Y2K38 problem has been described as a non-problem, given that we are expected
to be running 64-bit operating systems well before 2038. Well, maybe.
Just as Y2K problems arise from programs not allocating enough
digits to the year, Y2K38 problems arise from programs not allocating enough bits
to internal time.Unix internal time is commonly stored in a data structure using
a long int containing the number of seconds since 1970. This time is used in all
time-related processes such as scheduling, file timestamps, etc. In a 32-bit machine,
this value is sufficient to store time up to 18-jan-2038. After this date, 32-bit
clocks will overflow and return erroneous values such as 32-dec-1969 or 13-dec-1901.Machines
Affected Currently (March 1998) there are a huge number of machines affected.
Most of these will be scrapped before 2038. However, it is possible that some
machines going into service now may still be operating in 2038. These may include
process control computers, space probe computers, embedded systems in traffic
light controllers, navigation systems etc. etc. Many of these systems may not
be upgradeable. For instance, Ferranti Argus computers survived in service longer
than anyone expected; long enough to present serious maintenance problems.
Unix time is safe for the indefinite future for referring to future events, provided
that enough bits are allocated. Programs or databases with a fixed field width
should probably allocate at least 48 bits to storing time values. Hardware, such
as clock circuits, which has adopted the Unix time convention, may also be affected
if 32-bit registers are used.In
my opinion, the Y2K38 threat is more likely to result in aircraft falling from
the sky, glitches in life-support systems, and nuclear power plant meltdown than
the Y2K threat, which is more likely to disrupt inventory control, credit card
payments, pension plans etc.