| Rick Smith 2006-02-10, 6:55 pm |
|
"Michael Wojcik" <mwojcik@newsguy.com> wrote in message
news:dsi68u0283g@news1.newsguy.com...
[snip]
> That is, their requirements changed retroactively; or to be more
> formally precise, they have a general requirement ("implement the
> rules of DST") that remains in effect, but a derived specific
> requirement ("DST requires we change the hour at times T1 and T2")
> which has now been superceded.
>
> In such a case, the old behavior is no longer applicable for any user
> of the component, so the primary justification for specialization
> doesn't apply.
One of the features of FUNCTION CURRENT-DATE
is the difference between local time and UTC found in
the last five characters. Because UTC is a fixed reference,
any changes in local time that result from DST must be
reflected by an adjustment in the last five characters of
CURRENT-DATE. Thus, if CURRENT-DATE is fully
supported by the COBOL runtime and DST is supported
by the OS, COBOL programs could be made immune
from changes in the transition dates for DST. Or, so it
would seem.
|