I am using Pedestal.log for all logging in my newest project. I was just about to start investing in Metrics and Tracing using Pedestal.log, but the article says that these are deprecated. I’ll hold off on investigating / implementing these then.

For what it’s worth, I think it is really cool that the entire io.pedestal.log namespace is so small but has a fully usable implementation of logging, metrics and events. I am looking for ways to generate open-telemetry compliant metrics and traces from my app and hopefully io.pedestal.log would be the way to go. If there is a publicly available roadmap or plan for what changes need to be made, I’m happy to help with testing and contributions.

I'm intending to add a new library, io.pedestal/metrics to handle metrics and telemetry using open telemetry.

The current version is too closely tied to a specific metrics library.

And a .log namespace isn't the best place for metrics/telemetry!

okay, cool! I’m happy to try io.pedestal/metrics as soon as it is available, even development versions if they are available now.

Hey please consider this one: https://github.com/pedestal/pedestal/issues/666 This issue makes every pedestal log a edn string http://pedestal.io/pedestal/0.7-pre/reference/logging.html#_events_not_strings

So io.pedestal.log is a standalone jar, which could be used in place of clojure.tools.logging if you want MDC? Is that use encouraged, or is its existence as a distinct jar not necessarily a long-term design? (Is one invited to :exclude the dropwizard and opentracing baggage if using io.pedestal.log as a clojure.tools.logging substitute?)

Well, it's a bit more complicated than that, unfortunately. The pedestal.log library is stand-alone, but has the existing metrics and open-tracing dependencies and some of those types are exposed in the io.pedestal.log namespace which means they can't be easily removed without breaking changes.

It may come down to comments in the namespace that say "public API starts here" to determine what's a breaking change, and what's not. Really, a lesser-of-two-evils approach.