Tom

I've found a better option. Tokio tracing already has a Jaeger backend through open telemetry and its way easier to write

Oh just saw this. I already built it, it's pretty much just a curl with some bash commands then you set the environment variables. I'm having some issues with curl...

Just thought I'd post this here just incase anyone would like to use it. I will try and make a PR that integrates it into the build.rs aswell https://crates.io/crates/moltenvk_deps

I wonder if we can try and do this in two steps. 1. move the database as is. 2. do refactors. Because it will be a lot cleaner to review...

Maybe this would be worth a notion document? Also we could try out [github discussions](https://github.com/FuelLabs/fuel-core/discussions) although I've never actually used that. I have personally found using mermaid js in a...

One thing that I think we should really keep in mind during this design though is the fundamental limit that anything we do will only be tied to a key...

looks like an issue with coreaudio, what system is this on?

Here you can see two peaks where the flags are. They are around 90% cpu but only last about 18ms. This time period is 10s 

This is the responsible function: 9.00 ms 100.0% 7.00 ms `core::ops::function::impls::_$LT$impl$u20$core..ops..function..Fn$LT$A$GT$$u20$for$u20$$RF$$u27$a$u20$F$GT$::call::h1b01f83aac3f9eb9` It's just called by this: `9.00 ms 100.0% 0 s _$LT$nannou..audio..requester..Requester$LT$S$GT$$GT$::fill_buffer::h837708360a31ea69 ` So it's something in the audio thread...

Right now I'm slicing the function up into sections and measuring variance between the _average length_ that section runs and the _longest runs_. I'm also resetting every 1000 calls. So...