FRAME's misleading Validator/Collator terminology

[bernardoaraujor]

FRAME's pallets are being used for relay-chain as well as parachain implementations.

In many instances, the word Validator is used to name variables, traits, enums, etc. For the cases where the pallet is only expected to be used in a relay-chain context, that is usually fine (e.g.: staking and nomination-pools pallets).

However, there's many cases where the FRAME pallets are also used in a parachain context. The use of Validator terminology is potentially misleading there, because it's not obvious that Validator terms should actually be interpreted as Collator under those circumstances.

That can be even more misleading in the case of system-level parachains (cc @joepetrowski). A naive reader could maybe interpret those variables as something related to the relay-chain validators, or (even worse) simply equate Validators to Collators as if their concepts were actually identical.

I've actually witnessed the second scenario in our ecosystem recently (not really system parachain-related though). While it's not not our job to hold the hand of every naive developer, I still think there's room for us to decrease the potential for cognitive dissonances for developers diving into our codebase with fresh eyes.

Here's a list of FRAME pallets where I was able to detect the term Validator being used interchangeably with Collator (can be interpreted both ways, depending on context):

• session
• babe
• aura
• executive
• im-online

On the Cumulus side, collator-selection pallet is using Validator where Collator is the only possible interpretation. The only reason for that seems to be the fact that it imports frame_support::traits::ValidatorRegistration and all the other variables follow that as a convention.

A potential solution would be to rename FRAME's support::traits::validation module to something like support::traits::author, as well as all its Validator* variables (traits, etc) to Author*. That would impose a more generic naming convention on all the pallets listed above, removing the potential for contextual misinterpretations.

Cumulus companion: #73