Repository navigation
Conversation
hansieodendaal
left a comment
There was a problem hiding this comment.
Looking good.
Some comments to consider please.
Yes, we need the parent PRs to be merged first, then we can assess here. Although smaller PR's are easier to reason about, the size of your current PRs is not an issue atm. Where size do come in is that you should still categorize your scenarios into smaller feature and CI tag chunks. That being said, you still need a LEZ domain expert to look at your scenarios; my biggest contribution is approach, organization and Cucumber-edness. |
- prove non-inclusion from the post-admission tip; guard P-24 against instruction-layout drift - pin the supply account by fixture id; honor caller sequencer configs instead of clobbering priority_fee - derive wait timeouts from block cadence; generic pre-state account counts in P-20 - deduplicate wait loops, stake key derivation, and submit_stake steps; parallelize account reads
Add a Background step "chain waits give up after N blocks" that replaces the hidden WAIT_TIMEOUT_BLOCKS constant, and make the unchanged-accounts assertion name the accounts it compares, e.g. "the config, funding and ownership accounts are unchanged", instead of the fixed snapshot set
Use priority_fee_percent in the shared base config, read the minimum stake from the config's channel params, encode the raw Stake mirror with borsh and mover_account_id, and pass a program account id to the wallet's public send.
…iming Pass the stake funds account in every Stake pre-state list and assert the staked balance on it; claim the ownership account for P-13 through the token program, since a plain transfer no longer claims its recipient; donate from a genesis supply account in P-23, as plain transfers are fee-charged, and pin that the donation lands and stays outside the stake; shift the P-20 outline to 2 and 5 accounts around the new canonical four.
Split sequencer_registration.feature into three feature files, each a coherent group with its own CI tag, and add one workflow step per group to cucumber-integration-tests.yml following the run-cucumber-suite pattern from #779 and #793: - sequencer_registration.feature, @stake_registration_ci: the amount a first Stake moves (P-01, P-02, P-03, P-25, P-23) - stake_account_validation.feature, @stake_accounts_ci: the pre-state account list, its signers, owners and count (P-04, P-13, P-14, P-20) - stake_instruction_validation.feature, @stake_instruction_ci: the instruction bytes and call path (P-18, P-24) Feature names drop the em-dash tagline so the artifact_subdir path the suite action uploads on failure is plain ASCII. The non-inclusion protocol rationale stays in sequencer_registration.feature; the other two files point at it.
e5ccfbe to
ca3cd8f
Compare
- right behind the transaction under test and judges non-inclusion against it.
- stake_chain_caller: a new guest that forwards an instruction into another program as a chained call. Validated by P-17 (a Stake chained through it is rejected by sequencer_stake's caller-is-none guard); also unblocks P-19, U-13, F-10. The top-level transaction is fee-charged, so the rejection shows as inclusion with reverted effects, paid by a genesis supply account. - second native mover: the existing simple_balance_transfer. Validated by P-21 (Stake accepted through a non-authenticated_transfer mover). No claim on the funding account is needed: a debit is gated on the account's signature, which Stake passes through, not on its owner. - both test programs are deployed at runtime through program_loader and addressed by the header account the deployment claims; the scenario wallet gains a deploy request and a payer-aware program transaction.
cb6154b to
f7e1f09
Compare
Cucumber's run already parses --tags (with full tag expressions) and --name, which is what CI and local runs use. The env-var filter duplicated that with a weaker one and was ignored whenever --tags was passed.
Merge dev at 7e3f7f7 (CI: cache the resolved Bedrock node by its pinned revision, and pass one env pair per line to the integration-test prebuild step). The merge was conflict-free and touches only the CI workflow and the new resolve-bedrock-node action; the branch's sequencer registration prerequisites are kept as is.
RUSTSEC-2026-0385 (ctutils), -0386 (tokio AsyncFd I/O safety) and -0387 (crypto-bigint constant-time branches) were published after the last lockfile sync and fail `cargo deny check` on an unchanged lockfile. Bump tokio to 1.54.1, crypto-bigint 0.7 to 0.7.9 and ctutils to 0.4.6, which carry the fixes. crypto-bigint 0.5.5, reached through elliptic-curve 0.13 under k256 0.13, which the workspace pins directly and libp2p-identity also uses, has no patched 0.5 release; the fix only exists in the 0.7 line, which requires migrating to elliptic-curve/k256 0.14. RUSTSEC-2026-0387 is ignored in .deny.toml with that reason.
Merge dev at 0e5a0d9, which reports advisories instead of gating on them, checks them nightly, bumps tokio to 1.53.3 and ignores the ctutils and crypto-bigint 0.7.5 advisories rather than bumping them. This branch already bumps tokio to 1.54.1, crypto-bigint 0.7 to 0.7.9 and ctutils to 0.4.6, so the lockfile keeps those versions and the .deny.toml keeps only the RUSTSEC-2026-0387 ignore for the unpatched crypto-bigint 0.5 line. dev's ignores for RUSTSEC-2026-0385 and the yanked crypto-bigint@0.7.5 are dropped because neither crate version is in this lockfile anymore. dev's switch of unused-ignored-advisory to warn and its CI workflow changes are taken as is.
