crypto-org-chain / chain-main

Cronos POS Chain⛓: Croeseid Testnet and beyond development
https://cronos-pos.org/
Other
494 stars 349 forks source link

Problem: Pending decision on adding `feemarket` module to `chain-main` #758

Open devashishdxt opened 2 years ago

devashishdxt commented 2 years ago

This issue is to discuss about the feemarket module and add it to chain-main if needed as suggested by @yihuang.

yihuang commented 2 years ago

https://github.com/cosmos/cosmos-sdk/pull/11499 the multi-tier design implementation, which can work on cosmos-sdk 0.46 the multi-tier design adr: https://github.com/cosmos/cosmos-sdk/pull/10653

tomtau commented 2 years ago

good to be added, but the multi-tier one won't work until ABCI++ changes, right?

yihuang commented 2 years ago

good to be added, but the multi-tier one won't work until ABCI++ changes, right?

The block space reservation feature would need support from tendermint, I'm not sure if it's possible even with abci++, @JayT106 what do you think on this one? other than that, it's just a consensus fee and tx prioritization strategy, the outcome should be similar to eip-1559 in general, the eip-1559 is like a multi-tier system with many fine grained tiers which has same fee adjustment speeds.

yihuang commented 2 years ago

But without the block space reservation, it seems vulnerable to certain attack where one saturate block space with low tier txs to push up the price of higher tier tx whose price adjust at a faster speed. but one don't have direct incentives to do these kind of attacks though.

JayT106 commented 2 years ago

The abci++ indeed cannot solve the low tx fee tier squeezed issue. For the block space reservation feature, the Tendermint dev team may not want to implement this. The different apps will want different block proposing strategies I think.

And from the validator's perspective, the high processing fee makes more incentive to be a validator.

If App/Consensus protocol tx throughput can be improved, the block space reservation might not be that important.