fix max seat number computation #12
Conversation
It's still not 100% accurate, but it would give the result closer to the truth
| std::cmp::max( | ||
| seats_before_statelessnet, | ||
| protocol_config.num_chunk_only_producer_seats, | ||
| ) |
There was a problem hiding this comment.
I'm not sure if we should sum up these numbers or taking the maximum. Current value is for sure smaller than we need.
current <= maximum <= sum
For Statelessnet, the values are 20, 300, 320.
Instead of spending hours on the investigation, I suggest starting using maximum and fix it later again if we see the other issues here.
| protocol_config.num_chunk_only_producer_seats, | ||
| ) | ||
| } | ||
|
|
||
| /// This implementation is ported from near-api-js: | ||
| /// https://github.com/near/near-api-js/blob/bdbf839f54fbc399d7299da8cf9966bbdc426238/packages/utils/src/validators.ts#L24-L50 | ||
| fn find_seat_price_for_protocol_before_49( |
There was a problem hiding this comment.
Also, I think we don't need to check protocol version for the current change because protocol_config.num_chunk_only_producer_seats will be zero for older protocol versions, so it will not affect the result.
|
@telezhnaya I checked it on mainnet and it does not look right to me: 
The 100th seat (the last block producer seat is 816k NEAR at the moment, so you only need that to be the block producer, and you need even less to be a chunk producer!) To be fair, the main branch version is even more off: 
|
|
@telezhnaya What is our final call here? Do we need this PR to be merged? Can you double-check that the numbers match the expectations? |
|
No, not now |
Fixes #11
It's still not 100% accurate, but it would give the result closer to the truth.