[dhaval-supraoracles]

No description provided.

[isaacdoidge]

Review of the dev→1.34.0 merge at head aeb003df. The merge itself is largely sound — a three-way audit of the high-risk files (aptos_vm.rs, gas.rs, validator_txns/dkg.rs, both automation processors, the DKG framework modules, gas schedule, Cargo.toml/lock) found no silently dropped hunks from either parent, all dev-era Supra feature PRs (#290, #302, #338, #342, #348/349/351, #352, #355, #361, #366, #368/369) survive on the head, Supra feature flags 88–101 are correctly restored with original discriminants, and Supra gas params are intact (supra_stdlib.rs byte-identical, eth.trie.*/rlp.* preserved, LATEST_GAS_FEATURE_VERSION now 42). No dev-side tests were dropped by conflict resolution.

Beyond the two inline comments (one Critical, one High), items that need decisions:

1. The merge is 4 commits behind fork dev tip; three are correctness/security fixes (their omission looks like cut-point timing, not bad conflict resolution — but they should not ship missing):

• ce3cebc85a (#5/#6): reachable panic in native_verify_proof_eth_trie — head still has H256::from_slice(&root) (panics on non-32-byte root) at aptos-move/framework/src/natives/cryptography/eth_trie.rs:57 and .unwrap() at :66. Malformed proof input can crash a validator.
• 20a83be0d4 (#370/#371): exclude unrotated legacy-key validators from the validator set when the v2-key flag is active — absent (is_unrotated_legacy_key has zero occurrences on head). Validator-set safety fix in the same key-validation family this merge restores.
• 0c94987f (#13, current dev tip): legacy-key validators migrate to v2 and rejoin (builds on #370/#371); its regression test is also absent.
• 3fecf50d07 (#372): GenesisEvmContract→GenesisEvmTransaction rename for EVM call transactions.
  Recommendation: merge dev tip into this branch once more before landing.

2. Merge-ordering requirement: #303's head alone has Supra feature flags 97–101 clobbered by upstream discriminants (SUPRA_BLS_KEYS/SUPRA_BCFT_CERTIFICATES/SUPRA_DKG/SUPRA_TRANSACTIONS_INCLUSION_PROOFS/SUPRA_AUTOMATION_V2_1) — only this PR fixes that. #303 must not land without this PR; consider merging this branch into #303's so the broken intermediate state is unmergeable.

3. Topology note: private/feat/merge_with_dev (61db493c) is an ancestor of this head, not ahead of it — 912140a7 ("Fixed remaining test cases") is already included. No branch drift.

Findings that pre-date this merge or belong to the upgrade branch itself (persisted OnChainExecutionConfig shuffler compat, missing nonce_validation genesis init, coin.move FA-routing divergence, a pre-existing stake.move duplicate-key-check bug) are being raised on #303 / as separate issues, not here.

[isaacdoidge]

High — merge regression in supra-framework/sources/transaction_validation.move (automated_transaction_prologue_v2, ~line 396 on aeb003df). The file isn't in this PR's diff (the change came in with the earlier merge commit), so posting as a thread comment.

On dev, the prologue asserts automation_registry::has_sender_active_task_with_id_and_type(sender, task_index, task_type) outside the task_type != GST branch — it runs for every automated transaction and is type-aware. On this head, the assert (a) uses the type-less has_sender_active_task_with_id (which hardcodes UST via ..._and_type(sender, task_index, UST)) and (b) sits inside if (task_type != GST).

Consequences:

1. Gas-less (GST) automated transactions skip the active-task check entirely — a stopped, expired, or cancelled gas-less task's automated transaction still executes instead of being discarded with PROLOGUE_ENO_ACTIVE_AUTOMATED_TASK.
2. Non-UST charged task types (if/when any exist) are validated against the wrong type.

This looks like a conflict resolution that regressed to a pre-_and_type version of the assert. Fix: restore the dev form — type-aware check, outside the gas-fee branch. A test that registers a GST task, stops it, and asserts its automated transaction is discarded would lock this in.

[isaacdoidge]

Build break: cargo build --release (workspace) fails on the pinned toolchain (1.93.0 per rust-toolchain.toml).

error[E0283]: type annotations needed
   --> ecosystem/indexer-grpc/indexer-grpc-fullnode/src/convert.rs:983:44
    |  if supply.current_supply.max_value == u64::MAX.into() { ... }

u64::MAX.into() is ambiguous under 1.93.0's inference (fix: BigDecimal::from(u64::MAX) or an explicit turbofish). This blocks anything that pulls the processor lib into its build graph — including cargo build --release -p aptos (the CLI), which is the first step of unit_test.sh, so the framework test script cannot run as-is on this head. CI presumably doesn't do a full release workspace build, which is why it hasn't tripped.

(Ran into this validating the branch locally; cargo test -p aptos-framework, e2e-move-tests, and language-e2e-testsuite don't depend on it and are running separately.)