EDIT: this issue may be wrong: we may want to use collection_project rather than collection_env.
The CLI and Worker both currently generate configuration keys to support collections.
We need to add a new config key: collection_env. This will be used in the collections adaptor to request the collection for that specific environment.
In the CLI, we can accept env as an argument or load it from the project.yaml. I am comfortable only enabling this for v2 (it only matters on execute)
How does the worker know the env used in that run? It will either need to be passed as an option on the plan, or included in the claim payload. I think an option on the plan feels better. So load plan.options.env from the incoming Run object from Lightning, and feed that through to the credential.
EDIT: this issue may be wrong: we may want to use
collection_projectrather thancollection_env.The CLI and Worker both currently generate configuration keys to support collections.
We need to add a new config key:
collection_env. This will be used in the collections adaptor to request the collection for that specific environment.In the CLI, we can accept env as an argument or load it from the project.yaml. I am comfortable only enabling this for v2 (it only matters on execute)
How does the worker know the env used in that run? It will either need to be passed as an option on the plan, or included in the claim payload. I think an option on the plan feels better. So load
plan.options.envfrom the incoming Run object from Lightning, and feed that through to the credential.