##end closes it.
This may be used to extract documentation for macro as 'command'
and 'variables' such that cpack --help-command and --help-variable
does parse builtin modules files (CPack.cmake, CPackComponent.cmake,
...) in order to extract the corresponding doc.
QNX has the phtread stuff in the standard library. The best way would
IMHO be to check if a program that uses pthread_* can be successfully
linked without specifying any linker option before trying out the
different flags.
Change to consider a library embedded if it is found in a subdirectory relative to the
using executable/library. Previous commit considered them local.
This case is encountered when @rpath is used with framework libraries, which are inside a directory tree.
On dashmacmini2 the test showed output like this:
-- Found PythonInterp: /usr/bin/python (found version "Unknown option: --
usage: /usr/bin/python [option] ... [-c cmd | file | -] [arg] ...
Try `python -h' for more information.")
On my machine where python outputs "Python 2.7" this worked, but
PYTHON_VERSION_MAJOR, PYTHON_VERSION_MINOR, and PYTHON_VERSION_PATCH were all
set to "2.7".
Add some checks that the version output has the expected form before using it.
This allows the developer to tell FindPythonInterp which Python version should
be searched for. This allows the right version to be chosen for a project
without user assistance if there are specific requirements. This is especially
useful as it is common to have major versions 2 and 3 installed in parallel,
which are partly incompatible.
6856b4d Merge topic 'link-shared-depend-cycle-issue-12647' into check_symbol_exists
8e1f376 add a test for Check{,CXX}SymbolExists
813eca6 CheckSymbolExists: force the compiler to keep the referenced symbol
0df1942 Detect SGI MIPSpro compiler version with its id
a5e892c Document compiler version macro formats used for detection
d7c6f41 Detect HP compiler version with its id
3dd9fa9 Detect SunPro compiler version with its id
c198730 Detect Watcom compiler version with its id
5899b98 Detect Clang compiler version with its id
b8cfa65 Detect PGI compiler version with its id
6dae666 Detect IBM XL compiler version with its id
4080d55 Detect Borland compiler version with its id
2cc205a Detect Intel compiler version with its id (#11937)
a6d83cc Detect MSVC compiler version with its id
a662855 Detect GNU compiler version with its id (#6251)
fa7141f Add framework to detect compiler version with its id (#12408)
Otherwise the compiler may optimize out the reference to the symbol as the
previous version was not really using this. This leads to symbols that are
only in a header but not in the given libraries to be reported as present.
This came up on the first try to fix bug 11333 as "gcc -O3" would optimize
out the reference to pthread_create() so the correct library the symbol is in
was not detected.
The new test code was suggested by Brad King.
When we have no MPI compiler wrapper and search explicitly for the MPI
C++ library append it correctly to the list of libraries instead of
using a space.
Suggested-by: Mourad Boufarguine <bouffa@gmail.com>
This fix bug #12863 whose symptom was a lot of "warning: File listed twice"
printed out by rpmbuild when processing the spec file.
Signed-off-by: Eric NOULARD <eric.noulard@gmail.com>
First this fixes the bug that e.g. version "1.0.0" was shown as "1..". When
pkg-config was used to find OpenSSL the header file was parsed for the version
number even if pkg-config returned it already. Finally we also include the
patch level (i.e. the letter after the version number) in OPENSSL_VERSION.
When GIT_EXECUTABLE points at ".../Git/cmd/git.cmd" in an msysGit
installation we previously failed to detect the version number in a
subtle case. The "git.cmd" assumes 'chcp' is in PATH. It is typically
available at "C:\Windows\System32\chcp.com". On 64-bit Windows the File
System Redirector maps this location to "C:\Windows\SysWOW64\chcp.com"
for 32-bit processes. However, some Windows installations fail to
provide chcp.com at this path. Whenever git.cmd runs in a 32-bit
command shell, as it does under a 32-bit CMake binary, it reports
'chcp' is not recognized as an internal or external command,
operable program or batch file.
on stderr. Capture stderr separately so it does not affect parsing
of the version number.
