On 16/01/14 12:38, Tom Van Baak wrote:
The Multics clock design (a fixed bin (71), ie double word, representing
microseconds
since 00:00 01-01-1900) clearly informs the Unix one.
Was it 1900 or 1901? See:
http://www.multicians.org/jhs-clock.html
http://web.mit.edu/multics-history/source/Multics/ldd/bos/include/rdclock.incl.alm
http://web.mit.edu/multics-history/source/Multics/doc/info_segments/datebin_.info
http://web.mit.edu/multics-history/source/Multics/doc/info_segments/date_time_.info
"If clock is not a valid date, "01/01/01 0000.0 gmt Tue" is returned."
I spent years programming on GE-635 (GCOS/TSS). Multics used the GE-645 which
had a special external microsecond clock:
http://www.multicians.org/mgc.html#clock
Yes, the Multics clock is very much like the one UNIX adopted. GCOS used a more
traditional date and time format: 36-bits for date (mmddyy in BCD) and 36-bits for
time-of-day (in 15.625 us = 1/64 millisecond units). GCOS had a weird convention of
printing all times in hours with decimal places (no minutes or seconds). That was one of
the things that I liked about UNIX; it used hh:mm:ss notation. The other thing to note is
"Clock readings are Multics Greenwich mean time (GMT)". UNIX also adopted this.
Most OS's back then, including GCOS, ran in local time.
When I developed email in 1976 I encoded the BCD date (mmddyy) and BCD time
(hhmmss) into two 18-bit binary fields. This worked because the maximum
possible date was 123199, the maximum time was 235959, which just fit in the
maximum half-word (2^18 = 262144).
OK, so Y2K problems where all your fault. BUSTED! :D
235960 would actually be possible to encode there, as well as all local
time variants.
Cheers,
Magnus
_______________________________________________
LEAPSECS mailing list
[email protected]
http://six.pairlist.net/mailman/listinfo/leapsecs