QT Toolkit not available in versions 4-9

[ELC]

I want to setup a dockerfile using one of these official octave images as base, but I need the QT Toolkit enable.

I ran the following command with the latest version of each major from 4 to 9 and in every case the only toolkit available is gnuplot.

Is there a way to enable the QT toolkit? I understand that the qt libraries are already installed as part of the octave's dockerfile and I'm careful of using octave command line and not octave-cli.

Command to reproduce:
docker run --rm -it gnuoctave/octave:{VERSION} octave --eval "available_graphics_toolkits()"

The underlying hardware is the 2-core 8GB RAM 32GB Disk Github Codespace machine

Result:

ans =
{
  [1,1] = gnuplot
}

Versions tested:

• 4.4.1
• 5.2.0
• 6.4.0
• 7.3.0
• 8.4.0
• 9.4.0

[siko1056]

Do you still see only gnuplot, if you extend the docker command with the flags given in https://github.com/gnu-octave/docker?tab=readme-ov-file#starting-the-octave-gui

docker run \
  --rm \
  --network=host \
  --env="DISPLAY" \
  --env="HOME=$HOME" \
  --env="XDG_RUNTIME_DIR=$XDG_RUNTIME_DIR" \
  --user $(id -u):$(id -g) \
  --volume="$HOME:$HOME:rw" \
  --volume="/dev:/dev:rw" \
  --volume="/run/user:/run/user:rw" \
  --workdir="$HOME" \
  docker.io/gnuoctave/octave:9.4.0 octave --gui

[ELC]

I was able to identify the issue, I'm working on having a minimal working solution. The problem was not with the image itself but when running the image in a headless environment like MyBinder.org

I'll post the updates here.

My working solution is in this branch:
https://github.com/ELC/IPBookPlatform-environment/tree/octave

[ELC]

This is the minimal version of the dockerfile that works with Jupyter's Octave kernel and QT Toolkit:

FROM gnuoctave/octave:9.4.0

RUN apt-get update && \
    apt-get install -y git libsqlite3-dev libffi-dev xvfb \
                           octave-control octave-image octave-io octave-optim \
                           octave-signal octave-statistics \
                           libgl1-mesa-dri libglu1-mesa mesa-utils

COPY --from=ghcr.io/astral-sh/uv:0.6.3 /uv /uvx /bin/
RUN --mount=type=cache,target=/root/.cache/uv \
    --mount=type=bind,source=uv.lock,target=uv.lock \
    --mount=type=bind,source=pyproject.toml,target=pyproject.toml \
    uv sync --frozen --no-install-project --compile-bytecode && \
    ln $(find .venv/bin -type f) /usr/bin/ && \
    ln /usr/bin/jupyter-lab /usr/bin/jupyterlab

WORKDIR "/home/ubuntu"
COPY --chown=ubuntu:ubuntu . .
USER ubuntu

ENV DISPLAY=":1"
ENV OCTAVE_EXECUTABLE="/usr/bin/octave"

ENTRYPOINT ["bash", "-c", "set -e; Xvfb $DISPLAY -screen 0 1024x768x16 & sleep 2 && exec \"$@\" 2>&1 | tee .jupyter-server-log.txt", "--"]

# docker build -t my_image . && docker run --rm -it my_image octave --eval "available_graphics_toolkits()"

Considering my work was Python specific, I believe the only relevant points for the GNU Octave repo is that in order to support QT Toolkit I needed to:

• Install libgl1-mesa-dri, libglu1-mesa and mesa-utils
• Set up and run Xvfb in the entrypoint with a sleep to wait until it finishes startup

I think it may be helpful for others to add this information somewhere in the README. There are no more outstanding items from my side though and we can close the issue if no further action will be taken.

[siko1056]

Thank you @ELC , now I understand what you are trying to do and thank you for digging out the details for this use case.

I added a section to the README with the bare minimum needed to run this container image in an headless environment (e.g. Binder) 73334a9

https://github.com/gnu-octave/docker?tab=readme-ov-file#use-qt-graphics_toolkit-in-headless-environments