Skip to main content

Setup

We provide a docker container capable of building and running our measurement campaign. The container also functions as a Remote Desktop Server (RDP) with preconfigured development envioronments to inspet and edit the source code.

Install docker​

Build docker image​

In this repository, the devenv folder contains scripts and source files to recreate the development and measurement environment.

Dockerfile for the environment
FROM --platform=x86-64 linuxserver/rdesktop:ubuntu-xfce

ENV PUID=1000 
ENV PGID=1000
ENV TZ=Etc/UTC 

EXPOSE 3389

# Initialize misc
USER root
RUN apt update
RUN apt install git wget zip nano -y
RUN apt install openjdk-21-jdk -y
RUN apt install psmisc -y
RUN apt install webext-ublock-origin-firefox -y
RUN apt install build-essential -y


# Install ProbLog
USER root
RUN apt install -y python3
RUN apt install -y python3-pip
RUN pip install --break-system-packages problog
RUN pip install --break-system-packages PySDD

# Install eclipse
USER root
RUN wget 'https://www.eclipse.org/downloads/download.php?file=/technology/epp/downloads/release/2024-03/R/eclipse-modeling-2024-03-R-linux-gtk-x86_64.tar.gz' -O eclipse.tar.gz
RUN tar -xzf eclipse.tar.gz -C /opt
RUN rm eclipse.tar.gz
RUN echo '[Desktop Entry]\n\
Version=13.0\n\
Type=Application\n\
Terminal=false\n\
Icon=/opt/eclipse/icon.xpm\n\
Name=Modeling 2024-03\n\
Exec=/opt/eclipse/eclipse\n\
Name=Modeling 2024-03\n\
Categories=Development\n\' > /usr/share/applications/eclipse.desktop

# Install VIATRA and dependencies
USER root
RUN /opt/eclipse/eclipse -application org.eclipse.equinox.p2.director -repository https://download.eclipse.org/releases/2024-03 -installIU org.eclipse.collections.feature.source.feature.group/11.1.0.v20220705-1455,org.eclipse.xtend.sdk.feature.group/2.34.0.v20240227-0940,org.eclipse.xtext.redist.feature.group/0.0.0
RUN /opt/eclipse/eclipse -application org.eclipse.equinox.p2.director -repository http://download.eclipse.org/viatra/updates/release/2.8.0 -installIU org.eclipse.viatra.query.sdk.feature.source.feature.group/2.8.0.202310201322,org.eclipse.viatra.query.sdk.feature.feature.group/2.8.0.202310201322

# Install python packages for diagrams
RUN pip install --break-system-packages notebook
RUN pip install --break-system-packages pandas==1.5.3 seaborn==0.12.2 matplotlib==3.7.1 numpy==1.26.4 itables

# Install storm 
USER root
WORKDIR /opt
RUN apt install build-essential git cmake libboost-all-dev libcln-dev libgmp-dev libginac-dev automake libglpk-dev libhwloc-dev libz3-dev libxerces-c-dev libeigen3-dev -y
RUN wget https://github.com/moves-rwth/storm/archive/stable.zip
RUN unzip stable.zip
WORKDIR storm-stable
RUN mkdir build 
WORKDIR build
RUN cmake ..
RUN make -j12
ENV PATH=$PATH:/opt/storm-stable/build/bin
WORKDIR /opt
RUN rm stable.zip

RUN rm -rf /config/.cache

Before cuntinuing, ensure that the current working driectory is devenv od the repository. (ProbabilisticGraphQuery/devenv)

Build your own docker image with the following command:

docker build -t pgq-devenv .
Note
  • Downloading eclipse takes a lot of time. Be patient, or download the eclipse instance manually and replace RUN wget 'https://www.eclipse.org/downloads ... with COPY <your file> eclipse.tar.gz
  • Building storm model checker can take up to an hour. It also takes significant memory to compile. To lower resource usage, adjust the make -j12 flag in the dockerfile to use less CPU cores.

Alternatively, import the image using the following command:

docker image load -i <my image file>

Create container​

Create container with the following command.

docker run -d -u root ^
  -v \\wsl$\Ubuntu\tmp\.X11-unix:/tmp/.X11-unix ^
  -v \\wsl$\Ubuntu\mnt\wslg:/mnt/wslg ^
  --name=pgq-devenv-storm ^
  --security-opt seccomp=unconfined ^
  -p 4100:3389 ^
  --shm-size="1gb" ^
  -v %~dp0\..:/config/ProbabilisticGraphQuery ^
  pgq-devenv

(Delayed) You can stop the container with:

docker container stop pgq-devenv-storm

(Delayed) You can restart the container with:

docker container start pgq-devenv-storm
Note

If you use docker exec, don't forget to set the user to abc.

docker exec -u abc -it -w /config pgq-devenv-storm /bin/bash

Configuring container​

Download compile runtime dependencies​

docker exec -u abc -w /config/ProbabilisticGraphQuery/sources/hu.bme.mit.delta.api/libs/ pgq-devenv-storm ./download.sh
docker exec -u abc -w /config/ProbabilisticGraphQuery/sources/hu.bme.mit.inf.measurement.utilities/libs/ pgq-devenv-storm ./download.sh

Configure editor​

  • Connect to the running container from an RDP client (address: localhost:4100, username: abc, password: abc)
  • Launch Eclipse from Applications -> Development -> Modeling 2024-03
  • Pick a workplace. (Any directory works, but we recommend picking /config/ProbabilisticGraphQuery/devenv/workplace)
  • Import all projects from /config/ProbabilisticGraphQuery/sources (File -> Import -> General -> Existing Projects into Workspace -> Next -> Select root directory -> /config/ProbabilisticGraphQuery/sources -> Projects -> Select All -> Finish)
  • Change missing API baseline preferences for org.eclipse.viatra.query.runtime.rete to Warning or Ignore. (Window -> Preferences -> Plug-in Development -> API Baselines -> Missing API baseline -> Ignore)
  • Import launch configurations from /config/ProbabilisticGraphQuery/devenv/launch_configs (File -> Import -> Run/Debug -> Launch Configurations -> Next -> From Directory -> /config/ProbabilisticGraphQuery/devenv/launch_configs -> tick launch_configs -> Finish)

Using the container​

  • Export runnable jar for measurements: File -> Export -> Java -> Runnable JAR file -> Next -> Launch configuration (select any, e.g., SAT - Scale - hu.bme.mit.inf.measurement) -> Export destination: /config/ProbabilisticGraphQuery/measurements/pgq.jar -> select Extract required libraries into generated JAR -> Finish -> (Ignore warnings)