app: 💸 a mock consensus spend is performed #3891
Conversation
this makes a small (no-op) change to
`fmd_precision_within_grace_period`.
this will provide an event that can be used to identify why a clue did
not satisfy the consensus rules. using tracing's pretty formatter:
```
2024-02-26T21:26:49.825156Z ERROR penumbra_app::action_handler::transaction::stateful: invalid clue precision, clue_precision: 1, using_current_precision: false, using_previous_precision: false, within_grace_pe
riod: true
at crates/core/app/src/action_handler/transaction/stateful.rs:62
in penumbra_app::action_handler::transaction::stateful::fmd_precision_within_grace_period with current_fmd.precision_bits: 0, previous_fmd.precision_bits: 0, previous_fmd.as_of_block_height: 1
in penumbra_app::action_handler::transaction::check_stateful
in penumbra_mock_consensus::abci::deliver_tx
in penumbra_mock_consensus::block::execute with height: 1, time: 1708982809
```
cratelyn
force-pushed
the
kate/mock-consensus-test-sends-valid-binding-sig
branch
from
af63e0b to
0269234
Compare
cratelyn
changed the title
| // Build a transaction spending this note. | ||
| let tx: Transaction = { |
There was a problem hiding this comment.
a passing thought, but this part of the API is pretty long-winded. we talked a bit about this in our sprint planning meeting today, but i'd love to see a more convenient way to go about this for future test authors, without introducing a third planner API.
that is a design conversation that shouldn't block this PR, though.
There was a problem hiding this comment.
The API we'd like to use is the Planner in the view crate. We'd need to be able to pass the MockClient to its plan() method: https://rustdoc.penumbra.zone/main/penumbra_view/struct.Planner.html#method.plan
There was a problem hiding this comment.
x-ref #3896, i think?
There was a problem hiding this comment.
Clarification post 54c0b35 , where the transaction plan creation becomes much simpler: my point is that for simple tests, where we can write down the complete transaction plan explicitly, we should do so. In this case, we can, which is what 54c0b35 does.
If we need to do more complex test logic, test logic that is complex enough that we cannot simply write down the plan, then we should decide on a way to use the Planner API to plan transactions, rather than have another set of helper methods for generating plans. This may involve pulling in the full rust view server implementation, or continuing to use the MockClient. But for simple cases it's important to be able to test the chain logic without all the complexity of the rust view server.
This commit moves some common functionality into the MockClient itself, in order to simplify the layout of the test logic. Also, we add explicit invariant checks on the chain state.
hdevalence
temporarily deployed
to
smoke-test
GitHub Actions
Inactive
cratelyn
deleted the
kate/mock-consensus-test-sends-valid-binding-sig
branch
fixes #3788.
this branch builds upon the work in #3857 and #3866, filling out the remaining holes in the transaction plan. this now performs a spend, using the mock consensus engine.