Add Dockerfile

Dockerfile

@@ -0,0 +1,61 @@
+# Building this image will include the running of all benchmarks in a "cold
+# cache" state, and thus will take quite a bit of time (roughly 30-60 minutes,
+# with Ganache excluded).  The cache will be retained in the built image, such
+# that running the image will execute benchmarks with a "warm cache". For
+# example:
+#
+#   docker --build-arg ALCHEMY_API_KEY=... -t sim .
+#   docker run sim benchmark-hardhat
+#
+# BEWARE: This image should not be published because it will contain traces of
+# your Alchemy API key. (See https://stackoverflow.com/a/40762010/406289)
+# Future work can take advantage of Docker's support for secrets, but
+# unless/until there's a need to publish this image, that would be overkill.
+
+FROM debian
+
+ARG ALCHEMY_API_KEY
+
+RUN if [ -z "${ALCHEMY_API_KEY}" ]; then \
+      echo build argument ALCHEMY_API_KEY required; \
+      exit 1; \
+    fi && \
+    apt-get update && \
+    apt-get install --yes curl git make xz-utils yarnpkg && \
+    ln -s /usr/bin/yarnpkg /usr/bin/yarn
+
+# nix/dapptools don't like to install/run as root
+RUN useradd user && \
+    mkdir -p -m 0755 /home/user && \
+    chown user /home/user && \
+    mkdir -m 0755 /nix && \
+    chown user /nix
+WORKDIR /home/user
+USER user
+# nix/dapptools install scripts require this environment:
+ENV USER=user
+ENV BASH_ENV=/home/user/.profile
+ENV SHELL=/bin/bash
+SHELL ["/bin/bash", "-c"]
+
+RUN git clone https://github.com/mds1/convex-shutdown-simulation.git
+WORKDIR convex-shutdown-simulation
+
+RUN yarn && \
+    cp .env.example .env && \
+    sed -i "s!yourAlchemyApiKey!${ALCHEMY_API_KEY}!" .env && \
+    curl -L https://raw.githubusercontent.com/gakonst/foundry/master/foundryup/install | bash
+# Don't combine these sequential RUN commands because we need fresh shell
+# instances in order for each install script to take effect.
+RUN foundryup && \
+    curl -L https://nixos.org/nix/install | bash
+RUN curl -L https://dapp.tools/install | bash
+RUN dapp update
+
+# TODO: Benchmark ganache too, once it's fixed.
+# (See https://github.com/mds1/convex-shutdown-simulation/pull/4)
+RUN make benchmark-foundry && \
+    make benchmark-hardhat && \
+    make benchmark-dapptools
+
+ENTRYPOINT ["make"]

README.md

@@ -19,3 +19,4 @@ uses about 16M gas and performs a number of token transfers
 ## Tips
 
 - Set `export CLEAR_CACHE=1` in your `.env` file to clear the Ganache and Hardhat caches
+- Consider running the benchmarks via Docker. See the comment header in [the Dockerfile](./Dockerfile) for details.