test: mutation proof for #3310's install-warn test (DO NOT MERGE) - #615
Closed
MichaelTaylor3d wants to merge 10 commits into
Closed
MichaelTaylor3d wants to merge 10 commits into
MichaelTaylor3d wants to merge 10 commits into
Conversation
…d dig-rewards-coin Money-bug containment for dig_ecosystem#3303/#3286: exact `=` pins on dig-node-service's direct edge stop the resolved chia-sdk-driver/chia-sdk-types/ chia-puzzle-types versions moving under us. This constrains OUR compile target only -- Cargo.lock still carries a second, transitive chia-sdk-driver 0.30.0 and chia-puzzle-types 0.26.0 line via other crates, and this pin does not evict those. Containment, not correction. chia-protocol/chia-bls/chia-sha2 stay caret: an `=` on a wire type poisons every crate that depends on chia-protocol directly. dig-rewards-coin = "0.5" (not the ticket's stale "0.4"): 0.5.0 is the release that already refuses `epoch_seconds == 0` inside `read_distributor` itself. Refs #3310 Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
…-rewards-coin Adds RealRewardsChainPort (crates/dig-node-service/src/rewards/): distributor_report served for real over dig-wallet's CorroboratedChainSource and dig_rewards_coin::state::read_distributor, via a guarded read (read_distributor_guarded) that refuses launch constants carrying epoch_seconds == 0 BEFORE calling into dig-rewards-coin at all -- defense in depth over that crate's own identical refusal (state.rs:1015), since chia-sdk-driver-0.36.0's commit_incentives backfill loop never terminates on that value and has no await point a timeout could interrupt. store_id/root are recovered from the launcher's creating spend's CREATE_COIN memo (chain_source.rs), following dig-mirror-coin's read_parent_outputs pattern: authenticate the puzzle reveal against the coin's puzzle hash before running it, never trust an unauthenticated memo alone. The other four RewardsChainPort methods (funded_distributors, distributor_state, submit_entry_writes, spend_new_epoch) answer Unavailable -- out of this ticket's scope (funder-registry and prover-cycle work tracked separately). Installed once from server.rs's enable_chain_sync-gated block, logging a false (already-installed) return at WARN. Refs #3310 Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
CI's `cargo fmt --all -- --check` and `--locked` builds both failed on the prior commit: rustfmt wanted several closures/return-types reformatted, and Cargo.lock was missing the `chia-sdk-test`, `clvm-traits` and `clvmr` entries the new adapter's Cargo.toml lines require, which `--locked` refuses to backfill. Refs #3310 Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
…t, fix constants fixture
- LaunchCommentError::ChainSource/Malformed tuple payloads were flagged
dead_code by clippy: a derived Debug impl does not count as reading a
private field. Add a manual Display impl that formats each variant's
payload, and switch chain_port.rs's build_report call site from {error:?}
to {error} so the reason reaches a log reader.
- adapter_source_never_imports_withdraw_committed_incentives was
self-defeating: it include_str!s its own file and the test's own name
and assertion messages contain the literal string it searches for, so
it could never pass. Scope the scan to production_region(), everything
before the file's own #[cfg(test)] marker.
- launcher_spend_with's RewardDistributorConstants fixture set
reserve_inner_puzzle_hash/reserve_full_puzzle_hash to Bytes32::default()
without calling .with_launcher_id(launcher_id), which recomputes both
fields from curried tree hashes. chia-sdk-driver's
RewardDistributor::from_launcher_solution requires constants ==
constants.with_launcher_id(launcher_id), so this fixture deterministically
failed to decode on every invocation (not flaky/timing-dependent).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
…ll-path evidence Drives RealRewardsChainPort::distributor_report end to end against a distributor launched by dig_rewards_coin::launch_dig_distributor in chia-sdk-test's peer simulator, over a MockChainSource loaded from the simulator's real coin records/spends. store_id/root are asserted against the values launched with -- recoverable only by actually running the launcher's parent spend and decoding its CLVM memo, so no fixture shortcut can pass this test. Also covers install_reward_chain_port's single-install refusal (true then false, with the WARN server.rs's own call site logs). Node exposes no lighter test constructor to an external integration test crate, so the install-path test uses Node::from_env(), the same constructor openrpc_drift_guard.rs's own test uses. launch_fixture returns Box<dyn std::error::Error> rather than pulling in anyhow for one test file. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
…identity verdict ChainSource::parent_spend returning Ok(None) is a chain-source gap (the source does not yet hold the launcher's creating spend), not a genuine "this is not a DIG distributor" classification. Give it its own LaunchCommentError::ParentSpendUnavailable variant so chain_port.rs can map it to ChainPortError::Unavailable, agreeing with the other absence path (read_distributor_guarded's own Ok(None)), instead of rendering a transport lag as a definitive negative identity claim. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
- Add one bounded tracing::warn! on distributor_report's error path, firing once per failure->success transition (not per call, via an AtomicBool), so an installed-but-degraded chain source is no longer indistinguishable from "no port installed." - Extract current_distributor_epoch's arithmetic into a pure epoch_ordinal fn and unit-test it directly (non-zero multi-epoch case, the saturating_sub clock-skew branch, and the bare-launch zero case) -- the only non-trivial computed field in the report mapping, previously unexercised by anything but a zero-valued default. - Add launch_comment_error_to_port_error, wiring chain_source.rs's new ParentSpendUnavailable variant to ChainPortError::Unavailable, with a regression test, and a companion test confirming GuardedReadError::NonTerminatingEpochSeconds still maps to the named refusal (Other), never to Unavailable. - Add a unit test proving a failing chain source surfaces as a named ChainPortError::Unavailable, never Ok(_) with a default-valued report. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
install_reward_chain_port_refuses_a_second_install_with_a_warn previously captured and asserted against a warn it emitted itself inside the test's own closure -- deleting server.rs's real warn line would have left it green. Replace the self-emission with a source-text check against server.rs's own production region (the same shape adapter_source_never_imports_withdraw_committed_incentives already uses), so the assertion can only be satisfied by what server.rs actually ships. Also softens the A3 module doc's overstated "no fixture value can produce the right store_id/root" claim: true of this file today, not a structural guarantee. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
…e text not runtime behaviour Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Deliberately deletes the production warn in
server.rs's install-refusal branch to proveinstall_reward_chain_port_refuses_a_second_install_with_a_warn(PR #614) can actually fail. ExpectTest + coverageRED. THROWAWAY -- close and delete after the red is observed. Refs #3310.