Re: Software Engineering News Brief

["Norman H. Cohen" <ncohen@watson.ibm.com>]

Paul Eggert wrote (about the range of years 1901 to 2099):

> Actually, many kinds of applications require support for dates outside
> that range; it was shortsighted of the Ada standard-writers to prohibit
> implementations from having better standard date support than that.

But if some implementations have better "standard date support" than
others, it's no longer "standard", is it?  

The predefined Calendar package has a type for representing times and
dates in the range 1901 to 2099, but an implementation can provide a
nonstandard package that supports a different range, as long as the
implementation ALSO provides the standard package.  Similarly, a
programmer can easily write hiw own package supporting a different
range.

The intended use of the package Calendar, reflected in its design, is
the manipulation of dates and times during and near to the execution of
the program, most obviously to query the current time or to establish
the time at which some action should be performed.  (Ada programs that
use package Calendar construct dates prior to the advent of Ada are
quite rare.  The example of such a program on page 88 of Ada as a Second
Language is intended to be facetious, as is made clear by the text
introducing it.)  The predefined package Calendar was not designed for
calendrical calculations such as determining the day of the week on
which November 11 occurred in the year 1376, and does not provide
facilities that would be particularly helpful in making such
calculations.  If you were writing an application that required such
calculations, you would probably write your own, quite different,
package.

-- 
Norman H. Cohen
mailto:ncohen@watson.ibm.com
http://www.research.ibm.com/people/n/ncohen