CMake has a long-standing bug/feature (see [here](https://cmake.org/pipermail/cmake/2015-March/060204.html) and reply [here](https://cmake.org/pipermail/cmake/2015-March/060209.html)) which can mangle certain path variables by attempting to make them into relative paths if you try to set them with CACHE PATH.
Say you have your OpenCV download at `/path/on/my/computer/to/opencv/`. What actually happens is that if you try to set this variable by invoking CMAKE with `-DCMAKE_INSTALL_PREFIX=/my/desired/install/path`, what you end up is *not* `/usr/local/` and *not* `my/desired/install/path`, but instead, this monstrosity: `/path/on/my/computer/to/opencv/src/OpenCV-build//my/desired/install/path`.
That is, CMake attempts, for some reason, to turn the path that you passed into a path relative to `${CMAKE_BINARY_DIR}`. See the links I posted above: this is a known (and apparently unfixable) issue with CMake.
In OpenCV's case, among other potential issues, this leads to broken paths in `opencv_tests_config.hpp`, which can break the build or cause bizarre behaviour.
The fix for this issue, as stated in my links above, is to test that the variable hasn't been set yet with an `if(NOT DEFINED ...)` before attempting to set it. This is what I've implemented here. I admit I don't know enough about OpenCV's internals to know whether you *really* need to force the install to be in `/usr/local`, but as it stands right now you get *neither* a clean `/usr/local` path *nor* a customized `/my/desired/install/path`, but a broken mess. This change at least allows the user to customize their install directory.
In the meantime, there's a workaround for this, by explicitly defining the variable as a path with `-DCMAKE_INSTALL_PREFIX:PATH=my/desired/install/path`. But if this change can save anyone else the hours of headaches that I had today, I'll be happy.
* moved BLAS/LAPACK detection scripts from opencv_contrib/dnn to the main repository.
* trying to fix the bug with undefined symbols sgesdd_ and dgesdd_
* removed extra whitespaces; disabled LAPACK on IOS
Aravis several updates
* Fix adressing camera with id=0
* Aravis buffer property control & status added
* Modify of autoexposure algorith, ream frame ID from aravis + new properites
* Change of macro name
* VideoCapture now returns no frame on camera disconnecion
* Allow aravis-0.4 usage, proper camera object release.
- removed OpenCV_LIB_DIR* vars (they are broken and not required anymore)
- OpenCVConfig.cmake doesn't contain ANDROID/CUDA code if there is no such support
- removed OpenCV2_INCLUDE_DIRS_CONFIGCMAKE, merged into OpenCV_INCLUDE_DIRS_CONFIGCMAKE
- fix hard-coded relative paths for OpenCV_INSTALL_PATH
- removed OpenCV_TBB_ARCH
- switch OpenCVConfig.cmake into 2-level mode for Android SDK
PR #2968: cce2d998578f9c
Fixed bug which caused crash of GPU version of feature matcher in stitcher
The bug caused crash of GPU version of feature matcher in stitcher when
we use ORB features.
PR #3236: 5947519
Check sure that we're not already below required leaf false alarm rate before continuing to get negative samples.
PR #3190
fix blobdetector
PR #3562 (part): 82bd82e
TBB updated to 4.3u2. Fix for aarch64 support.
PR #3604 (part): 091c7a3
OpenGL interop sample reworked not ot use cvconfig.h
PR #3792: afdf319
Add -L for CUDA libs path to pkg-config
Add all dirs from CUDA_LIBS_PATH as -L linker options to
OPENCV_LINKER_LIBS. These will end up in opencv.pc.
PR #3893: 122b9f8
Turn ocv_convert_to_lib_name into a function
PR #5490: ec5244a
fixed memory leak in findHomography tests
PR #5491: 0d5b739
delete video readers
PR #5574
PR #5202
- added new functions from core module: split, merge, add, sub, mul, div, ...
