Publish is not executed before deploy is executed

[Nishikoh]

Describe the bug Publish is not executed before deploy is executed. I thought that the docker image would be published if the terraform code contained docker dependencies when performing experimental-deploy. However, contrary to my expectations, the publish was not executed.

Pants version 2.22.0

OS Ubuntu 22.04

Additional info

#pants.toml
[GLOBAL]
pants_version = "2.22.0"
backend_packages = [
  "pants.backend.build_files.fmt.ruff",
  "pants.backend.docker",
  "pants.backend.experimental.terraform",
  "pants.backend.python",
]

[python]
interpreter_constraints = ["==3.11.*"]

# main.tf
terraform {
  backend "local" {
    path = "/tmp/tfstate/baz.tfstate"
  }
}

provider "random" {
  version = "3.6.2"
}

resource "random_id" "id" {
  byte_length = 7
}

# BUILD
terraform_module(name="main")
terraform_deployment(name="deploy", root_module=":main", dependencies=[":app-docker"])

docker_image(
  name="app-docker",
  registries=[
        "ghcr.io",
  ],
  repository="<MY USER NAME>/pants-deploy-example",
  image_tags=["latest"],
  instructions=[
    "FROM alpine",
    "RUN echo 'hello world' > /hello.txt",
  ]
)

# log
$ pants experimental-deploy --experimental-deploy-publish-dependencies :deploy 
22:16:13.67 [INFO] Deploying targets...

Terraform used the selected providers to generate the following execution plan. Resource actions are indicated with the following symbols:
  + create

Terraform will perform the following actions:

  # random_id.id will be created
  + resource "random_id" "id" {
      + b64_std     = (known after apply)
      + b64_url     = (known after apply)
      + byte_length = 7
      + dec         = (known after apply)
      + hex         = (known after apply)
      + id          = (known after apply)
    }

Plan: 1 to add, 0 to change, 0 to destroy.
╷
│ Warning: Version constraints inside provider configuration blocks are deprecated
│ 
│   on main.tf line 14, in provider "random":
│   14:   version = "3.6.2"
│ 
│ Terraform 0.13 and earlier allowed provider version constraints inside the provider configuration block, but that is now deprecated and will be removed
│ in a future version of Terraform. To silence this warning, move the provider version constraint into the required_providers block.
╵

Do you want to perform these actions?
  Terraform will perform the actions described above.
  Only 'yes' will be accepted to approve.

  Enter a value: yes

random_id.id: Creating...
random_id.id: Creation complete after 0s [id=dzju5Fqrzw]

Apply complete! Resources: 1 added, 0 changed, 0 destroyed.

✓ infra/terraform/app/baz:deploy deployed

The publish log does not appear to be flowing.

https://github.com/grihabor/pants/blob/dd11c333e71976bb135f1b75b3abd26e27959bd2/src/python/pants/core/goals/deploy.py#L214

[Nishikoh]

When pants publish :app-dcocker is run by itself, it completes without problems.

$ pants publish :app-docker
22:24:31.91 [INFO] Completed: Building docker image ghcr.io/<MY USER NAME>/pants-deploy-example:latest
22:24:31.91 [INFO] Packaged app-docker.docker-info.json
The push refers to repository [ghcr.io/<MY USER NAME>/pants-deploy-example]
e97805dba242: Layer already exists 
63ca1fbb43ae: Layer already exists 
latest: digest: sha256:000271f8030b1409fe654d06416c5a67f506238fe9fc398611df7e8c704938b1 size: 734

✓ ghcr.io/<MY USER NAME>/pants-deploy-example:latest published.

[lilatomic]

Unfortunately, items in the dependencies field are not published prior to experimental-deploy. I agree that it would be good to be able to publish some dependencies before deploying. I thought there was another issue for this, but I can't find it ATM.

The Pants engine actually supports this, although the Helm backend is really the only one to use it (and only then for Docker containers) ref. I think that extending the Terraform backend to support this (or implementing it in Pants core) would be a straightforward change.