>>Usually the best indication for me that it's time to stop a long day is when I begin breaking stuff that worked before.
>>
>>>Hi everybody,
>>>
>>>Just curious if I beat the record today?
>>>
>>>Guess it's time to go home - nothing works anyway...
>
>Good point. Yesterday sometime to close to the end the project what worked stopped working at all on a separate unrelated file I didn't even touch.

It's about 13 years this week since my 2nd (and last) all-nighter, when I worked more or less through the day, starting at 11, taking a lunch, taking a nap, taking a walk and a beer, and then getting to understand what was it that I wanted to do. My mate left around 4AM, and I was there in the morning to see who's first to come to the office (the newspaper, then one of the HW guys, then the secretary). Effective keyboard time for that day must have been around 16 hours. The only justification for that is that the piece I wrote that night was then used almost everywhere, as an important piece of framework (browse with incremental search simulated with arrays and @say, with window region scrolling in FPD2.6 - fast as a snake). Because the next day I wasn't able to do much, my head buzzing with all the coffees I had, and my total time for that week wasn't much different from 40 hours.