We're probably far too late in the development cycle for it to be worth it now, but I now know that, as well as our little log! macro as done, we should have used the tracing crate instead. It's very well suited for what we're doing, with ways to output structured logs with nested contexts, much like we're doing ad-hoc now, but with a much cleaner style, better functionality, etc.. In particular, debugging would have been a lot easier if we could just sprinkle events everywhere at various logging levels, and turned up the verbosity outside of large experiments.
If there's any indication others want to use this for reasons other than reproducing our results, it would be worth it to make this change for maintainability going forward.
We're probably far too late in the development cycle for it to be worth it now, but I now know that, as well as our little `log!` macro as done, we should have used the [tracing crate](https://docs.rs/tracing/latest/tracing/) instead. It's very well suited for what we're doing, with ways to output structured logs with nested contexts, much like we're doing ad-hoc now, but with a much cleaner style, better functionality, etc.. In particular, debugging would have been a lot easier if we could just sprinkle events everywhere at various logging levels, and turned up the verbosity outside of large experiments.
If there's any indication others want to use this for reasons other than reproducing our results, it would be worth it to make this change for maintainability going forward.
We're probably far too late in the development cycle for it to be worth it now, but I now know that, as well as our little
log!macro as done, we should have used the tracing crate instead. It's very well suited for what we're doing, with ways to output structured logs with nested contexts, much like we're doing ad-hoc now, but with a much cleaner style, better functionality, etc.. In particular, debugging would have been a lot easier if we could just sprinkle events everywhere at various logging levels, and turned up the verbosity outside of large experiments.If there's any indication others want to use this for reasons other than reproducing our results, it would be worth it to make this change for maintainability going forward.