docker-compose.yml

version: '3.2'

# Docker Compose file for coordinating several dvmdostem containers.
#
# First you should build your images. This can be done individually or you can 
# use the docker-build-wrapper.sh script.
#
# Next create a .env file where you set the following variables:
#   $ echo "DDT_INPUT_CATALOG=/some/path/to/your/input-catalog" >> .env
#   $ echo "DDT_WORKFLOWS=/some/path/to/your/workflows" >> .env
#   $ echo "V_TAG=$(git describe)" >> .env
#
# Next start the containers:
#   $ docker compose up -d to start the containers
#
# Then you can run various programs inside the containers, see examples below.

services:
  # Some general dev tools - not expected to be used directly
  cpp-dev:
    build: .
    image: cpp-dev:${V_TAG}

  # The general dvmdostem dev enviorment. Intented to be used for 
  # most development work by mounting volumes of source code into 
  # the resulting container - essentially sharing the source code
  # folders from the host to the container so they can be edited 
  # on the host and used w/in the container run-time.
  dvmdostem-dev:
    image: dvmdostem-dev:${V_TAG}
    build: .
    tty: true
    ports:
      - "7001:7001"
    volumes:
       - sourcecode:/work
       - inputcatalog:/data/input-catalog
       - workflows:/data/workflows

  dvmdostem-autocal:
    image: dvmdostem-autocal:${V_TAG}
    build: .
    tty: true
    ports:
      - "7009:7009"
    volumes:
      - sourcecode:/work
      - inputcatalog:/data/input-catalog
      - workflows:/data/workflows


  # This is more of a stand-alone image, just meant to serve as a
  # compile time environment for creating the lean run image.
  # Copies source code into the image so that it doesn't need
  # volumes to run container and run make. Then once the dvmdostem
  # binary exists in this image, it can be copied into the lean 
  # run image.
  dvmdostem-build:
    image: dvmdostem-build:${V_TAG}
    build: .
    #command: tail -F /dev/null # keeps container running

  # The lean production image
  dvmdostem-run:
    image: dvmdostem-run:${V_TAG}
    build: .
    tty: true
    ports:
      - "7002:7002"
    volumes:
      - inputcatalog:/data/input-catalog
      - workflows:/data/workflows
      
  # Auxillary stuff that can't run on other container/images due to difficult
  # build dependencies (GDAL mainly).
  dvmdostem-mapping-support:
    image: dvmdostem-mapping-support:${V_TAG}
    build: .
    tty: true
    ports: 
      - "7003:7003"
    volumes:
      - sourcecode:/work
      - inputcatalog:/data/input-catalog
      - workflows:/data/workflows


volumes:
  sourcecode:
    driver_opts:
      type: none
      device: "${PWD}"
      o: bind
  inputcatalog:
    driver_opts:
      type: none
      device: '${DDT_INPUT_CATALOG}'
      o: bind
  workflows:
    driver_opts:
      type: none
      device: '${DDT_WORKFLOWS}'
      o: bind

## EXAMPLES

# In general there are 2 ways you can operate:
# 1) run a specific command in the docker container
#
# 2) start a bash terminal in the container and work with an interactive prompt 
#    in the container

# Runs the model in the service. Service is not a
# constant running process, so container starts, runs, 
# and exits. Doesn't handle Ctrl-C
#   $ docker-compose run dvmdostem-run dvmdostem --help

# Runs a basic ls in the service
#   $ docker-compose run dvmdostem-run ls /data/input-catalog

# Run in existing container (using exec as opposed to run does this)
#   $ docker-compose exec dvmdostem-run dvmdostem --help

# Several ways to compile
# 1) The parallel options seems very finnicky...
#    Mac Docker desktop seems to freeze a lot
#    $ docker-compose exec dvmdostem-run make -j2
#
# 2) Sign into interactive session and run make
#    $ docker-compose exec dvmdostem-run bash    
#    $ develop@256519d33e7c:/work$ make

# Run the io-view tool:
#   $ docker compose exec dvmdostem-mapping-support bokeh serve scripts/io_view.py