Dynamic Caching using AWS's ECR and the pulumi_docker_build package

[tnorlund]

Hi! I have a few python packages that I use as "base images" that I reference in Lambda functions. I tried asking this in slack, but I thought I'd try here too.

• How can I optimize cache_from/cache_to with ECR when using multi-stage builds?
• Any best practices for sharing base images across different Lambdas?
• How do you handle the image URLs vs static strings in the build_args for dev/prod environments.
• Any way to conditionally skip builds if the image already exists in ECR with the same content hash?