Argument-0002: Retention at Rank I

Member details

• Matrix username: @alexandre.balde@parity.io
• Polkadot address: 13jBAtYJar4xujPaEx41FxjSt9PqU7LqJRbySJiVdMtuWN42
• Current rank: 1
• Date of initial induction: 2025-10-28
• Date of last report: 2025-11-19
• Area(s) of Expertise/Interest: System Parachains, Polkadot Runtime

Reporting period

• Start date: 2025-12-01
• End date: 2026-02-28

Evidence

During this reporting period I continued my work on testing Polkadot protocol components through the Polkadot Ecosystem Tests (PET) project, while also contributing directly to the Polkadot SDK and the Fellowship runtimes CI infrastructure.

Polkadot SDK contributions

I contributed improvements to pallet testing and SDK maintenance:

1. Contributions to the scheduler pallet: added some clarifying comments to calls in the scheduler pallet, Lookup<T> checks in unit tests, and a missing test verifying rescheduling of a task to a full block (#10511, merged).
2. Added tests for SenderInSignatories/SignatoriesOutOfOrder error cases in pallet-multisig and clarifying notes to approve_as_multi/as_multi comments (#10843, merged). This was a follow-up from PET end-to-end testing (PET#305, PET#323).
3. Reworked the remote externality child key retrieval test, which was failing since the proxy pallet has no child storages (#10866, merged).

try-runtime-cli maintenance

In line with my focus on testing, I volunteered to become the maintainer of try-runtime-cli, a tool for validating runtime upgrades against live chain state. During this period I updated dependencies, fixed CI workflows and cut new releases (PRs).

PET CI infrastructure improvements

I worked to improve the reliability and performance of PET in its own repository, and in the Fellowship runtimes repository's CI pipeline, where it can resume its function as a regression testing tool across Polkadot and Kusama networks:

1. Reworked CI to use Chopsticks servers as RPC caches to avoid timeouts caused by throttling/stalling from production RPC endpoints (PET#504, merged). This addressed a recurring pain point where CI runs would fail due to network-level issues rather than actual test failures (tracking issue PET#506).
2. Follow-up refactoring to the Chopsticks caching approach to instead use Subway (thanks @xlc) as the RPC proxy/cache (PET#519, awaiting review).
3. Reworked the test.yml workflow in the Fellowship runtimes CI to use a test matrix with individual retry/timeout configurations, Subway-based RPC caching, and filtering to only run end-to-end suites for runtimes under the Fellowship's purview (runtimes#1068, awaiting review). This closes runtimes#1010.

PET test coverage expansion

I fixed an issue where the scheduler pallet would ignore calls scheduled manually via dev_setStorage because incompleteSince was not being set (PET#515, merged) — this was a subtle interaction between Chopsticks' storage injection and the scheduler's service_agendas scanning logic, which eliminates one common cause of spuriously failing tests. - This was minor, and although PET is not included in the Manifesto's scope as Fellowship work, this contribution should be considered in combination with the previous point: an effort to improve overall reliability of PET in runtimes' CI.

Voting record