Joel Ewing's responses to my strictures were, as I expected them to be, wholly
predictable in character.
Dates, even those in the comparatively respectable FIPs- and ANSI-compatible
Gregorian descending character-string, yyyymmdd, format inevitably get used;
and when they are used ad hoc disaster follows with the same inevitability.
Such questions as
o what is the day of the week of the date yyyymmdd?
o what is the date YYYYMMDD ten business days following the date yyyymmdd?
o how many business|working days are available between YYYYMMDD and yyyymmdd?
get asked and answered very badly when a format not designed for computation is
used. (I have a rogue's gallery/black museum of COBOL and C and PL/I
routines, written by people whose knowledge of calendrical computations was
inadequate, that 1) are wrong from time to time and 2) obtain their results
using schemes that are so convoluted and bizarre as to be all but unbelievable,
unless of course one has seen probability problems approached by undergraduates
who have not yet been taught what a Jacobian is.)
For reasons that elude me programmers who would never attempt to write their
own cosh or present-value subroutine have the chutzpah to do calendrical
computations badly.
Some of what Mr Ewing says is simply wrong. It does not really matter what
epoch origin is chosen provided trhat an organization chooses and adheres to
one. Conversions are then trivial. Or again, the argument that a DSN having
an unknown epoch origin is opaque while a yyyymmdd value is not is simply naif.
Incompletely or inadequately documented routines are, of course, problematic;
but they are problematic in a generic way; and the remedy for them is better
documentation not adherence to obsolete internal date formats.
The Y2k problem provided an opportunity to solve problems that was shirked.
Date-processing problems that should have been addressed with respectable
technology were instead papered over with dubious windowing schemes and the
like. What is past is prologue, and I know of no organization that did
significant Y2k remediation that is not now living with the disagreeable
consequences of having done it badly.
But enough! I have said what I wished to say about this problem.
John Gilmore Ashland, MA 01721-1817 USA
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
