Gerd Zellweger

we could have an "inherit" strategy that inherits range/scale etc. from the origin key

Ideally we should have buttons a user can click that sends them to the sandbox and populates the SQL code for a given demo. e.g., other apps I've seen essentially...

So no deletions for now _Originally posted by @ryzhyk in https://github.com/feldera/feldera/pull/2107#discussion_r1693993505_

[This part of the docs](https://www.feldera.com/api/change-a-pipelines-name-description-code-configuration-or-connectors) isn't really useful for a user: 

Profile dump not working when circuit is under high load

9

When the circuit is under high load, getting a profile dump fails with an HTTP timeout.

Strange config values

2

Any reason we add these as defaults to the config every time? IMO they should just "unwrap" to max values if missing? ``` max_output_buffer_time_millis: 18446744073709551615 max_output_buffer_size_records: 18446744073709551615 ```

Can we use the actual pipeline name instead of pipeline-uuid as the name in the pipeline process?

1

I think we should just use the name the user gives to the pipeline in `PipelineConfig.name` when we pass this from the manager to the pipelines. This would mean the...

manager can't copy binary for integration tests

4

Running the manager integration tests with ``` TEST_DBSP_URL=http://localhost:8080 cargo test integration_test:: --package=pipeline-manager --features integration-test -- --nocapture ``` and a separate manager instance with ``` cargo run --bin=pipeline-manager --features pg-embed ```...

Ideally this could be fixed > Why does the merger need to be shared across threads? Off-hand, I'd expect that the merger could be created, executed, and finalized in the...

Trying to start a previously failed pipeline does not work with the pipeline API

3

Trying to start a previously failed pipeline does not work with the pipeline API: (it needs to explicitly clear the error of the failed pipeline) ``` Running the training pipeline....