Henry de Valence
Henry de Valence
we might need to upsert rather than discard, if the TCT behavior is to retain the auth path for the _newer_ duplicate leaf, since in that case the TCT would...
The `CheckTx` method actually provides a numeric error code. However, we don't use it, because we don't explicitly enumerate our error types. This would be good to do, but unfortunately...
xref #3505
I think that the approach of creating synthetic `Begin/EndBlock`s is not a good one compared to doing JMT surgery. While doing surgery on the JMT to edit the version behavior...
Idea: what if we avoid surgery entirely by reading all of the keys out of the old database and writing them into a new one as part of a schema...
What's the advantage of doing this rather than providing a `.tar.xz` of the `pd` home directory?
Got it, I was confused by the term "snapshot" because CometBFT has a notion of p2p snapshot exchange, which we're not currently using.
Wait, why do we need to do that, instead of just not calling commit() at the end of InitChain? We know we'll never get a duplicate InitChain message from Tendermint,...
Hmm, but the app hash is app-defined, what if we simply returned [0u8; 32] in the InitChain response, and waited for the first Commit to compute a real one?
Alternatively, if the all-zero array is used by Go/Tendermint as a default value, we could use [255u8; 32] (all ones); either works just as well for our purpose, I think.