Crystal Durham

Oh no, the problem with this approach is that `ExtensionsMut` needs to have and gives out `&mut` access to all of the extensions on that span, so it needs to...

So this is where I got in a quick test; the `Data` portion of it works, but the interface of `Extensions[Mut]` causes problems https://github.com/tokio-rs/tracing/compare/master...CAD97:tracing:pool-exts

Concept for a smaller incremental improvement(?): Currently (IIUC), the `AnyMap` hashmap is reused, but the actual extensions' boxes aren't. A small improvement then is instead of `AnyMap![T => T]`, store...

~~breaking changes? #1612? #2048!?~~ I'll revive the branch and see what I can do to chase the lifetime errors out then 🙂 Also, https://twitter.com/mycoliza/status/1530356065610567686?s=20&t=q2_dwEk3_1GnWorF1F4pBg wants to get into a tracing-subscriber...

#2048 would fix this, though using a dynamic level is advised against in the current shape, as > Although it is possible to override the level, it is generally advisable...

Rebased over the tracing-attributes split.

[tracing-attributes 0.1.18](https://github.com/tokio-rs/tracing/releases/tag/tracing-attributes-0.1.18) has been published. Is there anything specific blocking this, or are we just waiting on person time?

Now that we have #1906, I guess I should look into capturing by-`Valuable` rather than by-`Value`.

Alright, this now supports capturing by-`valuable`, with the autoref order of `Valuable` > `Value` > `Debug` (This can be adjusted if desired; it might make sense to swap `Valuable` and...

One thing I just want to confirm is the way we want to handle it: the current order the conversions are tried is - by `Valuable`, then - by `Value`,...