@KonradStaniec MsgInjectedCheckpoint is currently been handled directly in PreBlocker. Why introducing this dummy handler would affect apphash?

@KonradStaniec MsgInjectedCheckpoint is currently been handled directly in PreBlocker. Why introducing this dummy handler would affect apphash?

Even though message is handled in PreBlocker, it is still transaction in the Block which means it gets executed during DeliverTx here - https://github.com/cosmos/cosmos-sdk/blob/8cc13ed2e2b4fca118b27cfaaaef71c2420aff06/baseapp/baseapp.go#L771. Thats why we see it as failed in explorers as its execution code is different than 0.

This is not part of the appHash but it is part of the LastResultsHash - https://github.com/cometbft/cometbft/blob/0d4d40f19dbf7e724b9a65b598b5431e733eee0c/types/block.go#L352

This means the node with with this change will produce and expects different headers that are in the current testnet, which will lead to chain split.

Actually , the easiest way to test this pr would be to try to sync the node build with this version with current testnet, this would verify:

1. Whether there is risk of chain split or not
2. Whether transactions which checkpoints are now considered successfull i.e have Code == 0

@KonradStaniec Thanks for the explanation. It all makes sense. I agree that this pr is not a urgent fix can can leave it for v2 but will test it out soon

Update after a sync test:

1. Syncing to test-net-5 failed at height 361 with error preblocker: failed to extract injected checkpoint from the tx set: failed to decode injected vote extension tx: proto: wrong wireType = 2 for field Status: tx parse error [cosmos/[email protected]/x/auth/tx/decoder.go:49], which is expected because in this pr, we have introduced a new attribute placehold_address in the MsgInjectedCheckpoint
2. I start a local testnet and start from scrach. At the height epoch * i + 1, got error 'out of gas in location: ReadFlat; gasWanted: 0, gasUsed: 1000: out of gas'. I think this is also expected as every tx in Cosmos should be charged with a basic gas as fee.

Based on the above, I'm not sure adding a dummy handler is a right way to go. Any ideas? @KonradStaniec @Lazar955 ?

I start a local testnet and start from scrach. At the height epoch * i + 1, got error 'out of gas in location: ReadFlat; gasWanted: 0, gasUsed: 1000: out of gas'. I think this is also expected as every tx in Cosmos should be charged with a basic gas as fee.

Ahhh true! to make it successful we would need to pass whole ante handler (gas and signatures). Thats a lot of custom work.

Tbh given that the only downside of not having this improvement it the fact that our custom transaction shows as failed in explorer (but it is possible to decode it and get its data) I would

• close the PR
• document our findings somewhere. (Maybe ADR in pm repo, describing this findings that it is not trivial to make this custom tx successful ?)

I agree with closing this PR.

I wondered maybe we should raise an issue to Cosmos as I think SDK should have a common solution for this other than each project is doing their own way. Or do we want to consult Cosmos friends first?