Argument-0008: Retention at Rank I


Report Date	2026/04/25
Submitted by	Damilare Akinlose

Member details

Matrix username: @dharjeezy/:matrix.org
Polkadot address: 12GyGD3QhT4i2JJpNzvMf96sxxBLWymz4RdGCxRH5Rj5agKW
Current rank: 1
Date of initial induction: 2024-04-29
Date of last report: 2026/01/18
Area(s) of Expertise/Interest: FRAME/Runtime & Node

Reporting period

Start date: 2026/01/19
End date: 2026/04/25

Argument

I am applying for Rank I retention.

In this cycle, my work focused on three areas. I added new try_state invariant checks to several consensus and reward related pallets. I landed a long-running overhaul of the litep2p ping protocol along with the connection keep-alive redesign that it required. I also continued progressing developer-facing work on block number providers, fungible-trait migrations, and QUIC transport interoperability. Beyond code, I published a deep-dive write-up of the ping work on the Polkadot Forum so that the design rationale would be available to the wider ecosystem.

Below is a summary of my key contributions:

Reliable & Resilient

`litep2p`: Spec-Compliant Ping Protocol & Substream Keep-Alive Redesign

After several review cycles, my overhaul of the litep2p ping protocol was merged in PR #416. The change brings the implementation in line with the libp2p ping spec. Pings are now repeated rather than sent only once on connect. Inbound and outbound substreams are reused for subsequent requests, which is required for smoldot compatibility. Ping substreams are also excluded from the connection keep-alive mechanism, so idle connections can be closed on timeout when only ping and identify substreams remain open. I tested the change across litep2p to litep2p, litep2p to pre-PR litep2p, and litep2p to smoldot. As a result, dead-peer detection now happens within roughly 25 seconds, which is meaningful for validators identifying unreachable peers during consensus.

To document the design and the cascading keep-alive redesign that the ping work required, I authored a Polkadot Forum article: A deep dive into overhauling litep2p's ping protocol. The post walks through periodic pinging, spec-compliant random payload verification, the new tagging of protocols as either keep-alive or non keep-alive (separating application protocols from infrastructure ones), and the backwards-compatibility considerations for older litep2p nodes.

Introduce `try_state` Hook for Pallet Authorship

I added a try_state hook to pallet-authorship to verify a key storage invariant. The implementation was merged in PR #11215 and closes part of the broader issue #239 initiative to add try_state compliance across pallets in the Polkadot SDK.

Introduce `try_state` Hook for Pallet BABE

I implemented a try_state hook for pallet-babe covering all of its key storage invariants. The change was merged in PR #11216 and continues the systematic effort to make consensus-critical pallets verifiable under runtime checks.

Introduce `try_state` Hook for Pallet BEEFY MMR

I extended the try_state initiative to the BEEFY MMR pallet in PR #11734. The PR defines and asserts the invariants that hold for the pallet, and is currently open and under review.

`try_state` for Pallet Atomic Swap: No Invariants to Assert

Following up on the work I started last cycle for pallet-atomic-swap in PR #10473, I went back through the pallet's storage and logic to determine which invariants a try_state hook should enforce. The conclusion was that the pallet has no invariants worth asserting at the storage level: its single PendingSwaps map only records open swap offers, and the pallet's correctness is enforced at the extrinsic boundary rather than through any cross-storage relationship. I therefore closed the PR with that finding, which removes pallet-atomic-swap from the list of pallets still pending try_state coverage under issue #239.

Flexible & Developer-Focused

Migrate `pallet-conviction-voting` to Fungible Traits

I authored PR #11142 to migrate pallet-conviction-voting from the legacy Currency and LockableCurrency traits to the modern fungible traits, replacing locks with freezes via fungible::MutateFreeze. The change replaces type Currency with type Fungible and type RuntimeFreezeReason in the pallet's Config, introduces a FreezeReason::ConvictionVoting composite enum, and converts all lock operations (set_lock, extend_lock, remove_lock) to their freeze equivalents. To preserve existing user state, I added a SteppedMigration (LazyMigrationV0ToV1) that lazily converts existing locks to freezes at runtime upgrade, together with a v1_migration_step benchmark and a storage version bump. This contributes to the broader fungibles migration tracked in issue #226.

Expose Block Number Provider Name as a Constant

I continued work on exposing the BlockNumberProvider identifier as a runtime constant in PR #9091. The PR extends the BlockNumberProvider trait with an IDENTIFIER constant and adapts the #[pallet::constant] macro to accept attribute parameters, so that it can fetch values from associated constants on types. The end result is that DApp developers can reliably distinguish between local and Relay Chain block numbers in metadata. Review iteration is ongoing.

Interoperable & Compatible

`litep2p`: Support Dialing DNS Addresses Over QUIC

I extended the QUIC transport in litep2p to resolve DNS multiaddresses (/dns/, /dns4/, /dns6/) before connecting, bringing it to feature parity with the existing TCP and WebSocket transports. The implementation in PR #542 closes issue #406 and removes a long-standing limitation that previously required QUIC peers to be addressed by raw IPs.

Voting record

Ranks	Activity thresholds	Agreement thresholds	Member's voting activities
I	90%	N/A	I have voted on 0 out of 0 referenda in which I was eligible to vote (i.e undefined% voting activity). Out of xx referenda in which members of higher ranks were in complete agreement, I have voted in line with the consensus x times (i.e undefined% voting agreement).
II	80%	N/A
III	70%	100%
IV	60%	90%
V	50%	80%
VI	40%	70%

Misc

Question(s):
Concern(s):
Comment(s):