Arthur runs on Docker. To build it, run the following:

docker build -t jaycode/arthur .

Arthur application is comprised of three servers:

• The main app and workspace app, both are located in the same docker as below:

  docker run -dit -v /c/path_to_project/:/media/disk -p 8000:8000 -p 8888:8888 -p 6006:6006 -p 49152:49152 --name arthur jaycode/arthur
  

  In this case /c/path_to_project/ is the directory that contains '/Arthur/', the main directory of this application. It is important to follow this similar structure (i.e. app's root dir's parent virtually linked with container) to ensure external libraries are running without issues.

• A Redis server for session sharing between main and workspace apps. Run the following docker command to set it up:

  To display errors from running that code:

  docker logs $ID
  

  To store session in persistent data, run this instead (not needed but good for debugging session values):

  ID=$(docker run --name arthur-redis -d -v /c/path_to_project/Arthur/redis_data:/data -v /c/path_to_project/Arthur/dev/redis.conf:/usr/local/etc/redis/redis.conf -p 6379:6379 redis redis-server /usr/local/etc/redis/redis.conf --appendonly yes)
  

  session:eyJhY3RpdmVfcHJvamVjdCI6InJpc2t5In0.CT36Hg.vX7ffotqpm1y6bMXgN3K3YisTiQ

Port information:

• 8000: Arthur main app.
• 8888: iPython notebook.
• 6006: Theano's Tensorboard.
• 49152: Arthur workspace app.
• 6379: Redis (used in second docker above)

To access a docker container, run the following:

docker exec -ti arthur bash

Before actually running the server, make sure mongodb server is up and running. In localhost this is done by simply running command mongod.

And finally, run the server by going to /media/disk/Arthur and running:

python server.py

And create another instance to run:

python server-workspace.py

Grunt is used to allow compass to update scss as well as uglifying javascript files. First cd to root directory (the same directory that contains this readme file, then install grunt with npm install grunt. Other npm modules will then be asked to be installed (such as grunt-sass and grunt-contrib-watch).

When all npm modules have been installed, grunt watcher can be run by command grunt inside root directory which will detect updates to scss and javascript files then compile them into css file readily used in app.

All scss and javascript files are editable from directory dev/front_end.

When adding new modules, vendor all the pip *.tar.gz into vendor/ with following command:

pip install --download vendor -r requirements.txt

ipython notebook documents in /test are there to systematically test and document your tests.

To test out library code at its lowest level, run the code directly e.g.

python libs\arthur\document.py > result.txt

result.txt will then contain output of your test.

Current recommendation is to run database server outside of docker. To have MongoDB database server up and running, do the following for Windows dev machine (assuming docker vm is running):

1. Open "Network Connections".
2. Find "VirtualBox Host-Only Network Status".
3. Click "Details..." and take note on IPv4 Address.
4. Enter this ip address as mongodb config (e.g. D:\mongodb\config.txt file), under the setting net.bindIp. Enter this next to local IP e.g. "127.0.0.1,192.168.56.1".
5. From inside docker, it should now be possible to connect with command mongo 192.168.56.1.

Run the following script for installation (make sure to turn the database server on first).

python dev/install.py

Run pylint [path] > pylint.txt from root directory to check code cleanliness at given path.

Arthur uses Sphinx to write documentation. All documentation files are located in docs/ directory. Sphinx will automatically searches for docstrings within the code and compile them into documentation. To compile the files, run make html from within docs/ directory.