Glenn Cole ([log in to unmask]) wrote:
: All this time/hardware clock/software clock discussion is giving
: me a headache.  Do you realize how many ways one can get the date
: in a COBOL program (most of which apply to other languages as well)?
: Let's give it a whirl:
:         1. CURRENT-DATE constant
:         2. CURRENT-DATE function (POST85)
:         3. CALENDAR   intrinsic (not recommended ;)
:         4. HPCALENDAR intrinsic (5.5 pp4)
:         4. DATELINE   intrinsic (can't manipulate results easily)
:         5. ALMANAC    intrinsic (includes DOW, which is nice)
: Is there some "rule of thumb" that says when it is advisable
: to use each?  I'll try a first pass:

I'll respond from a COBOL perspective:

: 1. CURRENT-DATE constant
:         - COBOL standard
:         - simplest to use
:         - no allowance for multiple time zones
:         - provides only 2-digit year

Not exactly a constant, it's what COBOL calls a special register.
It is not in the language standard, but is an extension supported
by a number of vendors, including HP.  I don't recommend it.

: 2. CURRENT-DATE function
:         - COBOL standard
:         - more complex than CURRENT-DATE constant
:         - supports multiple time zones (set TZ!)
:         - provides 4-digit year

This is the way I recommend to get the current date in COBOL.
It is the only ANSI-standard method that provides a 4-digit
year.  Yes, the TZ variable needs to be set correctly; consider
doing this in a system-wide logon UDC.  And the system time
zone offset has to be right:  check it with the SHOWCLOCK command,
change it with the CLKUTIL utility (safest) or the SETCLOCK command
(if you understand it).

: 3. CALENDAR intrinsic
: 4. HPCALENDAR intrinsic
: 5. DATELINE intrinsic
: 6. ALMANAC intrinsic

You didn't mention the ACCEPT statement (ACCEPT ... FROM DATE).  It's
standard COBOL, but provides only a 2-digit year.

Walter Murray
Hewlett-Packard
Support Technology Lab
COBOL II/iX Project