Topcoder - Marathon Match New Scoring System - Final PoC

For this contest, we are looking at trying some new technologies to assembly the new scoring system for Marathon Match challenges.

For that we would like to experiment the CLING tool, a c++ language interpreter.

Prerequisite

Build tools and OS

node 8
node-gyp
cmake 3.2+
g++ 7+
Python 2.7
Operation system: MacOS, Linux(recommend to use Ubuntu 18.04)
docker and docker-compose (optional if you want to test with docker image)

Require xcode under macOS or proper C/C++ compiler toolchain, like GCC under linux.

Please ensure node-gyp tool is installed successfully to build Node.js native addon later.

C++ libraries

Install dependencies

Please follow steps of cling build page to build and install cling manually. You may also use package tools such as Homebrew and apt-get instead of building it yourself.

Here's a brew under macOS example:

brew install cling

Ensure that cling was properly installed by typing cling command:

$ cling

****************** CLING ******************
* Type C++ code and press enter to run it *
*             Type .q to exit             *
*******************************************
[cling] $

Please follow steps of nlohmann json github page to build and install nlohmann json library manually. But, once again, I would suggest you to use the package tool.

Here's a brew under macOS example:

brew tap nlohmann/json
brew install nlohmann_json

Install dependencies notes for Ubuntu

You can download latest valid package match your version of Ubuntu from cling download page. If you are using Ubuntu 18.04 you can download ling_2018-10-27_sources.tar.bz2.

Unzip the source code to a folder, like /usr/local/cling/0.5_1.
Install dependent libraries sudo apt install libtinfo-dev libz-dev to make the make pass.

You will need the following actions after building the code:

Make a link to cling's include directory (for example /usr/local/cling/0.5_1/include) into /usr/local to make the interpreter work correctly.
Update library path. Run: echo "absolute path of /usr/local/cling/0.5_1/lib" > /etc/ld.so.conf.d/cling.conf && sudo ldconfig and use ldconfig -p to make sure node addons could find libcling.so successfully.

Build node addons code

Edit env.sh to configure environment variables for cling and nlohmann/json correctly.

Cling_LIB_\DIR is the directory that contains lib folder of cling tool. Cling_DIR is the directory that contains file ClingConfig.cmake. LLVM_INSTALL_PREFIX is the directory that contains c++-analyzer, ccc-analyzer file and lib folder of cling tool. NLOHMANN_JSON_INCLUDE_DIR is the c++ header folder that contains nlohmann/json.hpp.

Now run following commands to build the source code(it will compile addons codes during npm install process):

source env.sh
npm install
npm run lint # lint with airbnb code style

You can run npm run rebuild to rebuild after install process is over.

If there is no errors you will find node addons MarathonMatchScoring.node would be generated in folder build\Release. You can quickly test by run bash demo.sh 1 and you can check successful result for file test_files\case1_result.txt

If you can run successfully you can then run npm test in same console or make sure related environment are configured rightly.

Build demo code

npm run rebuild-demo

The target executable MarathonMatchScoring would be generated in folder demo.

Running notes

You must preload libcling to resolve run errors like undefined symbol: _ZN4llvm11raw_ostream6handleEv.

If you still meet errors, you can frist try to use absolute path for LD_PRELOAD under linux or DYLD_INSERT_LIBRARIES under osx. For example, export LD_PRELOAD=$CLING_LIB_DIR/libcling.so, or you may use export LD_LIBRARY_PATH=$CLING_LIB_DIR(may also use sudo ldconfig later) and then LD_PRELOAD=libcling.so <application> under linux. Or you may use similar feature DYLD_LIBRARY_PATH(export LD_LIBRARY_PATH=$CLING_LIB_DIR) & DYLD_INSERT_LIBRARIES (DYLD_INSERT_LIBRARIES=libcling.dylib <application>) under osx with SIP is turned off.

Please notes DYLD_INSERT_LIBRARIES will not extend by subprocess so you cannot run mocha directly under OSX and you have to call node node_modules/.bin/mocha.

Please notes some recent OSX with SIP enabled may not work for DYLD_INSERT_LIBRARIES doesn't work in recent versions of Mac OS.

Possible fix from Library not loaded is sudo ln -s /Applications/Xcode.app/Contents/Frameworks/libcling.dylib $CLING_LIB_DIR/libcling.dylib Last option for recent OSX need to copy libcling.dylib to running directory with cp $CLING_LIB_DIR/libcling.dylib .·(recommend actually to avoid many issues for osx).

I test codes on OSX 10.14 and Ubuntu 18.04 and in both of them everthing worked fine.

Docker notes

To save your time I provide docker image you can build with docker-compose build and then run docker-compose run mm You have to run commands npm install && npm run lint && npm test && npm run rebuild-demo again one by one after run docker-compose run mm command.

You can then follow Validation.md to verify codes build in Ubuntu 16.04 docker image.

cccvvv2012 / cpp-mm-scoring