32-bit 2038 issue

[Tom Peterson (USG)]

Has anyone considered what to do about time zone files and the 32-bit 2038 
issue?  This comment from zic.c sums up the problem:

> /*
> ** The tz file format currently allows at most 32-bit quantities.
> ** This restriction should be removed before signed 32-bit values
> ** wrap around in 2038, but unfortunately this will require a
> ** change to the tz file format.
> */

With Y2K bringing attention to time-related problems, there's an increasing 
awareness/concern about the 2038 issue as well.  People know it's not just a 
system clock setting issue.  The limitations affect projected dates used by 
applications as well.  Problems are already cropping up where financial 
institutions can't generate amortization data for 40, 50, etc year loans and 
the like.  I expect there are other practical examples as well.

As new 64-bit UNIX platforms are emerging, there's a trend toward 64-bit 
time_t's.  While this certainly eliminates the 2038 limit, it also creates a 
problem for time zone data files.  A time zone file filled with data spanning 
64-bits of seconds-based time would be huge, not to mention the performance 
problems of accessing it.  Some ideas for addressing this problem are:

1) Expand the date/time boundaries for time zone files to a specific range.  
This would have to balance file size vs anticipated application needs.  
Perhaps an upper limit corresponding to the year 2100 or 2200 would suffice?  
Regardless of the range set, this approach only delays the problem.

2) Find some way to mimic what the zic compiler does at run-time.  This might 
involve generating some kind of algorithmic data for each time zone which can 
be used to calculate future transition times, etc on-the-fly.  I'm not sure 
this is possible or practical given the varied complexity of time zone rules.

3) Other ideas???

Also, with 64-bit time_t's, the tm struct becomes the limiting factor.  
Boundary checks now become an issue for localtime()/gmtime() instead of 
mktime().  This is a problem for standards at a mininum.  While mktime() has 
the (somewhat ambiguous) -1 return value for non-representable times, no such 
provision is defined for localtime()/gmtime().

Any ideas or precedent for addressing this issue?

thanks,
- Tom

=====================================================================
Tom Peterson                   | DIGITAL UNIX Development Environment
Digital Equipment Corporation  | Phone: (603) 884-7550
110 Spit Brook Road ZKO3-2/W17 | FAX:   (603) 884-2257
Nashua, NH 03062-2698          | Email: mailto:tomp at zk3.dec.com