I once spent quite a long time tracking down a bug that would only occur if the time between 10:00 and 12:59 because of the extra character that occurred in the date formatting, leading to an unexpected line wrap that led to a lookup bug later on during a session.
If I'd known that time of day was the variable causing the behavior change, it would have been fairly easy to resolve—but that wasn't the case. Instead it was only resolved by the large-scale collection of failure and success cases coupled with the careful correlation of bytes sent & received across an entire session.
If I'd known that time of day was the variable causing the behavior change, it would have been fairly easy to resolve—but that wasn't the case. Instead it was only resolved by the large-scale collection of failure and success cases coupled with the careful correlation of bytes sent & received across an entire session.