Skip to content

Latest commit

 

History

History
245 lines (192 loc) · 9.66 KB

README.md

File metadata and controls

245 lines (192 loc) · 9.66 KB
Raw

Building NITRO

CMake is the preferred build method. Version 3.14 or better is required.

Sample Build Scenario

mkdir build
cd build
cmake ..
cmake --build . -j
cmake --build . --target install
ctest

Problems and Configurations

  • If your system compiler does not fully support C++11, you may have to specify a different one during the configure step. e.g.

    cmake -DCMAKE_C_COMPILER=/some/path/gcc/4.9.1/bin/gcc -DCMAKE_CXX_COMPILER=/...../bin/g++ ..

  • Pass -DCMAKE_INSTALL_PREFIX to set the install location.

  • Python and C++ bindings are built. Just make sure the relevant tools are on your PATH. Java and MATLAB bindings are now in the archive directory and no longer built.

  • See the coda-oss CMake build README for further build configuration information, particularly for Python-related details. The same options there may be passed to Nitro.

  • Build types Release, RelWithDebInfo, and Debug may be chosen

    • On Linux, debug symbols are available by default (RelWithDebInfo). Configure build type with -DCMAKE_BUILD_TYPE

    • On Windows, release type should be configured during the build and install steps

      cmake --build . --config Release -j
cmake --build . --config Release --target install

      The CMake default build type Debug may not work with Python, unless the Python installation includes debug versions of the Python libraries.

  • Regenerating python bindings

    • Currently, cmake configuration for regenerating swig bindings is incomplete and use of waf is required.

    • The .regenerate_python_bindings.py script is wrapper around waf can be run to quickly update these files.

      python3 .regenerate_python_bindings.py DEBUG_PY_BINDINGS=1 python3 .regenerate_python_bindings.py # see all details

    • This is typically required when project dependencies (e.g. CODA-OSS) are updated.

  • If the CMake build system does not support a required feature that Waf does, create an issue or a pull request!

Building with Waf

Waf is the legacy build system. Below are all of the options available.

> python waf --help
waf [command] [options]

Main commands (example: ./waf build -j4)
  build    : builds the project
  clean    : removes the build files
  configure: configures the project
  dist     : makes a tarball for redistributing the sources
  distcheck: checks if the sources compile (tarball from 'dist')
  install  : installs the build files
  uninstall: removes the installed files

Options:
  --version             show program's version number and exit
  -h, --help            show this help message and exit
  -j JOBS, --jobs=JOBS  amount of parallel jobs (8)
  -k, --keep            keep running happily on independent task groups
  -v, --verbose         verbosity level -v -vv or -vvv [default: 0]
  --nocache             ignore the WAFCACHE (if set)
  --zones=ZONES         debugging zones (task_gen, deps, tasks, etc)
  -p, --progress        -p: progress bar; -pp: ide output
  --targets=COMPILE_TARGETS
                        build given task generators, e.g. "target1,target2"
  --enable-warnings     Enable warnings
  --enable-debugging    Enable debugging
  --enable-64bit        Enable 64bit builds
  --enable-doxygen      Enable running doxygen
  --with-cflags=FLAGS   Set non-standard CFLAGS
  --with-cxxflags=FLAGS
                        Set non-standard CXXFLAGS (C++)
  --with-defs=DEFS      Use DEFS as macro definitions
  --with-optz=OPTZ      Specify the optimization level for optimized/release builds
  --libs-only           Only build the libs (skip building the tests, etc.)
  --shared              Build all libs as shared libs
  --disable-symlinks    Disable creating symlinks for libs
  --disable-java        Disable java (default)
  --with-java-home=JAVA_HOME
                        Specify the location of the java home
  --require-java        Require Java lib/headers (configure option)
  --nopyc               Do not install bytecode compiled .pyc files (configuration) [Default:install]
  --nopyo               Do not install optimised compiled .pyo files (configuration) [Default:install]
  --disable-python      Disable python
  --require-python      Require Python lib/headers (configure option)
  --enable-openjpeg     Enable openjpeg

  configuration options:
    -b BLDDIR, --blddir=BLDDIR
                        build dir for the project (configuration)
    -s SRCDIR, --srcdir=SRCDIR
                        src dir for the project (configuration)
    --prefix=PREFIX     installation prefix (configuration) [default: '/usr/local/']

  installation options:
    --destdir=DESTDIR   installation root [default: '']
    -f, --force         force file installation

  C Compiler Options:
    --check-c-compiler=CHECK_C_COMPILER
                        On this platform (linux) the following C-Compiler will be checked by default: "gcc icc suncc"

  C++ Compiler Options:
    --check-cxx-compiler=CHECK_CXX_COMPILER
                        On this platform (linux) the following C++ Compiler will be checked by default: "g++ icpc
                        sunc++"

Sample Build Scenario

> python waf configure --enable-debugging --prefix=installed
> python waf build
> python waf install

Enabling a debugger

-g and its variants can be achieved at configure time using the --enable-debugging switch at waf configure time

Memory Debugging

To ease debugging and memory leak detection, macros are used to malloc, realloc and free information. NITF_MALLOC(), NITF_REALLOC(), and NITF_FREE() should be used instead of the stdlib.h functions.

If you defined NITF_DEBUG during compilation (using configure, give the argument --with-defs="-DNITF_DEBUG" and this will occur automatically), you will get an memory image information dump every time you run an executable, named memory_trace.<pid> where <pid> is the PID of the process you just ran. There is a verification tool located in nitf/tests/verify, called mem_sane.pl. If you run mem_sane.pl with the memory trace as the single argument, you will get a formatted output of all memory that is questionably allocated or deallocated in the nitf library's calls. Please, please, please check your stuff.

C Conventions

In order to keep the C code easy to program and debug, and above all, OO, we stick to certain conventions herein:

  • All constructors must be passed an error On failure, they return NULL, and populate the error

  • All destructors must NULL set the object after they are done deleting, and should check the object prior, to make sure that it has not been deleted already. This means, of course, that all destructors take a pointer to the object. In practice, usually most of these, then, take double pointers (where usually you pass it a pointer by address)

  • All objects are in structs with an underscore in front of their name, and a typedef to the real name (.e.g., struct _nitf_PluginRegistry => nitf_PluginRegistry)

  • All functions that are non-static should be wrapped in a NITFAPI(return-val) or NITFPROT(return-val) for protected data.

  • This allows for easy macro definitions in order to control the decoration algorithm for windows, and to assure that the import decoration and export decoration are identical (otherwise we cant use them)

  • IMPORTANT: The difference between NITFAPI() and NITFPROT() is that the C++ code binding generator exposes API() calls and ignores PROT() calls.

  • All enumerations and constants have a NITF/NITF20/NITF21 prefix. Along these lines, all functions and objects are prefixed with a 'namespace' (nitf/nitf20/nitf21).

Platforms

While the ultimate goal is to be cross-platform and cross-language, the C and C++ layers get the most support.

The Python layer gets some use for scripting convenience.

The MATLAB and JAVA layers have not been touched in years; they are no longer built, code remains in the archive directory.

TREs need to be coded in C (only).

Before you commit

  • Create a unit test for your all code you are adding

  • Compile and test. (ctest)

  • A clang-format script is available at externals/coda-oss/.clang-format. Use it.

  • Doxygen on root directory and view in browser the doxygen code (in nitf/doc/html/).

Doxygen Commenting

Please make an effort to write doxygen comments. I know, especially in C, that doxygen has some issues. However, its the best, cheapest thing we have, and its important to have the APIs documented. It will save me the trouble of fixing it later, which will make me eternally grateful.

NITF Library Users: General Issues

NITRO handles TREs by loading dynamic libraries at runtime. Therefore, you need to make sure NITRO can find them.

  • If you are building from source, the location will be compiled in, and you don't have to do anything extra.

  • If you are working from a binary release, you will have to tell NITRO where the plugins are by setting the NITF_PLUGIN_PATH enviornment variable. This should look something like <install>/share/nitf/plugins.

  • If you wish to use a custom TRE location, you can also specify that with NITF_PLUGIN_PATH.

Contact

February 2022, Dan Smith Maxar