Networks naming scheme

[cor]

We'd like to within our Nix configurations unambiguously refer to specific network setups with a consistent naming scheme.

Parameters

The way I see it, we have the following parameters defining a network:

Network        = union | ethereum | osmosis | ...
Target         = mainnet | testnet | devnet
ValidatorCount = 4 | 8 | 16 | 32 | 64 | ...
Environment    = local | online
Sync           = genesis | fork

Full Naming Scheme

Networks can be referred to using the following name:

networks.Network-Target-ValidatorCount-Environment-Sync

For example:

networks.union-devnet-4-local-genesis

"A Union Devnet with 4 validators running in a Local environment starting from Genesis"

Or

networks.ethereum-mainnet-64-online-fork

"Ethereum Mainnet with 64 validators running in an Online environment starting from a Fork"

Shorthand Aliases

Since there are specific setups that are most commonly used, and because not all combinations make sense, we have the following shorthand aliases:

union-devnet = networks.union-devnet-4-local-genesis
ethereum-devnet = networks.ethereum-devnet-8-local-genesis

union-testnet = networks.union-testnet-16-online-genesis

(More will follow)

[PoisonPhang]

Overall, this makes sense to me. However, I'm struggling to identify the need for Environment.

Won't Mainnet and Testnet always be online while Devnet will always be local?

[benluelo]

Do we have any usecases for locally running the union testnet? And would running it locally not make it a devnet?

[benluelo]

Also, what would running a "forked state" of the testnet look like?

[PoisonPhang]

I still have some work to do to determine how we can fork main/test net and run it locally, but yes. That use case makes sense.

A valid argument exists for that being a devnet once forked, but I'm fine calling it a local testnet or mainnet.

Thanks for the clarification sir.

[benluelo]

Is an "online" devnet possible?

[cor]

Is an "online" devnet possible?

This is one of the combinations that does not make sense, as mentioned in the aliases section: "and because not all combinations make sense"

[benluelo]

Any reason for Train-Case as opposed to kebab-case?

[cor]

Ah yeah meant to do kebab case but accidentally inserted the case used for the types. Will update

[cor]

Fixed

[benluelo]

I would recommend either moving the validation count to the end or pre/post-fixing it with something (i.e. 64-val), since it could be confused for the revision number on cosmos chains

[cor]

Good point, I agree that we should avoid this confusion

[benluelo]

How does the "fork" Sync work?

[cor]

These renames are pending @PoisonPhang 's attempt to replace the validator set after statesyncing

[mkduggirala]

wondering about the necessity of the 'Environment' parameter. akso implementing validation checks to prevent unsupported parameter combinations

[Manikantad1310]

Nice project and well effort of deva, i appreciate the hard work