I personally prefer my dockerfiles that are meant to run a single binary to have that binary as the entrypoint. This way I just have to pass the flags directly and don't need to know how the binary was named inside the image.
Compare with this change
docker run eigenda-proxy:latest --memstore.enabled
to what we currently have
docker run eigenda-proxy:latest ./eigenda-proxy --memstore.enabled
This is wordy for no reason, but also I need to inspect Dockerfile manually to check where the binary is located.
This is a breaking change however, given that RaaS's or whoever has this setup in a docker-compose file somewhere will need to change their CMD to work with this (just remove the name of the binary)
I personally prefer my dockerfiles that are meant to run a single binary to have that binary as the entrypoint. This way I just have to pass the flags directly and don't need to know how the binary was named inside the image.
Compare with this change
to what we currently have
This is wordy for no reason, but also I need to inspect Dockerfile manually to check where the binary is located.
This is a breaking change however, given that RaaS's or whoever has this setup in a docker-compose file somewhere will need to change their CMD to work with this (just remove the name of the binary)