See also msysGit issue 358:
http://code.google.com/p/msysgit/issues/detail?id=358
Note that FindGit prefers "git.cmd" over "git.exe" because it sets up
the proper HOME environment variable necessary for Git ssh connections
to work the same as they do from the Git bash prompt.
Decode decimal digits from _SGI_COMPILER_VERSION or _COMPILER_VERSION to
compute version number components. See documentation at:
http://predef.sourceforge.net/precomp.html
The MSVC, HP, XL, SunPro, Watcom, Borland, and Intel compilers specify
their version number in components encoded in a single integer value.
Document the components that we use to compute version numbers.
Decode hex digits from __SUNPRO_C and __SUNPRO_CC to compute the version
number components. Note that the constant encodes decimal digits as hex
digits (never larger than 9). We represent them as decimal after
extraction. See documentation at
http://predef.sourceforge.net/precomp.html
Although the documented version number format is
0xVRP where V = Version, R = Revision, P = Patch
it holds only though SunPro C/C++ version 5.9. Later versions have
a two-digit revision (minor) number so their format is 0xVRRP.
When cmake searches for Python libs in Windows it searches in:
[HKEY_LOCAL_MACHINE\\SOFTWARE\\Python\\PythonCore\\${_CURRENT_VERSION}\\InstallPath]/libs
However, the information might not always reside there. The information
could also reside in:
[HKEY_CURRENT_USER\\SOFTWARE\\Python\\PythonCore\\${_CURRENT_VERSION}\\InstallPath]/libs
when one installs Python for a single user and not for all users.
Fix typo introduced in commit 66a08c10 (more uniform approach to enable
language, 2004-08-26). The optimization option should be /O2 for
Release configurations and /O1 for MinSizeRel.
Suggested-by: He Yuqi <yuqi.he@gmail.com>
This addresses Bug 11882 which provided a sample implementation for adding
support for cusparse. I went ahead and added all the libraries I thought
appropriate.
Added support for additional import paths during protoc invocation
time to the PROTOBUF_GENERATE_CPP public macro via a new
PROTOBUF_IMPORT_DIRS optional variable.
Patch courtesy of Miroslav Kes <mkes@ra.rockwell.com>
The default OS X 10.4 linker incorrectly searches for dependencies of
linked shared libraries only under the -isysroot location. It fails to
find dependencies of linked shared libraries in cases such as the
ExportImport test. It produces errors like:
/usr/libexec/gcc/i686-apple-darwin8/4.0.1/ld: warning can't open dynamic library:
libtestLib3Imp.dylib
referenced from: /.../ExportImport/Root/lib/libtestLib3lib.1.2.dylib
(checking for undefined symbols may be affected) (No such file or directory, errno = 2)
/usr/libexec/gcc/i686-apple-darwin8/4.0.1/ld: Undefined symbols: _testLib3Imp
referenced from libtestLib3lib expected to be defined in
libtestLib3Imp.dylib
or with CMAKE_SKIP_RPATH off to enable install_name in the Export side:
/usr/libexec/gcc/i686-apple-darwin8/4.0.1/ld: warning can't open dynamic library:
/Developer/SDKs/MacOSX10.4u.sdk/.../ExportImport/Export/impl/libtestLib3Imp.dylib
referenced from: /.../ExportImport/Export/libtestLib3lib.1.2.dylib
(checking for undefined symbols may be affected) (No such file or directory, errno = 2)
/usr/libexec/gcc/i686-apple-darwin8/4.0.1/ld: Undefined symbols:_testLib3Imp
referenced from libtestLib3lib expected to be defined in
/.../ExportImport/Export/impl/libtestLib3Imp.dylib
Note how "/Developer/SDKs/MacOSX10.4u.sdk" is prepended to the dependent
library path.
Commit 2cff26fa (Support linking to shared libs with dependent libs,
2008-01-31) and commit 82fcaebe (Pass dependent library search path to
linker on some platforms, 2008-02-01) worked around the problem by
defining platform variable CMAKE_LINK_DEPENDENT_LIBRARY_FILES. It tells
CMake to link to dependent libraries explicitly by their path thus
telling the linker where to find them.
