ICS7: Inconsistencies within spec

[josef-widder]

Surfaced from Informal Systems IBC Audit of cosmos-sdk hash cosmos/cosmos-sdk@90e9370

• headers contain height in unit64 while some lines above it is defined as (epoch,height) pair

• The note on "would-have-been-fooled logic appears out of context in the text. When restructuring the ICSs we should consider reorganizing the information.

• upgrades are Tendermint-specific and not mentioned in ICS002.

• Proposals are not discussed in the specification.

• genesis is not discussed in the specification.

• The function VerifyClientState does not appear in the spec.

• This error handling is not present in the specification.

• This error handling is not present in the specification.

• In all the Verify functions the specification checks latest height and frozen. These checks are done in the implementation in produceVerificationArgs. Perhaps the spec could mirror that modularity.

[cwgoes]

Generally I agree - however, I am not convinced that the IBC specification should include anything about governance, which in my view is an implementation detail of the Cosmos SDK (and the Cosmos Hub specifically). Perhaps there's some clean way we can abstract it or mention that a governance function could be implemented, but I do not think we should e.g. specify the logic of the proposal handler in the ICS specifications.

For genesis, I think we can include it, but as "initialisation logic" that needs to be run before any of the handlers or such.

[josef-widder]

For genesis, I think we can include it, but as "initialisation logic" that needs to be run before any of the handlers or such.

I think adding something on initialization logic would be great.