-
Notifications
You must be signed in to change notification settings - Fork 3.6k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
WIP: SDK2 #328
Comments
querying with proofs will first require a common structure format (with interfaces) for supporting any variety of Merkle proof types, in a series (e.g. for SimpleTree (block header) -> SimpleTree (multistore) -> IAVLTree (iavl)) |
Additionally I think there is an opportunity to lump some common initialization/execution functionality together (such as found in |
Functioning CLI/REST #324 also. I think the last line is not needed for sdk release. It can merge without ibc/staking/governance, and they can be built on top of it, as different projects. Also for merkle proofs of trees, please look at ibc spec v0.3.1, Appendix C. There it specifies a generic binary format to represent any merkle proof and chain them together. It should be compatible with patricia tree proofs as well, but haven't proven that. |
I think at least queries without proof is essential before announcing it. You can put data in, but you can't see it.... |
For now people can send "read" transactions. It's not ideal but let's get something workeable out there and start rapidly iterating on bringing in the necessary new features |
Please explain how a "read" transaction works? |
I think query (without proofs) is super essential. Otherwise, it looks like an ugly hack. And really, they should not be that hard. At least for an MVP implementation. |
A read transaction is a consistent query (ie. a query that passes through consensus). For instance in basecoin, we'd have to define a This is how merkleeyes was written for Jepsen. Query isn't sufficient because it can return stale reads - ie. you might be querying from a node thats behind and meanwhile the value has changed. So for proper testing, you actually need "read transactions", or what some call "consistent reads". Obviously for basecoin that'd be ridiculous, and I'm not proposing we do it. We could and should and must do it for the key-value store app we build, ie. the new merkleeyes-jepsen-app on SDK. Anyways, no one disagrees that queries are essential. Let's get them in ASAP. But let's not block on it - folks can start building and unit testing their apps without it. |
closing in favour of sprint boards and milestones 🕺 |
UPDATE: Look at https://github.com/cosmos/cosmos-sdk/projects/2
Everything on the SDK2 branch:
Pre-announcement
Post-announcement
The text was updated successfully, but these errors were encountered: