Conversation
| - Improved tracing span fields ([#1650](https://github.com/0xMiden/miden-node/pull/1650)) | ||
| - Replaced NTX Builder's in-memory state management with SQLite-backed persistence; account states, notes, and transaction effects are now stored in the database and inflight state is purged on startup ([#1662](https://github.com/0xMiden/miden-node/pull/1662)). | ||
| - [BREAKING] Reworked `miden-remote-prover`, removing the `worker`/`proxy` distinction and simplifying to a `worker` with a request queue ([#1688](https://github.com/0xMiden/miden-node/pull/1688)). | ||
| - Fixed `TransactionHeader` serialization for row insertion on database & fixed transaction cursor on retrievals ([#1701](https://github.com/0xMiden/miden-node/issues/1701)). |
There was a problem hiding this comment.
Moved this previous entry here as it is better suited here.
There was a problem hiding this comment.
I'm going to need some more context in how this is being used? If its just for testing why not just implement a very basic proving server?
As a general rule I'm not very comfortable having "libraries" exposed from the node since we're a binary application. Changes here would not be considered breaking.
There was a problem hiding this comment.
I'm going to need some more context in how this is being used
It's used on the testing prover of miden-client
If its just for testing why not just implement a very basic proving server
It's just for testing, so implementing everything from scratch is a possible solution. I've pushed a possible implementation to the miden-client update PR 755123e. It's not that more complicated so we can close this PR and just roll with it.
There was a problem hiding this comment.
We were using this a library in the client also to benefit for rust artifacts caching in the CI, the same we do with the node using its components as library. I agree that it is not completely necessary, best case scenario we should use the actual binary from the node to ensure that the integration works.
There was a problem hiding this comment.
Yeah, I think we briefly discussed this with @juan518munoz: We could clone the binary and run it, it's just a bit more cumbersome to manage revisions and artifact caching (with the upside that it tests the integration better)
There was a problem hiding this comment.
I mean if its for our own internal testing purposes; I don't mind it so much.
I'm just thinking about our "internal" RPC client that then got used by actual teams and I'd prefer to not have that be an option. If exposing this is easiest for you then lets do it, and just note in the readme and doc comments that its not stable.
After #1688 the miden-client lost the capability of using the testing prover due to it's previously used structures/methods becoming private or removed. This PR re-adds this functionality.