Re: [h-e-w] Almost a bug :)

[Benjamin Riefenstahl]

Hi Karel, Richard,


"Sprenger, Karel" <address@hidden> writes:
> In my diary file I have the following special dates (which I picked
> up from somewhere during the millenium shuffle):

;-)

Some trivia here:

> Jan 19, 2038 03:14:07 End-of-time for 32-bit machines (milliseconds
> since January 1, 1970 overflows 4 bytes)

Should be "seconds since January 1, 1970 overflows 4 byte sigend
integer".

> Feb 06, 2040 06:28:15 End-of-time for old Macintoshes (seconds since January 
> 1, 1904 overflows 4 bytes)

More exactly "seconds since January 1, 1904 overflows 4 byte unsigned
integer".

> Feb 06, 2106 06:28:15 End-of-time for 32-bit machines (seconds since
> January 1, 1970 overflows 4 bytes)

Probably "seconds since January 1, 1970 overflows 4 byte unsigned
integer".


1970-01-01 is the start of the POSIX epoch used for time_t.  This
epoch is also used by most C RTLs these days.  According to POSIX
time_t should be signed, not unsigned, but than most C RTLs on Windows
don't claim to implement POSIX completly.

I think that 1980-01-01 is the start of the BIOS clock of the IBM-PC
hardware. 

In theory Win32 can do a much larger time range, its API uses 8 byte
integers for representation and although this is counting nanoseconds
(!), I believe that still gives a range of more than 10000 years or
somthing like that ;-).  I'm not sure how much of this range
Win95/98/Me actually can cover, as it most likely needs to convert to
some other unit internally.  NT/2000/XP should be able to handle the
whole range.

> So, although some problems may arise from MS code, others are
> inherent to 32-bit chips.  I didn't consider this to be a problem
> for myself as I hope to be using 64-bit machines by then :-)

So do the POSIX proponents.  I have heard though that in banking and
insurance there actually is software out there that needs to calculate
30 years into the future, and some of those packages may be in trouble
in 2007.


so long, benny
-- 
Energis-ISION
Benjamin Riefenstahl
Harburger Schlossstrasse 1
D-21079 Hamburg

eMail: address@hidden
Web: http://www.energis-ision.com