- Video: https://www.youtube.com/watch?v=pMksa4d9a_Y
- Previous: 2025-08-19 Technical Fellowship OpenDev Call
Summary
-
500ms blocks approaching finish line, separating block latency from core resource usage
-
Elastic Scaling launched on Polkadot - celebrating Polkadot 2.0
-
AH migration code complete, waiting for reviews and tests - Kusama deadline October 7th
-
Fellowship retention vote loophole fixed for 32 initial members
-
JAM gas cost model: SIMD implementation progressing, basic algorithm translated
-
JAM conformance fuzzing: 18-19 teams enrolled, daily bug reporting, preparing for milestone one audit
-
Proof of possession work: added support for two-proof crypto types
-
Hard pressure referendum passed, implementation ready for review
-
Fungible trait migration ongoing for multiple pallets
-
Polkadot Builder Party hackathon announced for early October
Basti (Rank 6)
-
500ms blocks
-
published blog post explaining the approach
-
can now see the finish line, no more blockers
-
still writing tests and discovering improvements
-
key innovation: separating block production latency from resource allocation
-
previous system: block production married to core count (12 cores = 12 blocks)
-
new approach: independent concerns - latency vs compute resources
-
-
examples of flexibility:
-
200-500ms blocks with only one core (high latency, low compute)
-
three cores at native speed enables continuous 6-second production
-
-
goal: not wasting resources while maintaining low latency
-
-
host functions discussion
-
coordinating rollout strategy for multiple new host functions
-
batching BLS proof of possession and new allocator host functions
-
cannot activate without coordination due to breaking changes
-
need to communicate with parachains for upgrade coordination
-
considering Westend testing before mainnet activation
-
Elastic Scaling celebration
-
referendum passed, feature live on Polkadot
-
4 validators still need to upgrade (old version causing disputes)
- if they misbehave, validators get ignored onchain for rest of era (no slashing yet)
Oliver (Rank 3)
-
Fellowship housekeeping
-
fixed retention vote loophole
-
32 members from initial seeding never required retention votes
-
could stay on rank indefinitely without proving activity
-
3 members voted on zero out of 276 possible referenda (rank 3)
-
slowed down all referendum times
-
fix merged in last release, should be “trimmed down and slim” in few months
-
-
Asset Hub migration
-
code complete, waiting for reviews and tests
-
critical deadline: October 7th for Kusama
-
need to merge this week to meet timeline
-
planning two separate releases (Kusama, then Polkadot)
-
releases one month apart
-
allows incorporating Kusama feedback into Polkadot release
-
provides agility if continuous testing finds issues
-
-
Jan Bujak (Rank 3)
-
JAM gas cost model
-
re-implementing SCALE codec in SIMD
-
finished translating basic algorithm (still buggy, needs debugging)
-
tedious process but progressing well
-
next steps:
-
debug current implementation
-
optimize and benchmark
-
potentially write AVX-512 variant for more speed
-
final version will go into Graypaper
-
-
Muharem (Rank 2)
-
Asset Hub migration
-
working full-time on AH migration
-
collaborating with Oliver
-
Davide Galassi (Rank 2)
-
JAM conformance testing
-
working on Parity’s JAM implementation
-
focused on preparing proper fuzzing for milestone one audit
-
current fuzzing not official process, but extensive:
-
18-19 teams enrolled
-
running daily fuzzing
-
public bug reports in jam-conformance repo (hosted under Davide’s GitHub)
-
active conversations with teams about compliance
-
teams can reproduce and fix issues
-
this is not yet the official process, since it runs for just a few minutes
-
-
official audit process:
-
will fuzz each implementation for multiple days
-
duration depends on machine performance
-
likely running on Toaster infrastructure (not personal machine)
-
each team fuzzing time varies: faster teams get less time, slower teams get more
-
-
published performance dashboard
-
comparing teams’ performances
-
milestone 1 not about performance, but understanding ratios important
-
need same block import count across all implementations for same level of confidence
-
-
-
Oliver: asked about reusing infrastructure for official process
-
mini-fuzz tool released today
-
enables teams to self-assess before submission
-
uses fuzzing protocol version 1 (will be used for official audit)
-
-
automation challenges
-
current scripts somewhat rough (Python scripting)
-
manual launch process not optimal
-
will need DevOps team for scaling
-
more teams coming will need better automation
-
-
Seyed Lavasani (Rank 1)
-
proof of possession work
-
fundamental discovery: some crypto types need two proofs instead of one
-
required structural changes to crypto types
-
added new proof type (previously everyone generated one signature)
-
PR coded and approved by Basti
-
final step: 2-year-old PR from Basti needs update
-
-
BLS and Westend discussion
- after final PR, could potentially have BLS on Westend
-
host functions discussion with Basti
-
coordinating BLS host functions with new allocator RFC
-
need to batch host functions together for coordinated rollout
-
cannot just activate without communication (breaks parachains)
-
parachains need backporting or feature upgrades
-
last host functions added ~2020-2021, so not frequent
-
-
RBLS(?) library optimization
-
working on efficiency improvements
-
academic interest: “want fastest in the world”
-
not a bottleneck, but want optimization
-
important to complete before adding host functions (implementation gets pinned after)
-
Dom Valles (Rank 1)
-
“hard pressure” proposal work
-
referendum just passed
-
implementation up for review
-
not particularly complicated
-
hoping to merge and include as patch after AH migration
-
-
timeline discussion
-
targeting release after Polkadot AH migration (early November)
-
could potentially do mid-October separate release
-
psychological importance of having as separate release noted
-
-
next up: RFC 150 - voting while delegating
-
has draft implementation in SDK
-
complements RFC for easier review
-
received reviews from Basti and others
-
Dom just returning from 5-day vacation, will address reviews
-
Ludovic Dominguez (Rank 1)
-
new Fellowship member (joined ~2 weeks prior)
-
background
-
started as game developer (Unity, C#)
-
moved to web3 during COVID
-
3 years in EVM ecosystem (smart contracts, NFTs, web3 games)
-
attended PBA5 Singapore, stayed in Polkadot
-
now working at PBA full-time
-
-
current work
-
migrating pallets from old currency framework to fungible trait
-
low priority task that team lacked resources for
-
coordinating with Pablo on same migration work to avoid duplication
-
Pablo (Rank 2)
-
fungible trait migration
-
published forum post proposing fungible interface for storage deposits
-
analogous to current fee payment via asset conversion
-
uses ticket system: convert to ticket for deposit, convert back on refund
-
prioritizing migrations by difficulty and impact
-
-
pallet-identity migration
-
identified as simplest migration candidate
-
easiest in terms of technical feasibility, migrations, and impact
-
currently working on implementation
-
-
coordinating with Ludovic to avoid duplicate work
Additional Notes
-
Polkadot Builder Party Hackathon
-
launching first week of October (~2.5 weeks from call, October 6th)
-
two tracks:
-
web2 developers: using parachains as services (payments, robotics, AI, storage, compute, gaming, social, knowledge management)
-
web3 developers: build your own parachain
-
-
treating parachains as services, presenting “web3 cloud” concept
-
seeking hackathon project ideas from Fellowship
-
forum thread: “Polkadot Builder Party Ideas Hackathon Ideas”
-
opportunity for Fellows to host workshops (can hook up with Parity/Web3 Foundation)
-
workshops become part of back catalog for future reference
-