I'm using the same Docker base image, "python:3.10.10-slim-bullseye@.sha256:...", in several different

BUILD

files. Is there a way to define this base image in only one place and reuse it across all my

BUILD

files? I've considered using environmental variables, but I'm not sure if that's the best approach. Does anyone have any suggestions on how to achieve this?

So is this inside the dockerfile that you are using it or for a docker related target?

Docker related target.

For any target e.g.

docker_environment

, you can use

___defaults___

https://www.pantsbuild.org/docs/targets#field-default-values

I see. That way I can change the default values. But, I have figured out how to define a global env variable in pants.toml; how can I access it from BUILD files?

Can you give an example of what and how your target looks like where you are using this base image repeatedly?

I have already created a macro that generates the docker images (for aws lambda). It looks like this:

def lambda_docker(name, handler, repository, registries, image_tags, **kwargs):
    entry_point, handler_fun = handler.split(":")
    entry_app, _ = entry_point.split(".")
    entry_dir = build_file_dir().parts[-1]

    pex_binary(
        name="bin-srcs",
        entry_point=entry_point,
        layout="packed",
        include_requirements=False,
        include_tools=True,
    )

    pex_binary(
        name="bin-deps",
        entry_point=entry_point,
        dependencies=["3rdparty/python:poetry#awslambdaric"],
        layout="packed",
        include_sources=False,
        include_tools=True,
    )

    base_image = "python:3.10.10-slim-bullseye@sha256:fcf375288c9348c9708cc7ea3d511b512224219fdc164b6960b3ce85288e1cbf"

    docker_image(
        name=name,
        repository=repository,
        dependencies=[":bin-srcs", ":bin-deps"],
        registries=registries,
        image_tags=image_tags,
        instructions=[
            f"FROM {base_image} as deps",
            f"COPY {'.'.join(build_file_dir().parts) + '/' + 'bin-deps.pex'} /bin-deps.pex",
            "RUN PEX_TOOLS=1 /usr/local/bin/python3.10 /bin-deps.pex venv --scope=deps --compile /app",
            f"FROM {base_image} as srcs",
            f"COPY {'.'.join(build_file_dir().parts) + '/' + 'bin-srcs.pex'} /bin-srcs.pex",
            "RUN PEX_TOOLS=1 /usr/local/bin/python3.10 /bin-srcs.pex venv --scope=srcs --compile /app",
            f"FROM {base_image}",
            "COPY --from=deps /app /app",
            "COPY --from=srcs /app /app",
            "WORKDIR /app",
            'ENTRYPOINT [ "/app/bin/python", "-m", "awslambdaric" ]',
            f'CMD ["{entry_dir}.{entry_app}.{handler_fun}"]',
        ],
        source=None,
    )

So I suppose you want to use it inside the instructions?

Sorry for the delay; had some bitbucket issues. Yes, and also in the build environment definition:

+docker_environment(
+    name="lambda_docker_env",
+    platform="linux_x86_64",
+    image="python:3.10.10-slim-bullseye@sha256:fcf375288c9348c9708cc7ea3d511b512224219fdc164b6960b3ce85288e1cbf",
 )

If it was just docker_image then you can use this https://www.pantsbuild.org/docs/docker#build-arguments

[docker]
build_args = [
  "BASE_IMAGE=value1",
]

# Dockerfile
ARG BASE_IMAGE
FROM $BASE_IMAGE

But that doesn’t help with

docker_environment

It's not well documented, but both the

./pants

script and scie-pants (

pants

) support sourcing a

.pants.bootstrap

file and gathering exported env vars from there. See https://github.com/pantsbuild/pants/discussions/17633 If using scie-pants (

pants

) the more efficient and standard

.env

files are supported as well. See: https://github.com/pantsbuild/scie-pants#features

Yes; I have considered creating a custom local image; that seems like the best option at the moment.

Yes. Although I'm not sure if

env("X", "optional default")

is supported in 2.15.0 or just 2.16.x.

I am already using 2.16.0.dev7

Ok, you should be able to use that inline for target field values.

You could also define this as a constant “macro”: https://www.pantsbuild.org/docs/macros