Damien Mathieu

Running benchmarks on a private repository seems like a bad approach. Contributors wouldn't be able to know what we run, or when. Nor would they be able to add new...

It looks like this is fixed with #2997, since we now have a single PR for all dependabot upgrades, created by whoever runs the workflow.

It seems like this issue is resolved with #2997, since we now have a single PR for all dependabot upgrades.

This should include an automated test

How about using the logr interface instead of a custom-made logger that will only be used here?

You mention you're updating the span name to follow semantic conventions. Is there a spec specifying those conventions for kafka, or examples of other libraries using the new name?

I'm not sure I understand how this is an update to "follow messaging semantic conventions" if no other libraries is doing the same thing, and no conventions are defined?

Unfortunately, the lint job is a bit flaky at the moment 😢 I reran it. But it seems the failure now is indeed related to your changes.

The metrics SDK is currently under heavy revamping to be moved out of beta (see the [new_sdk branch](https://github.com/open-telemetry/opentelemetry-go/compare/new_sdk/main)). I think this kind of change should wait for that work to...

We already support environment variables for configuration, as [specified](https://github.com/open-telemetry/opentelemetry-specification/blob/f124b21de091f3d668e4ce11b954ec17e3da231f/specification/sdk-environment-variables.md).