forums.silverfrost.com

[mecej4]

Here is a short function that calculates the day of the week in the Gregorian (prolated, if appropriate) calendar. The returned value: (1=Sunday,7=Saturday); The inputs: the year (positive integer), month (1 to 12) and the day of the month (1 to 31, user to check for valid value before invoking function)

[simon]

You may need to add a check for the year. A positive integer is not going to give correct results for years before the transition from the Julian to the Gregorian calendar, and when that transition occurred depends on which country you are in, so it can get a real mess! however, I suspect that for most applications the year of interest will be sufficiently recent.

[mecej4]

Thanks for your cautionary notes, they are quite appropriate. George Washington has two birthdays for such reasons, and I even read that a Dutch military unit won a battle before the date marked on their written orders, which reached them many days after the battle was over.

Another potential source of error/confusion is the matter of whether the first day of a week is Sunday (as in the US, UK, etc.), Monday (EU countries, ISO), Saturday (many Islamic countries) or some other day.

[mecej4]

Here is a list of years in which May 31 occurs on Sunday. This list may be used to create a test program for verifying the DOW(y,m,d) function.
1970, 1981, 1987, 1992, 1998, 2009, 2015, 2020, 2026, 2037

[Kenneth_Smith]

Here is a function to do the required check before calling DOW.

[mecej4]

Thanks Kenneth, for providing your code for checking the input arguments to DOW() .

[LitusSaxonicum]

Aren't there routines for doing this in the FTN77 library?

[mecej4]

[mecej4]

Here is a little puzzle that can be solved using a DOW() function.

Find years in the range 1901 to 2099 for which:

(i) the weekdays of January 1 and December 31 are the same, and
(ii) December 13 is a Sunday.

Verify, for the years found, that the 13th days of February, March and November are Fridays.

Hint: The year 2026 is one such year .

[Kenneth_Smith]

In the DOW() function, in the line:

[mecej4]

We could also use the MODULO() intrinsic function instead of MOD():
dow = modulo(v,7)+1

This may not suit users of Fortran 77.
I, too, have found myself making errors with code involving signed integer values passed as arguments to MOD(). Not enough practice, I suppose!

[simon]

[mecej4]

See https://www.nottingham.ac.uk/manuscriptsandspecialcollections/researchguidance/datingdocuments/juliangregorian.aspx

where you can even see an image of part of the orders and text with a partial explanation. The same page also contains another example involving France and Britain:

"What this meant in practice was that a reply written in Britain to a letter sent from France could apparently be dated from before the original was sent!"

[arctica]

This is an interesting problem that gets more complicated pre-1582. The code works for 1582 or later, but falls over for earlier dates. From what I found, Sunday was the first day of the week from 321CE after Emperor Constantine.

[PaulLaidler]

The Gregorian calandar started in 1582.