Deployment environment variable map

[johnhoran]

I'd suggest that the environmental variables should be defined as a map instead of in a list. e.g. instead of

resource "astro_deployment" "hybrid" {
  environment_variables = [{
    key       = "key1"
    value     = "value1"
    is_secret = true
  }]
}

it would be

resource "astro_deployment" "hybrid" {
  environment_variables = {
    key1 = {
      value     = "value1"
      is_secret = true
    }
  }
}

There are two benefits to this, firstly keys need to be unique anyway, so defining them as a map instead of a list implicitly enforces this in terraform. However the primary reason is because we would then be able to utilize terraforms lifecycle.ignore_changes meta argument to ignore changes in certain attributes. e.g.

resource "astro_deployment" "hybrid" {
  environment_variables = {
    key1 = {
      value     = "value1"
      is_secret = true
    }
  }
  lifecycle {
    ignore_changes = [
      environment_variables["key2"]
    ]
  }
}

This is desirable for us because we have some environmental variables managed by our CICD pipeline, and others would be more suited to terraform. Terraform does support ignore_changes applied to array elements, however relying on the order of the variables feels fragile.

[vandyliu]

Hi @johnhoran

Thanks for the request.
This is actually quite a smart change and I think would benefit a lot of other people as well.
Not sure of the timeline of when this can be added but definitely will be!