Is there a way to have a shell_command that can tell whether it is being invoked as a dependency of a publish or package goal? I'm working on making a macro that creates AWS ECR via shell_command as suggested here: https://pantsbuild.slack.com/archives/C046T6T9U/p1660713334302559

that information is not exposed to the command, but it is available in rule code, so a plugin is able to know this, and as such, it is possible to add this to the

shell_command

target as env var or something like that. It would imply though that the command will have to re-run for each goal as the cache key would then depend on the goal being run.

Hmm, so the issue I was trying to resolve was that I want to be able to build images without needing to be logged into AWS. I did a hack that just adds "; exit 0" to the end so that it always succeeds, but it sounds like that would run into the same issue. Is there a way to specify that I never want to cache a target?

I don’t think we have this per target (you can always ask pants to run without cache

--no-cache

), but for targets that support using env vars, I can think of work arounds to make them uncachable by providing the current time as the env var value for example..

This is essentially just a one-off command, right? Why not just make the repository manually the first time or using the terraform `aws_ecr_repository` resource?

not a docker expert, but I think I want a repository per docker image

not a docker expert, but I think I want a repository per docker image

I believe that is correct. The ECR terminology is confusing for no apparent reason.