To FootThere is an excellent article "Delphi for Time Travellers" by Brandon Smith in the June 1998 issue of The Delphi Magazine (pub. iTec, UK) issue 34 pp.24 ff., which refers to my work "misctime", now in Date Miscellany I, programs/, etc. Since the article was written, there have been changes on my Web site; there is one technically significant typo and there are things which I would myself have put a bit differently. This, therefore, is general comment on the article.
Where the article indicates that my code was hard to follow, I have added comment to my program files.
P 24 c 1 : Correct, but must be read carefully : the 4000-year rule is (AIUI) in the old Britannica, but it is not in a modern one, is not official and is not Gregorian.
P 25 box : JDN 2,400,000 starts at noon GMT 1858/11/16.
P 25 box : For the day of the week, the number should be not 999,990 but 999,999,990 - I now use 987,654,321 (it's the same, modulo 7) for safety. Weekdays are Sun=0 to Sat=6 here. File basedate.pas, from the floppy disk, has 999999990 correctly.
P 26 c 1 : Not at Nice, but at Nicaea, in Bithynia, which is in the north-west of Asia Minor.
P 26 c 3 : It says "... not valid for dates prior to 15 October 1582"; this depends on one's chosen calendar, of course; few countries actually changed calendar in October 1582.
P 26 L 2 : Near the end, the ≤1752 code could be executed unconditionally, with the result amended if >1752 - less code.
P 28 c 1 : It refers to my Pascal program MJD_DATE; much of the code cited has now been moved to the unit DATEPROX.
P 30 c 3 : "... puzzled ... month array started with March ..." : Shifting Jan & Feb into the previous year means that the Leap Day comes (or not) only at the end of the calculational year; this is a simplification.
P 31 c 1 : "... type real ..." - I don't know about Delphi, but in Pascal type real is slow in comparison with IEEE types.
P 32 box 5 : DATEPROX can now do the irregular Leap years; and has faster routines for Gregorian only CMJD<->YMD.