Code coverage audit, rework to obtain coverage w/o large E2E tests

[cwgoes]

A lot of our coverage is provided by two large E2E tests (MASP & ledger transactions), which should be mostly replaced with / supplanted by unit / prop tests with proper mocks so that we can reach sufficient coverage.

[tzemanovic]

to my knowledge make test-unit-coverage skips e2e

[cwgoes]

Hmm - in that case @batconjurer or @mariari can you clarify the concern here?

[tzemanovic]

I agree that MASP should have unit / integration / prop tests. There's nothing to catch issues in it besides e2e tests and the lack of these tests is being a serious issue for the more complex integrations

[brentstone]

@tzemanovic @batconjurer @Fraccaman how relevant is this issue still?

[Fraccaman]

I think coverage is still low especially for MASP.

[tzemanovic]

I went through coverage report from #3704: https://app.codecov.io/gh/anoma/namada/pull/3704/tree/crates

In our crates, the least coverage comes from sdk (14466 missed lines), node (6593 lines), ibc (2186), apps_lib (2011). I think we can improve the coverage a lot in these if we include integration tests, which should be possible (WIP in #3704).

Specific section that need attention:

• account
  • auth & storage mods
  • priority: medium
• ibc
  • lib - a lot of the ibc functionality is only tested in e2e
  • priority: medium
• node
  • shell/governance.rs - should also be able to move this into governance crate now
  • priority: medium
• sdk
  • masp - can we also move it (with shielded-sync sub-mod) into shielded_token?
  • priority: high
• shielded_token
  • vp & validation mods
  • priority: high
• token - done in Add test coverage on token transfer struct #3709
  • Transfer methods
  • priority: high
• tx
  • a lot of functionality inside the types mod is not covered
  • it also looks like the types mod contains too much uncategorized logic that should be refactored
  • priority: high