Merge dev at 8aea254, which archives the integration tests without debug info in CI. The merge is conflict-free: dev only touched the GitHub workflow files, and the branch's cucumber sequencer registration tests and dependency bumps are untouched.
Merge dev at fe93529: a failed channel read is now an error item rather than the channel's end in bedrock, the indexer and the zone indexer; the indexer's status is dumped on a catch-up timeout; bedrock logs the size of every published block; chain_state gains zone indexer tests. The merge had no conflicts; the branch's dependency bumps in Cargo.lock are preserved alongside dev's new chain_state async-trait entry.
Merge dev at c830502, which moves the cucumber CI run onto a build job plus a per-suite matrix. No conflicts: dev touched only the workflow and composite action, and the matrix tags match the suites this branch already splits the stake scenarios into.
Merge dev at e0da95a into the sequencer registration prerequisites branch. The only conflict was the import block of the wallet test app, where dev's sequencer-wallet FFI integration added the SequencerClient import next to this branch's MAX_SEGMENT_DATA_LEN import; both are kept.
Merge dev at 68ad874 into the sequencer registration prerequisites branch. Dev brings atomic deposits in the sequencer, the channel_config actor renamed to channel_signing, the core FFI HTTP client and the wallet storage u128 nonce fix. No conflicts: dev did not touch the cucumber stake scenarios, the wallet test app or the advisory pins this branch carries, and the only shared file, Cargo.lock, merged cleanly with the branch's tokio, crypto-bigint and ctutils bumps intact.
Merge dev at 1806e87 into the sequencer registration prerequisites branch. The merge applied cleanly: dev's fee-aware genesis funding and private transfer fee changes do not touch the stake Cucumber steps, the wallet app's runtime program deployment, or the advisory ignores this branch carries. Cargo.lock keeps the branch's tokio, crypto-bigint and ctutils bumps alongside dev's new dependencies.
The 2026-10-26 bump of crypto-bigint, ctutils and tokio moved the guest image ids, so the committed artifacts no longer matched and the artifacts check has failed on every run since. dev clears the same advisories with ignore entries and keeps the old versions, so take dev's lockfile and ignore list as they are.
Merge dev at 952bd6d into test/sequencer-registration-prerequisites. Dev brings the CI changes that move integration shard retries inside nextest and give each shard its own archive group. The merge touched only the CI workflow and its discovery scripts; the branch's cucumber stake and sequencer registration work is unchanged.
Merge dev at 5562eee (drop the sequencer_bootstrap and bridge nextest serial groups, Bedrock confirmation depth 3 with the slow-finality variant at depth 10). The merge applied cleanly with no conflicts; the branch's cucumber stake scenarios and wallet helpers are unchanged.
Merge dev at bafd07b (bump logos-blockchain to 0.3.5-rc.1) into test/sequencer-registration-prerequisites. The merge was clean: dev's changes land in the clock program, bedrock and channel-signing actors, sequencer config and the bridge integration test, none of which overlap the stake-scenario cucumber steps and wallet helpers this branch adds. integration_tests type-checks with its tests on the merged tree.
Merge dev at 5cebe5a (size-metered privacy-preserving transaction storage fees, pinned bridge_lock zones, cross-zone messages from any program, actor metrics) into the sequencer registration cucumber suite. The merge applied cleanly with no conflicts; the branch's own changes are limited to the cucumber stake scenarios, their step definitions and the wallet test app, and the integration_tests crate still type-checks on top of dev.
Merge dev at 14cb5d0, which moves the testing framework imports under the logos-zone-sdk crate. The only conflict was the import block of the integration-test wallet app: keep the branch's program_loader_core import next to dev's zone-sdk re-exports.
Merge dev at dcbd318 (proven fee tips). The merge applied without conflicts; the stake cucumber scenarios and wallet test helpers of this branch are unchanged on top of it.
🎯 Purpose
Add the chained-caller test prerequisites to continue with cucumber sequencer integration tests: a
stake_chain_callertest guest, runtime deployment of test programs throughprogram_loaderfrom the scenario wallet, and paid-by submission against a deployed program account.The second-mover prerequisite (P-21) originally in this PR was dropped on the merge with
dev:Stakenow moves the funds natively and no longer chains into a caller-chosen mover, so there is no second mover to exercise. P-19 has no target either, asConfirmStakeis gone from the program.⚙️ Tests added
stake_instruction_validation.feature—@stake_instruction_ci🧪 How to Test
🔜 Future Work
To add remaining tests planned for sequencer registration these prerequisites need to be developed first:
Related tests will be added in upcoming PRs.
📋 PR Completion Checklist
Mark only completed items. A complete PR should have all boxes ticked.