feat: cost estimator #68
Conversation
| } | ||
|
|
||
| impl fmt::Display for ExecutionStats { | ||
| fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { |
There was a problem hiding this comment.
I thin it would be good to print this out in a comma separated format (like a CSV file), something like this: https://github.com/succinctlabs/rsp/blob/main/bin/host/src/execute.rs
| ENV PATH="${GOPATH}/bin:${PATH}" | ||
|
|
||
| # Set up Go environment | ||
| WORKDIR /optimism |
There was a problem hiding this comment.
I feel like the correct way to do this is to put in the optimism repo & then build the docker file from there and push the image to docker hub?
I think it's semantically a bit incorrect to clone github repos in a dockerfile lol
There was a problem hiding this comment.
Yeah after the server is finalized can do this.
| Ok(response.ranges) | ||
| } | ||
|
|
||
| async fn start_span_batch_server() -> Result<Child> { |
There was a problem hiding this comment.
nit: I feel like what's more "semantically correct" is to execute the docker command first & then this bin, instead of starting the server in bin/cost_estimator.rs
| let rollup_config = RollupConfig::from_l2_chain_id(l2_chain_id).unwrap(); | ||
|
|
||
| // Fetch the span batch ranges according to args.start and args.end | ||
| // TODO: If the ranges are greater than 20 blocks, we will have to split them in a custom way. |
There was a problem hiding this comment.
Is this because our executor is slow for large blocks?
|
|
||
| let prover = ProverClient::new(); | ||
|
|
||
| // TODO: These should be executed in parallel. |
There was a problem hiding this comment.
yeah that's a good callout--I feel like you could do a par_iter and use different threads since each executor is just single-threaded
* write in parallel * write in parallel * wip, wtf * add * feat: cost estimator --------- Co-authored-by: Ubuntu <[email protected]>
Overview
Add a cost estimator tool that outputs the relevant statistics across a range of blocks.
Simplify the pipeline for surfacing the next relevant block.
Simplify program block execution logic, given that only one payload is produced at a time.
Add more resilient logic for timing out the execution of the native host.