- Generated on Wed Oct 26 2022 14:15:33 for CAPD DynSys Library by
1.8.17
|
CAPD DynSys Library
5.2.0
|
For CAPD users:
libwxgtk2.8-dev package).libX11-dev package).libgmp, libgmp-dev, libgmpxx, libmpfr, libmpfr-dev.liblog4cxx10-dev).libboost1.53-all-dev or libboost-all-dev).To use CAPD developer version one require also
The CAPD package was tested on Linux distributions: Debian, Ubuntu. Please check Dockerfiles to see our setup.
1) Install C++ compiler from Command Line Tools (it is also shipped with XCode) by doing one of the following:
2) If you have an M1 chip (filib interval library does not support it), then do one of the following
Use Rosetta (it is an Intel processor emulation and can slow down performance)
You can choose to have two versions of Terminal. One running with Rosetta and the other to run without it. To do so simply duplicate the Terminal application, and choose one of the terminal application to be under Rosetta and the other without it. Use the Rosetta terminal for the below installation steps.
–without-filib switch when calling configure script. It will use internal CAPD interval package instead of filib library. It is not yet well tested. It is good for development.If you have an Intel mac, ignore this step.
3) Install Homebrew
For OSX you can find a few package managers, we recommend Homebrew.
4) Install GNU tools and libraries, that CAPD depends on.
Optional step - graphical library X11 support:
version="2.7.11"
curl -L -o /tmp/XQuartz-${version}.dmg "https://dl.bintray.com/xquartz/downloads/XQuartz-${version}.dmg"
hdiutil attach /tmp/XQuartz-${version}.dmg
sudo installer -verbose -pkg /Volumes/XQuartz-${version}/XQuartz.pkg -target /
ln -fs /opt/X11/include/X11 /usr/local/include/X11
ln -fs /opt/X11/lib/libX11.dylib /usr/local/lib/libX11.dylib
ln -fs /opt/X11/lib/pkgconfig/{x11,xproto,kbproto,xcb,pthread-stubs,xau,xdmcp}.pc /usr/local/lib/pkgconfig/
Verify previous steps with pkg-config:
5) Optional/Experimental Recently OSX version of gcc/g++ switched to clang engine. We have not tested it for compatibility with our code. You can compile the code, but we are not sure about compatibility with roundings etc. You can experiment with original version of gcc/g++ which you can install using following commands:
The current version of CAPD uses automated compilation tools therefore it needs linux like shell console and selection of other tools. We recommend use of MinGW - Minimalist GNU for Windows.
We do not recommend Windows platform for development: shell scripts execution in MSYS is very slow(!) because of fork() emulation.
Nowadays we use cross-compilation technique to build binaries for Windows. We use MXE with Docker, see our Dockerfiles . However, long time ago we were able to prepare MinGW enviroment on Windows for CAPD compilation using the following:
NOT supported!!!
1) Download and install mingw-get-setup.exe. Set location or use default C:\MinGW. Select 'Basic Setup' on the left panel. Then mark: mingw-32-base, msys-base. You may see a few errors related to dependencies - just ignore them.
2) Click 'Start' and to 'Run' prompt past (change C:\MinGW if you use different settings):
Answer two times y.
MinGW is installed! To verify: open
and run
Please make sure that
returns something like:
if not, then close MSYS shell and execute once again:
3) Now we need to install a few packages. In the MSYS shell (msys.bat) execute following commands:
The CAPD compilation should be performed from msys shell.
(Optional)Please download Boost library from here. ATTENTION For Windows download ZIP archive. On the Boost webpage you can find details about the installation process. You need to do it from native windows command line shell (Start -> Run -> cmd.exe). For us enough is to execute:
You may see a lot of warnings, do not care about them - just keep waiting and be patient. UNLESS you see an error - then let us know.
Next, form the MSys shell (replace [path_to_boost] with MSys path to the Windows folder with boost, in our case /c/boost_1_55_0/ ):
1.8.17