* rr record
* rr replay until crash or exception (set up catchpoint for that)
* set up hardware watchpoint for borked data that caused the issue
* work backwards in the data flow through watchpoints and reverse-continue
I would say this fundamentally changed how I approach debugging.
