Terraform Vars Cascade Processing

[Tensho]

Does it make sense to process terraform.tfvars files in all parent directories? Consider the next example:

prod
  us-east-1
    prod
      mysql
        terraform.tfvars # 1 (particular module level variables)
  terraform.tfvars # 2 (region level variable)
terraform.tfvars # 3 (account level variables)

Currently, there is only find_in_parent_folders(), which picks the first found terraform.tfvars file in the parent directories. In the example above we can merge only 1 and 2. It would be nice to adjust the function or create a new one (to not break backward compatibility), which would merge 1, 2, and 3.

People achieve described behavior with the hooks, but I think that should be in the "core".

Here is a comment from other thread describing a similar thing.

[brikis98]

Supporting multiple includes was intentionally omitted to avoid over-complicating reasoning about the code, how configs are merged, etc. If we can find a simple way to do it in the future, it may be worth doing it. #707 adds support for it, but #466 needs to be resolved before any new such features.

[jpreese]

FWIW I am looking for this functionality as well. Only allowing a single terragrunt.hcl can be very limiting. The scope of what you want to include in the inputs block is going to increase as you traverse deeper into the folder structure (global variables -> region variables -> environment variables -> module variables).

This sort of approach doesn't really seem possible at the moment.

[Tensho]

@brikis98 Is it still something that correlates with the future Terragrunt plans?