Neither test runner verifies the transaction rejection reason against the fixture's expectException:
- statetest:
expectException is reduced to a bool at load (o.exception = j.contains("expectException")); the runner only asserts the tx was rejected and the state root matches.
- blockchaintest: an invalid block caused by a tx just
continues on any rejection (if (!res.rejected.empty()) continue;) without comparing the reason.
So a transaction rejected for the wrong reason still passes. We now match block-level reasons (#1556); we should do the same for TransactionException.*.
Needs a mapping from evmone's tx ErrorCode (plain geth-style messages) to EEST TransactionException.* constants — those messages can't simply be renamed since t8n/state-test output relies on the geth-compatible form.
🤖 Generated with Claude Code
Neither test runner verifies the transaction rejection reason against the fixture's
expectException:expectExceptionis reduced to a bool at load (o.exception = j.contains("expectException")); the runner only asserts the tx was rejected and the state root matches.continues on any rejection (if (!res.rejected.empty()) continue;) without comparing the reason.So a transaction rejected for the wrong reason still passes. We now match block-level reasons (#1556); we should do the same for
TransactionException.*.Needs a mapping from evmone's tx
ErrorCode(plain geth-style messages) to EESTTransactionException.*constants — those messages can't simply be renamed since t8n/state-test output relies on the geth-compatible form.🤖 Generated with Claude Code