Resolve cyclic module dependencies

[kocubinski]

Relates to Phase 4 of #11899.

#13012 (comment) identifies all cyclic dependency relationships between /x/* modules in the SDK. These relationships must be acyclic for the completion of Phase 4 in the above epic.

• /x/auth <-> /x/authz
• /x/auth <-> /x/feegrant (@atheeshp )
• /x/auth <-> /x/staking - x/auth should not depend on x/staking #16467
• /x/auth <-> /x/simulation
• /x/auth <-> /x/bank - x/auth should not depend on x/bank #16464
• /x/auth <-> /x/gov
• /x/authz <-> /x/staking - x/authz design should not require a cyclic dependency to function #16463
• /x/authz <-> /x/bank - x/authz design should not require a cyclic dependency to function #16463
• /x/authz <-> /x/gov
• /x/bank/ <-> /x/staking/ - x/bank should not depend on x/gov, x/staking, and x/distribution #16577
• /x/bank/ <-> /x/distribution/ - x/bank should not depend on x/gov, x/staking, and x/distribution #16577
• /x/bank/ <-> /x/gov/ - x/bank should not depend on x/gov, x/staking, and x/distribution #16577
• /x/distribution <-> /x/gov - x/distribution should not depend on x/gov #16469
• /x/distribution <-> /x/staking - x/staking should not depend on x/distribution #16471
• /x/gov <-> /x/group - (note: there is a false positive here from x/gov/README.md)
• /x/gov <-> /x/params - since x/params is deprecated this can be ignored.
• /x/gov <-> /x/mint - x/gov and x/mint should not depend on each other #16579
• /x/gov <-> /x/staking - x/staking should not depend on x/gov #16472
• /x/simulation <-> /x/staking - ignoring for now pending completion of simulation ADR

Tasks

Preview Give feedback

1. x/authz design should not require a cyclic dependency to function #16463
   T:Sprint +1
   
2. x/auth should not depend on x/bank #16464
   T:Sprint +1
   
3. x/auth should not depend on x/staking #16467
   T:Sprint +1
   
4. x/distribution should not depend on x/gov #16469
   T:Sprint +1
   
5. x/staking should not depend on x/distribution #16471
   T:Sprint +1
   
6. x/staking should not depend on x/gov #16472
   T:Sprint +1
   
7. x/bank should not depend on x/gov, x/staking, and x/distribution #16577
   T:Sprint +1
   
8. x/gov and x/mint should not depend on each other #16579
   T:Sprint +1
   

[tac0turtle]

It seems that moving some interfaces to the core module or to types would resolve many of these along side the test changes. Would we prefer they move to types or core?

[aaronc]

It seems that moving some interfaces to the core module or to types would resolve many of these along side the test changes. Would we prefer they move to types or core?

[aaronc]

I think any module related interfaces belong in a types or api package rather than core. In my mind, core should include only include framework code that has no state machine logic. Since we already have the api module, it might make sense to consider how that effort fits into all of this.