Upgrade to opentelemetry 0.22

[djc]

Motivation

Would like to upgrade to opentelemetry 0.22.

Solution

This seems mostly straightforward. Upstream deprecated opentelemetry_jaeger::new_pipeline() (which is used in an example) in favor of the OTLP collector for Jaeger; given that we already have an OTLP example should we just delete it?

[jtescher]

Looks good, seems like the jaeger -> otlp exporter switch needs to happen to avoid the deprecation errors

[djc]

I was suggesting just removing the Jaeger example -- are you saying it should be switched to OTLP? (I noticed there's already a fairly elaborate OTLP setup in one of the other examples, so not sure it makes sense to have both.)

[jtescher]

Can remove too, whichever is easier

[djc]

I'm not sure it makes to continue to pursue minimal-versions here if we don't have a similar CI job in the upstream opentelemetry repo to keep it from regressing this.

[jtescher]

yeah true

[rbtcollins]

I think like asked for in #99 the silent API break of i64 histograms should be added to a Changelog entry.

Something like

Breaking Changes

* Upgrade to v0.22.0 of opentelemetry. For list of breaking changes in OpenTelemetry, see the [v0.22.0 changelog](https://github.com/open-telemetry/opentelemetry-rust/blob/v0.22.0/opentelemetry/CHANGELOG.md).
* ia64 histograms will silently downgrade to k:v exports. This is not statically detected. Code like the following produces ia64 histograms:

\```
        tracing::info!(histogram.abcdefg_auenatsou = -19_i64);
\```

[djc]

I've added a changelog entry and bumped the version number. @jtescher do you think this is good to release?

[rbtcollins]

CI failure is
https://github.com/tokio-rs/tracing-opentelemetry/actions/runs/8055902534/job/22003813184?pr=100#step:5:399-415

Cow imported twice.

[djc]

Yeah, was looking at this last night but it's not actually clear to me where the second (or rather, the first) import is.

[valkum]

Yeah, was looking at this last night but it's not actually clear to me where the second (or rather, the first) import is.

A borrow::Cow was recently added by a PR to layer.rs. Wondering if this is some rebase GitHub thing. The CI used 8b31f0a which is not the HEAD of your branch.

[djc]

Yeah, was looking at this last night but it's not actually clear to me where the second (or rather, the first) import is.

A borrow::Cow was recently added by a PR to layer.rs. Wondering if this is some rebase GitHub thing. The CI used 8b31f0a which is not the HEAD of your branch.

Ahh, that explains it, thanks!

[vriesk]

Hi! Can this be merged now?

[djc]

(@jtescher also, I would be happy to help maintaining this if you're running short on time/energy.)

[jtescher]

@djc thanks! I don't have admin on this repo, could get added in discord maybe? but I made you an owner on crates.io 👍