- added function replacement mechanism
- added example of HAL replacement library
- IPP is disabled by default when compiler is mingw (couldn't make it
work)
- fixed some warnings
- fixed some `__GNUC__` version checks (for correctness and convenience)
- removed UTF-8 BOM from hough.cpp (fixes#5253)
This retains the desirable quality of not including paths to CUDA libraries
from the build system into the config files, and has two major advantages:
* It removes the need to use link_directories, which doesn't guarantee that
the libraries from the supplied directory will be used (there may be
libraries with the same names earlier in the search path).
* It removes the need to put -L entries into OPENCV_LINKER_LIBS. This variable
is used with target_link_libraries, where such entries are treated as linker
flags, so doing this is unportable. I remove the support for -L entries
from OpenCVGenPkgconfig.cmake, as well, to discourage adding them in the
future.
It's pretty much a simplified copy of the Linux script, lacking fancy colors.
Also, I had to drop Python testing, because it's not easy to pass the Python
module location to the script, and I have no pressing need to run the Python
tests at the moment.
(cherry picked from commit c1e3ca170e)
Conflicts:
CMakeLists.txt
2) disable IPP on 32-bit Linux when OpenCV is built as shared libs. Otherwise we get linker errors
3) disable IPP's minMaxIdx 32-bit floating-point flavor in a hope that it fixes some test failures
- Substituted HAVE_WINRT with WINRT
- Fixed compilation issues in ocl.cpp and parallel.cpp
- Fixed compiler issue for WP8: "C2678: binary '+' : no operator found which takes a left-hand - Fixed gitignore
- Added #ifdef HAVE_OPENCL to remove compiler warnings in ocl.cpp
- Used NO_GETENV similar to '3rdparty\libjpeg\jmemmgr.c;
- Added ole32.lib for core module (for WindowsStore 8.0 builds)
- Made OpenCV_ARCH aware of ARM
Signed-off-by: Maxim Kostin <v-maxkos@microsoft.com>
It's pretty much a simplified copy of the Linux script, lacking fancy colors.
Also, I had to drop Python testing, because it's not easy to pass the Python
module location to the script, and I have no pressing need to run the Python
tests at the moment.
IPP can be switched on and off on runtime;
Optional implementation collector was added (switched off by default in CMake). Gathers data of implementation used in functions and report this info through performance TS;
TS modifications for implementations control;
Fix indentation in output that made it look like changes were dependent on WinRT when they are independent libraries.
Defaults needed flipping otherwise undesired behavior. Change is tested with combinations.
Fixed and tested
Windows Phone v8.0/v8.1 SDK for Universal Windows Apps (Windows Phone v8.1 Silverlight App support not included) and fix initial cache causing problem
If both Python 2 and Python 3 are found, then build bindings for both of
them during the build process. Currently, one version of Python is
detected automatically, and building for the other requires changes the
CMake config.
The largest chunk of this change generalizes OpenCVDetectPython.cmake to
find both a Python 2 and Python 3 version of Python. Secondly, the
opencv_python module is split into two modules, opencv_python2 and
opencv_python3. Both are built from the same source. but for different
versions of Python.
When GTK3 is not found, HAVE_GTK was being set to TRUE. This edit
ensures that HAVE_GTK is only set if GTK3 or GTK2 (meeting minimum
version requirements) is present.
Selection logic for printing 'GTK: No' when the libraries are not found
has also been removed so the message is printed when GTK is not found or
selected in common with other libraries.
Changes committed:
modified: CMakeLists.txt
modified: cmake/OpenCVFindLibsGUI.cmake
During merging of conflicting versions of this file, I erroneously
deleted several lines in the GUI reporting section. This is repaired
in this commit.
specific code. As a result of this, HAVE_GTK3 no longer needs to be
exposed.
The use of HAVE_GTK, and HAVE_ GTK3 have been changed to mirror the
method used by HAVE_QT and HAVE_QT5.
On branch gtk3
Changes to be committed:
modified: CMakeLists.txt
modified: cmake/OpenCVFindLibsGUI.cmake
modified: cmake/templates/cvconfig.h.in
modified: modules/highgui/src/window.cpp
modified: modules/highgui/src/window_gtk.cpp
