Re: date overflow in year 2038

[Dr. David Alan Gilbert]

* Eric Blake (address@hidden) wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> According to David M. Dowdle on 2/13/2009 8:45 PM:
> > clouded:~> date -d "Fri Jan 19 03:14:08 UTC 2038" +%s
> > date: invalid date `Fri Jan 19 03:14:08 UTC 2038'
> > clouded:~>
> > 
> > 03:14:07 is apparently when 32bit time_t hits MAXINT
> 
> Yep, and that's why many newer systems are switching to 64-bit time_t.
> But changing the size of size_t is an ABI difference, so not one that you
> can easily port to older kernels.
> 
> > I'd rank this as low priority, but people doing things like 30 year
> > mortages will be hitting this already.
> 
> Not a bug in coreutils, but an inherent limitation in your kernel.  Just
> like the old limitation that you couldn't have a file larger than 2GB
> until you upgraded to a kernel with 64-bit off_t support.

Erm really? Sure the kernel might keep the date in 32bit but he wasn't
trying to read or set system time - so the kernels view of time is
mostly irrelevant.  I'm guessing coreutils uses a lot of libc for
it, and I suspect that is where there maybe the limitation.

(I've seen some systems that really get upset if their time
is set to after 2038 - so please be careful if you are trying
that).

Dave
-- 
 -----Open up your eyes, open up your mind, open up your code -------   
/ Dr. David Alan Gilbert    | Running GNU/Linux on Alpha,68K| Happy  \ 
\ gro.gilbert @ treblig.org | MIPS,x86,ARM,SPARC,PPC & HPPA | In Hex /
 \ _________________________|_____ http://www.treblig.org   |_______/