Replying to a message of Dan Ceppa to Bob Ackley:

 DC> -> On 18 Jul 09  03:16:48, Bob Ackley got back to Dan Ceppa 
 DC> -> Re: Y2K Bug in OLR

 DC>> I think the year was 2038, or something like that.  BW's problem was 
 DC>> it held only 2 bits for the date.

 BA>> Two bits??  With two bits you can count all the way up to three
 BA>> (beginning at zero).  With one byte (8 bits) you can count up to 255
 BA>> and with two bytes (16 bits) you can count up to 65535 (both cases
 BA>> starting with zero). 

 DC> Whatever it was, the field for the date in the programming was set up 
 DC> to accept 2 digits for the year in thw data record.  

That's probably two bytes, then.  If it's stored as ASCII rather than binary then
two digits is all you get.  Mainframes had that problem in that a lot of data (on
IBM systems) used a 'packed decimal' format (two EBCDIC numeric characters in
one byte) rather than binary, and only allowed two digits (one byte) for storing a
year value.  *That* was the 'Y2K' problem that the mainframe world feared.

At about 2330 on December 31, 1999, some top management people (not including
my boss) started wandering into the company's computer room where I was running
the usual overnight production.  Midnight came and went, and the IBM mainframe
(which I now own) didn't even hiccup.  By 0015 they'd all left.  I still wonder what
they thought they could have done if the thing had thrown up at midnight, none were
systems programmers (in fact at the time the company didn't *have* a systems
programmer, he'd resigned earlier that year to take a position with another company
and was never replaced), and we didn't have an IBM service contract.

--- FleetStreet 1.19+