One can interact with this system via docker-compose
. If you want to download prebuilt images (totaling around 1.6GB in size) of these services, issue docker-compose up --no-build
at the root of this repository. If you prefer to build the images locally, issue docker-compose up --build
. Either way, eventually you'll have two services available locally: donu
on port 8000, the webserver exposing our functionality via HTTP request-response workflows, and a Jupyter Lab instance on port 8888 that interprets exposure
, our measure and experiment language. See the docker-compose
logs for the link (containing a secret key) to access this Jupyter instance via a web browser.
Working with AlgebraicJulia
(for model stratification or PNC gromet simulation) on a clean copy of this repository, without docker-compose
, requires first running ./scripts/run-algebraic-julia.sh --build
. The building process may take a while.
Some of the models you might expect donu
to serve, if you're using it locally, are symlinked to from a submodule (demoRepo
). The classic git submodule update --init
should populate it.