Don't we need grades for blockchain developers like a developer, a core developer and finally, a hardcore developer?
Evgeniy Danilenko
1,500 posts
Evgeniy Danilenko
@EugeneDVV
Developing a default framework for low latency cross-chain applications @PelagosNetwork, a former core Blockchain development in EF, @0xPolygon, @ErigonEth
- It's been 2 days after @0xPolygon hard ford. So far no reorgs deeper then 10 blocks polygonscan.com/blocks_forked . We expect that now most of reorgs will be less then 16 blocks. It's just a mitigation of the problem, not real finality, although it looks like quite efficient mitigation
- Evolution, not revolution. One of goals of @0xPolygon PoS is stability. However, last year we figured out how to evolve, keeping stability. It's possible to do a lot with backward compatibility. Eg Heildall-Bor communication: we had lack of observability and stability,
- Replying to @HouseofChimera @FantomFDN and 7 othersIn my developer opinion, Lachesis is the best consensus I've ever seen. For DeFi we need instant finality, which Lachesis has, that also is essential for API. I tried working with Avalanche many times and finality was always an issue. I end up writing my own layer upon API.
- I strongly believe that @0xPolygon PoS developers need to be highlighted. Starting to write random things that were implemented or achieved by them. @0xkris16 working hard to test milestone finality - the very first type of finality Bor is going to have - also he's implementing
- More atomic locks resulted in x8 performance gain for @0xPolygon transaction pool. Now network can handle more transactions before it'll influence block building time. With block interruption features it effectively makes Polygon protected from spam.
- It might be a mistake to think that blockchain development is fundamentally different from typical web2. I think updating a chain is considerably easier than dealing with Android phone users that have a 3-year-old outdated app and no intense to change a phone)
- Implementing zks from the very scratch (curve addition, scalar multiplication) is such a beautiful thing to do. So clean and wonderful. It's one of the greatest achievements of crypto.
- Many thanks to @0xSharma for finishing github.com/maticnetwork/b… . Tests spamming transactions showed extreme block time stability! Just a few millisecond difference with the aim block time of 2s.
- One of big changes in 0.3.4 Bor @0xPolygon is going to be transaction pool performance optimizations. It took us a huge amount code and testing to do github.com/maticnetwork/b… . We were able to reduce caused by txpool block latency p50,p75 by 8-10 times, p99 - 4-5 times.
- Currently I am in Polygon and for me there is no any long-term solution, except of #Erigon. Literally all blockchain clients share the very same data model with same issues. Only Erigon made research and implementation of a better solution.Replying to @0xNiveda3) Switch to Erigon client (long term solution) Erigon client (github.com/ledgerwatch/er…) now natively supports Bor consensus. The state data is as small as 4.5 TB for Polygon PoS Mainnet! All functionality (including additional RPC calls that Bor implements) have been tested.
- Everyone is talking about scaling transaction, but nobody is talking about scaling Ethereum API. However, go-ethereum in many, many cases provides just few requests per second. E.g eth_call. How would we scale DeFi without scaling API?
- One of the main overhead in Ethereum transaction execution is golang GC(this is true for @ErigonEth , @0xPolygon , @go_ethereum ). This might be solved with github.com/ethereum/evmone , but will introduce a lot of overhead from CGo. Hovewer, golang asm doesn't have such overhead
