systematic-debugging

Intermittent bugs usually indicate race conditions or timing issues, such as missing await statements or unordered async callbacks. First write a deterministic test that reproduces the failure reliably, then probe with targeted logging to confirm the hypothesis before changing code.

Use git bisect to isolate the faulty commit in O(log N) steps. Mark the current broken state with git bisect bad, mark a known-good release with git bisect good, then test each midpoint commit Git checks out until the exact offending commit is found.

This pattern indicates state leakage, typically caused by global variables, cached HTTP clients, or missing cleanup in component unmount logic. Inspect shared mutable state between operations and add logging probes to confirm which state persists incorrectly.

Start with the top line for the fatal error, then scroll past framework and node_modules frames to the first function you wrote. That line shows where the failure surfaced; trace upward through the call stack to find why invalid data was passed in.

Silent failures usually come from empty catch blocks, unhandled promise rejections, or conditional rendering that hides missing data. Audit catch handlers and promise chains, and add explicit logging at each stage of the failing operation.