neurorecon

Repository for building OpenRecon containers.

Local setup on macOS

Install and start Docker Desktop. If Python 3 or 7-Zip is not already available, install them with Homebrew:

brew install python p7zip

From the repository root, create and activate the local Python environment:

python3 -m venv .venv
source .venv/bin/activate
python -m pip install jsonschema packaging

Confirm that Python and pip both resolve inside this repository's .venv:

command -v python3
python -m pip --version

Both paths should contain /neurorecon/.venv/. Activation applies only to the current terminal, so activate the environment again after opening a new one.

Choose one recipe and run the build from its directory. For example:

cd recipes/qsmxt

# Prefer the local Docker cache, then fall back to the remote image.
/bin/bash ../build.sh

# Or require an already-cached local Docker image.
/bin/bash ../build.sh --local-cache

build.sh refuses to install Python packages unless a virtual environment is active. Missing jsonschema and packaging dependencies are installed into that environment using the same Python interpreter that runs the build.

When using --local-cache for an offline build, the script now prompts for which artifact(s) to create: OpenRecon, FIRE, or both.

Each recipe build now emits two distributable artifacts by default:

• OpenRecon_<vendor>_<name>_V<version>.zip
• FIRE_<vendor>_<name>_V<version>/

Optional FIRE overrides can be exported from a recipe params.sh when needed:

• fireFreeSpaceMb
• fireStartupCommand
• fireSearchString
• fireHostname
• firePort
• fireBundleName

Experimental raw acquisition return

../build.sh --experimental-raw-return permits research labels with both emitter=raw and injector=raw. The option extends validation for that build without changing the stock schema. It does not supply a native ICE adapter or make raw acquisition reinsertion compatible with the stock scanner injector. Ordinary builds continue to reject raw-return labels.