Unfortunately the workaround had the side effect of linking dependent
libraries and defeats most benefits of LINK_INTERFACE_LIBRARIES.
Fortunately OS X 10.5 and above do not need to find transitive
dependencies at all so we can avoid the workaround on Modern OS X.
Previously we linked C, Fortran, and ASM shared libraries compiled with
the HP compiler using a direct invocation of the linker (ld). This
behavior was left historically from support for an ancient HP C compiler
that did not know how to create shared libraries. Fortran shared
libraries need to be linked with the compiler to get the language
runtime library dependencies as is already done for C++.
Update the HP-UX-HP* platform information to use the compiler front end
when linking shared libraries. This works on modern HP tools and
produces correct behavior. If there is a need to support older tools
again we can add a special case for them.
...if it matches "windres", as opposed to being exactly equal to "windres"
Cross-compiling windres compilers are named something like
"i686-w64-mingw32-windres" (for example)
automoc now defaults to strict mode, also with Qt4, i.e. it behaves as
the documentation says by default. I also inverted the switch
CMAKE_AUTOMOC_STRICT_MODE to CMAKE_AUTOMOC_RELAXED_MODE.
Docs and test adapted accordingly.
Alex
The User may now specific a list of file that shouldn't be
automatically handled by CPack but specified by the user.
Like %config(noreplace) or specific %attr.
The concerned files/dir lines will be removed from the set
automatically handled by CPack.
The mingw32-make tool does not handle parenthesis in the path to a
source file consistently. When CMake is installed in a typical location
like "c:\Program Files (x86)\CMake 2.8\" the mingw32-make tool fails on
the FortranCInterface detection project sometimes with errors like
>mingw32-make -f CMakeFiles\myfort.dir\build.make CMakeFiles/myfort.dir/my_module.f90.obj
mingw32-make: *** No rule to make target `x86)/CMake 2.8/share/cmake-2.8/Modules/FortranCInterface/my_module.f90)',
needed by `CMakeFiles/myfort.dir/my_module.f90.obj'. Stop.
due to parens in the path to the FortranCInterface source directory.
However, the behavior varies with the file name of build.make:
>copy CMakeFiles\myfort.dir\build.make CMakeFiles\myfort.dir\build2.make
>mingw32-make -f CMakeFiles\myfort.dir\build2.make CMakeFiles/myfort.dir/my_module.f90.obj
[ 3%] Building Fortran object CMakeFiles/myfort.dir/my_module.f90.obj
Tested with
>mingw32-make -v
GNU Make 3.82
Built for i386-pc-mingw32
Work around the problem by copying the whole FortranCInterface source
directory in to the project build tree.
2d11951 Merge branch 'master' into AutomocIncludedDotMocFileHandling
1eca18f automoc: add documentation for CMAKE_AUTOMOC_STRICT_MODE
bc278ce automoc: fix line length
62e223e automoc: add variable CMAKE_AUTOMOC_STRICT_MODE, to enable strict parsing
40c5167 automoc: accept even more .moc files in non-strict mode
c207f5d automoc: also accept other files when .moc is included in non-strict mode
9c0df72 automoc: add a StrictParseCppFile(), which is only qmake-compatible
174bf35 automoc: move the code for finding headers into separate function
8507eae automoc: fix handling of included _p.moc files
7ada172 automoc: some more linebreaks for the warnings for better readability
3b93e26 automoc: add extra check whether the header contains Q_PRIVATE_SLOT
4745715 Add a test case for the use of Q_PRIVATE_SLOT.
bde4edb automoc: add special handling for including basename_p.moc, with test
74ab0f6 automoc: move some code from the big parsing loop into separate functions
bc7560e automoc: add test for including a moc_abc_p.cpp file
30fd8e6 automoc: add test for including the moc file from another header
...
