Naming convention for configurations (env, toml, cli args and more)

[Briancbn]

A continuation of #17 (comment)

MQTTSettings decoupled from TaskOrchestratorSettings as it is now read separately within client::clients::mqtt itself instead of from main.

But for the environment variables.

MQTT_CLIENT__* more like a global configurations shared across multiple services that uses MQTT, like AWS_SECRET_ACCESS_KEY. I don't think this is the intention here.

Whereas it is more fitting to be within the context of "Task Orchestrator".

We can also do something like

TO__SERVER__HTTP_HOST  # Server host (http)
TO__SERVER__HTTP_PORT  # Server port (http)
...

TO__MQTT_CLIENT__HOST
TO__MQTT_CLIENT__PORT

TO__GRPC_CLIENT__HOST
TO__GRPC_CLIENT__PORT

Let me know your thoughts @EthanKuai @DillonChew98

[EthanKuai]

Good point. Just researched; should be an easy change.

@DillonChew98 I will remove TaskOrchestratorSettings struct. For this reason will open a new PR.

[DillonChew98]

Yes, this was the idea behind having the extra nesting of TaskOrchestratorSettings. Since we are decoupling MQTT and maybe eventually the rest of the clients, we could do away with:

struct Settings {
  task_orchestrator: TaskOrchestratorSettings {
    http: HttpSettings,
    amqp: AmqpSettings,
}

and have it be flattened to:

struct Settings {
  http: HttpSettings,
  amqp: AmqpSettings,
}

then add a TO prefix to the load_base_configuration config builder instead:

builder = builder.add_source(config::Environment::with_prefix("TO").prefix_separator("__").separator("__");

[EthanKuai]

I chose prefix "RMF2_TO", or should I stick with "TO"?

[EthanKuai]

FYI: Within #18 config.rs changes, it is terribly simple to update TOML section headers to prefix with RMF2_TO., let me know if that is preferred.

[DillonChew98]

FYI: Within #18 config.rs changes, it is terribly simple to update TOML section headers to prefix with RMF2_TO., let me know if that is preferred.

Don't think this is required.

I personally am fine with it just being TO, @Briancbn wdyt?

[EthanKuai]

Perhaps should be consistent with other components?

[amqp.consumer]
queue = "@RECEIVE@-rmf_schedule"

Here uses rmf_schedule, so perhaps RMF_TO?

[Briancbn]

Here uses rmf_schedule, so perhaps RMF_TO?

This is legacy code haha, please ignore. I think RMF2_TO better.

The rest of the RMF2 services can follow the same convention, rmf2_vda5050_master -> RMF2_VM.
@sauk2 any thoughts?

[sauk2]

Here uses rmf_schedule, so perhaps RMF_TO?

This is legacy code haha, please ignore. I think RMF2_TO better.

The rest of the RMF2 services can follow the same convention, rmf2_vda5050_master -> RMF2_VM. @sauk2 any thoughts?

Looks good to me!