Skip to content

Move Parallel TX Signalling to SeiChain DeliverTx call#291

Closed
BrandonWeng wants to merge 2 commits into2.0.0betafrom
bweng-move-signalling
Closed

Move Parallel TX Signalling to SeiChain DeliverTx call#291
BrandonWeng wants to merge 2 commits into2.0.0betafrom
bweng-move-signalling

Conversation

@BrandonWeng
Copy link
Contributor

@BrandonWeng BrandonWeng commented Oct 7, 2022

Describe your changes and provide context

Previously we were signaling and blocking for each handler in the message. In theory, this is fine as we are supposed to be processed synchronously based on the access operation. However, due to us not having enough granularity in the resource dependency mapping, blocking on messages is not sufficient as the ordering for when each message is unblocked (between different TXs) could be different. This results in different gas fees or different bank AVL tree shapes as the node insertion order matters.

Therefore, as we do not have more granular support for per-message concurrency, we should be okay with just blocking on the entire transaction and ensuring that no r/w can occur unless the entire TX is ready to be processed.

Note that the Gas used is based on the size of the data returned, depending on the ordering of the message process it's possible that some may result in the same bank data being returned with less digits, and therefore costing different gas.

With these PRs merged, we should be able to then define more granular access operations for the message types we want to support e.g Bank Send should have access operations for read to sender, writing to both sender and receiver, and the contractAddr. This is operating on the assumption that the majority of transactions will not have overlapping resources. If it doesn't achieve the results we desire then we can spend more time optimizing for this

sei-protocol/sei-cosmos#35
sei-protocol/sei-cosmos#37
#287

Testing performed to validate your change

Ran the load test 20 iterations with 5 rounds each and saw that the chain didn't get stuck and from previous outputs of the delivered, the messages are being processed sequentially as expected

@BrandonWeng BrandonWeng closed this Oct 7, 2022
@BrandonWeng BrandonWeng deleted the bweng-move-signalling branch December 20, 2022 14:57
masih pushed a commit that referenced this pull request Sep 29, 2025
## Describe your changes and provide context
This properly adds the ante deps message access ops (with index -1) to
the map used for comparator validation. Prior to this, we were
regenerating the antedeps, but this is a more correct fix.

## Testing performed to validate your change
localsei verification
masih pushed a commit that referenced this pull request Sep 30, 2025
## Describe your changes and provide context
This properly adds the ante deps message access ops (with index -1) to
the map used for comparator validation. Prior to this, we were
regenerating the antedeps, but this is a more correct fix.

## Testing performed to validate your change
localsei verification
masih pushed a commit that referenced this pull request Oct 1, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants