You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Basically same problem statement as #7518 . Currently taker fee sends are taking extra logic, gas, and sync time by being on every swap. (2.2% of block sync time on mainnet right now)
Instead we could just take the taker fee from the user, and send them to a "holding address". We can then send funds from that holding address to stakers + community pool once per day, at the beginning of epoch. We just need to be careful that its impossible for a normal address to send to this module account. (As then they could spam it with too many denoms)
As we want to implement more complicated taker fee distributions, e.g. kickbacks for some pairs, I am imagining this architecture becoming that there is one address per "type" of taker fee distribution, based on how we later choose "type" of fee distribution.
I am anticipating this to be a 1.5% block sync improvement on its own. (As we are still doing one token send still and calculating the taker fee. However it is removing a community pool send, and updates to two trackers, so 3 database writes and their respective, notable, CPU logics)
The text was updated successfully, but these errors were encountered:
Basically same problem statement as #7518 . Currently taker fee sends are taking extra logic, gas, and sync time by being on every swap. (2.2% of block sync time on mainnet right now)
Instead we could just take the taker fee from the user, and send them to a "holding address". We can then send funds from that holding address to stakers + community pool once per day, at the beginning of epoch. We just need to be careful that its impossible for a normal address to send to this module account. (As then they could spam it with too many denoms)
As we want to implement more complicated taker fee distributions, e.g. kickbacks for some pairs, I am imagining this architecture becoming that there is one address per "type" of taker fee distribution, based on how we later choose "type" of fee distribution.
I am anticipating this to be a 1.5% block sync improvement on its own. (As we are still doing one token send still and calculating the taker fee. However it is removing a community pool send, and updates to two trackers, so 3 database writes and their respective, notable, CPU logics)
The text was updated successfully, but these errors were encountered: