Proof integration fix store #1505
Conversation
eshelB
requested review from
Cashmaney,
toml01,
assafmo and
liorbond
as code owners
| @@ -5,7 +5,7 @@ go 1.19 | |||
| replace ( | |||
| // dragonberry | |||
| github.com/confio/ics23/go => github.com/cosmos/cosmos-sdk/ics23/go v0.8.0 | |||
| github.com/cosmos/cosmos-sdk => github.com/scrtlabs/cosmos-sdk v0.45.13-0.20230726163133-dc2b5eab1db5 | |||
| github.com/cosmos/cosmos-sdk => github.com/scrtlabs/cosmos-sdk v0.45.13-0.20230808154448-1e7ad22811a4 | |||
There was a problem hiding this comment.
Seeing this scrtlabs/cosmos-sdk@1e7ad22811a4, is there a way to pass height-1 when we call this? I'm not sure what are the performance implications.
There was a problem hiding this comment.
If we pass height-1 it breaks all other queries other than wasm
There was a problem hiding this comment.
wasm doesn't care as it has the cache
There was a problem hiding this comment.
I told Itzik about this, we thought maybe we should see what happens and if there are performance implications, we know what to look for. Also memory implications, as now queries have the whole db. But probably none of these implications exist as these are all pointers to subsections of the iavl store?
There was a problem hiding this comment.
If we see performance issues, we could probably even create a context with both height and height - 1 instead of the whole store.
There was a problem hiding this comment.
I tried to find a way to do this but it seems that the multiStore there didn't support more than one specific version, only all of them
uses a version of the cosmos sdk that passes a multistore to queries which contains height-1 as well, not only the last height