Repository navigation
Support explicit repeat proxy deployment for isolated tests #3
Description
Carry forward upstream keep-network/hardhat-helpers#38: keep-network/hardhat-helpers#38
The helper rejects a deployment name already present in hardhat-deploy's records. Upstream reported this when testing L2TBTC independently of deployment scripts and running the test suite twice.
Investigate a documented opt-in fresh-deployment mode or test fixture/record isolation. Keep the existing duplicate-deployment guard as the default for production scripts.
Acceptance criteria:
- Reproduce repeated test runs with a persisted local deployment record.
- Explicitly define whether the option replaces a record, uses a separate deployment name, or resets test-only records.
- Exercise storage/layout validation and ensure failure preserves the previous record.
- Document the distinction between fresh deployment and an idempotent replay of an existing deployment.
Activity
mswilkison commentedon Sep 10, 2026
Implemented in the two maintained release-line PRs: #6 (ethers v5) and #7 (ethers v6).
Both include optional base/upgrades entry points, explicit fresh proxy deployment with record preservation on failure, real-chain upgrade regression coverage, and verification against existing artifacts. The v5 task now passes noCompile with etherscan 3.1.8+; v6's supported verification task already avoids main compilation. V6 preparation also leaves the canonical record unchanged until execution is confirmed.
Validation passed on all four supported test stacks: 55 tests per v5 stack and 71 per v6 stack, plus typechecks, builds, lint/format checks, packaging, and clean tarball consumers without the optional OpenZeppelin Upgrades plugin.
Leaving this issue open for the two PRs to merge; the implementation work is complete. README and MIGRATION.md describe the API and consumer changes.