Teach CMakePlatformId.h to construct an "INFO:compiler_version[]" string
literal from macros COMPILER_VERSION_(MAJOR|MINOR|PATCH|TWEAK) to be
defined in CMake(C|CXX)CompilerId.(c|cpp) for each compiler. Provide
conversion macros DEC() and HEX() to decode decimal or hex digits from
integer values. Parse the version out of the compiler id binary along
with the other INFO values already present.
Store the result in variable CMAKE_<LANG>_COMPILER_VERSION in the format
"major[.minor[.patch[.tweak]]]". Save the value persistently in
CMake(C|CXX)Compiler.cmake in the build tree. Document the variable for
internal use since we do not set it everywhere yet.
Report the compiler version on the compiler id result line e.g.
The C compiler identification is GNU 4.5.2
Report CMAKE_(C|CXX)_COMPILER_(ID|VERSION) in SystemInformation test.
ae62a1c Test CMAKE_GNUtoMS option in ExportImport on MinGW and MSys
afb00fe Add CMAKE_GNUtoMS option to convert GNU .dll.a to MS .lib
61e8629 Factor makefile generator link rule lookup into helper function
a603250 Load platform files that need to know the ABI when possible
ecd8414 Fortran: Detect pointer size in gfortran on MinGW
0efe602 TinyCC: Add default compilation flags (#12605)
ec636e2 TinyCC: Add compiler info for shared libs on Linux (#12605)
1f49d72 Recognize the Tiny C Compiler (#12605)
This allows you to have more than source file with the same name but different
directories. The intermediate and configuration files are now in this same directory.
Nvcc can emit '/path' instead of '//path' which can cause a lot of grief later. We test
to see if the file exists, if it doesn't then we see if the file exists with '/'
prepended. Files that don't exist won't be added to the list.
Fix a long outstanding bug when a file in the dependency list wasn't found. This bug
wouldn't reset the dependencies, so the makefile would still want the missing file when
building. The work around was to configure twice, but this is no longer necessary.
Teach the Windows-GNU.cmake platform file to look for Visual Studio
tools matching the target ABI. Add an extra step to the link command
for shared libraries and executables that export symbols and on which a
new GNUtoMS property is set (initialized by the CMAKE_GNUtoMS option).
Tell the GNU linker to output a module definition (.def) file listing
exported symbols in addition to the GNU-format import library (.dll.a).
Pass the .def file to the MS "lib" tool to construct a MS-format DLL
import library (.lib).
Teach the install(TARGETS) command to install the MS import library next
to the GNU one. Teach the install(EXPORT) and export() command to set
the IMPORTED_IMPLIB property pointing at the import library to use the
import library matching the tools in the importing project.
Load platform files named in CMAKE_<lang>_ABI_FILES for each language
once the ABI sizeof(void*) is known. During the first configuration
this is after the test for working compiler and ABI detection checks.
During later configurations the ABI information is immediately available
because it has been saved in CMake<lang>Compiler.cmake.
Use __SIZEOF_POINTER__ which the GNU Fortran compiler defines at least
on 64-bit MinGW. Assume default size 4 on MinGW if gfortran does not
define the size.
Brad King
James Bigler
David Cole
Stephen Kelly
Rolf Eike Beer
Philip Lowman
Alex Neundorf
Clinton Stimpson
Marcin Wojdyr
Kitware Robot
Andy Piper
Zack Galbreath
Alexey Ozeritsky
Peter Kümmel
Eric NOULARD
Jonathan Klein
Mario Bensi
Charlie Sharpsteen
Kurtis Nusbaum
Tobias Bieniek
Matt McCormick
Anthony J. Bentley
Minmin Gong
David Faure
Modestas Vainius
Patrick Gansterer
Ben Boeckel
Tom Hughes
Kashif Rasul
Peter Collingbourne
Bill Hoffman
Droscy
Yury G. Kudryashov
Mike McQuaid
Nicolas Despres
Matthias Kretz
Mattias Helsing
Deborah Pickett
Christian Andersson
Michael Wild
Jason Erb
Tim Gallagher