Comments on: Year 2038, another Y2k?? http://www.fwdemails.com/2007/08/10/year-2038-another-y2k/ A collection of funny, cool & intersting forwarded emails Mon, 23 Jan 2012 23:47:42 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: Semi http://www.fwdemails.com/2007/08/10/year-2038-another-y2k/comment-page-1/#comment-5134 Semi Fri, 08 May 2009 13:20:17 +0000 http://fwdemails.com/2007/08/10/year-2038-another-y2k/#comment-5134 Ah, yes, that old chestnut... don't worry, everyone knows about it (http://en.wikipedia.org/wiki/Year_2038_problem) and there is already a move towards using 64-bit time_t in future. ... and btw, you are completely wrong about the signed integer representation. The first bit is not normally a sign bit at all, a signed integer is normally stored in two's compliment allowing add and sub instructions to work the same on both signed and unsigned values. I think the real problem is that real-world implementations of time_t are not consistent with what they do about negative time values, hence the last bit can not be used reliably. Interesting reminder though. Ah, yes, that old chestnut… don’t worry, everyone knows about it (http://en.wikipedia.org/wiki/Year_2038_problem) and there is already a move towards using 64-bit time_t in future.

… and btw, you are completely wrong about the signed integer representation. The first bit is not normally a sign bit at all, a signed integer is normally stored in two’s compliment allowing add and sub instructions to work the same on both signed and unsigned values. I think the real problem is that real-world implementations of time_t are not consistent with what they do about negative time values, hence the last bit can not be used reliably.

Interesting reminder though.

]]>