what a thread! Far from buttons, dropshadows and frames counting.
And who is involved? Craig. No comment.
Last time I read about such problems was around 2000.
Was, besides the 'y2k-problem' also connected to file systems (especially JHFS+).
I had a source explaining this popular in large. Can't find it just now.
Anyway, a good summary is here, explaining the jump to negative numbers from
an integer overflow.
Now it's hard to solve Craig's 32535244799 problem.
At first we have
log2(1+2*32535244799) = 35.9 , so 2^35 < 1+2*32535244799 < 2^36.
Could be 2^15 seconds and a 10^6 subdivision (microseconds) what requires 20 bits,
That is the Microseconds of 2^16 seconds have dimension above 35 and below 36 bit.
[2^15*2^20 =] 2^35 < 1+2*32535244799 < 2^36 [= 2^16*2^20]
And 2^15 is clearly derived from the old 'Apple-paradise joke':
32 bit = plus-minus 15 bit plus 1 bit for the sign plus one bit for (the) apple.