jaskij

> I looked into this request some time ago, but unfortunately, due to the way that etl::delgate works, it cannot accept runtime function pointers. Ah, sorry, I didn't explain it...

Oh, I do, but I don't use `etl::delegate` often enough for this to be intuitive, so I always loose twenty-thirty minutes figuring out build errors whenever this comes up. So...

> I'll put it on my TODO list. I'll make a PR, hopefully before Christmas, sounds like a great first issue to get my feet wet with contributing.

Between Christmas chaos and my mental state, I utterly forgot about this, sorry. I'll get to it over the weekend.

What you describe here really reminds me of the design of [`valuable`](https://docs.rs/valuable/latest/valuable/) - while it doesn't get much traffic, it has unstable support in `tracing`, one I have used to...

> I'm still torn on what should be "standard" Personally I would be inclined to follow [OpenTelemetry](https://opentelemetry.io/docs/reference/specification/metrics/) since it is an effort to standardize both metric and tracing data. >...

For the internally handled option, I would expect, at least: - an easy way to convert `UnitName` and `Unit` to a string, for the (common?) case where the underlying protocol...

You are right that with the `Custom` escape hatch, the set of units directly represented in `metrics::Unit` can probably be small and lean. That said, `Custom` probably needs to be...

> Like, it's technically correct, no question, but it would definitely be weird. I guess this implies we might need to make prefix either optional or have a `None` variant....

Just looked at commit activity, and there's barely anything in `main` since the last point release, so I'm assuming this tool is in maintenance mode? Or just doesn't need new...