Eliza Weisman

I imagine the wire format changes are going to be something like this: - add a "runtime ID" to identify runtimes - add a "watch runtimes" RPC that returns a...

In general, I think we should try to prefer whichever approach makes the `console` CLI do more of the work, and avoid making the application do anything that we _can_...

We may want to do some actual benchmarking, here. It could be good to measure the performance impact of adding the `tracing-causality` library to a busy application (i.e. in a...

@seanmonstar > [...] assuming that the util function does essentially the same thing [...] I'm pretty sure it does, because IIRC, I basically copied and pasted the (at the time...

The biggest issue, I think, is that if the program is exiting, and we drop some guard that's held in `main`, we have almost certainly _already_ dropped the Tonic gRPC...

@briankung yeah, we can _definitely_ do that --- it should be fairly straightforward to detect when individual _tasks_ panic, when using a runtime where task panicking is captured. When a...

It's possible that this data is also collected by tokio's runtime metrics feature (https://github.com/tokio-rs/tokio/issues/4073), which I would like to start integrating into the console. It might be worth looking at...

I think the issue is that the task's span is (incorrectly) being kept alive by the child task's span. We'll probably want to change the `tokio` instrumentation so that task...

Another option that could be worth considering is giving the task spans some way of explicitly indicating that the task has terminated (e.g. it could `record` a field value like...

cc @seanmonstar