The twist I’m exploring is to keep debug logs buffered during normal execution and only release the relevant context when an error occurs. So you get the diagnostic value of DEBUG without continuously storing all the noise.
I built a small Go library to experiment with this approach, but I’m particularly interested in where the model breaks down in real-world systems.
The twist I’m exploring is to keep debug logs buffered during normal execution and only release the relevant context when an error occurs. So you get the diagnostic value of DEBUG without continuously storing all the noise.
I built a small Go library to experiment with this approach, but I’m particularly interested in where the model breaks down in real-world systems.