Talk:Last Friday of each month

From Rosetta Code
Revision as of 17:34, 11 August 2012 by Walterpachl (talk | contribs) (→‎1582: proletic should be proleptic?!?)

command line

Why does the task specify that the year is supplied from the command line? This seems to conflate two totally separate language capabilities--date arithmetic and command line handling. For the PHP example, I'm just ignoring the command line requirement. --Showell 17:37, 20 January 2012 (UTC)

task name

of year? not of month?--eMBee 13:43, 7 November 2011 (UTC)

It looks like it's all of the last fridays of the months of a given year? --Rdm 14:52, 7 November 2011 (UTC)
yes, exactly, which is why i find the current title confusing. something like Last Friday of every month would be more to the point. otherwise, i thought the original title was sufficient. something about fridays. it is clear enough that this is about the calendar. is more details really needed?--eMBee 14:59, 7 November 2011 (UTC)
English is not my born language, but I also find the title "Last Fridays of year" confusing. I understand it as 'really' the last fridays of a given year, for example the last three Fridays of 2012 are 2012-12-14, 2012-12-21 and 2012-12-28. Blue Prawn 21:10, 9 November 2011 (UTC)
exactly my thoughts too. so what's a better title? "last friday of each month?" (i thin kthe fact that it is for a year is less interesting, the same solution would work for any period of time).--eMBee 04:11, 10 November 2011 (UTC)
I'm not particularly attached to the current name; I just thought that the original (“Last Fridays”) wasn't good enough either. Suggested changes are welcome. –Donal Fellows 08:49, 10 November 2011 (UTC)

C version

This discussion is about the program shown here:

<lang c>#define _XOPEN_SOURCE

  1. include <stdio.h>
  2. include <time.h>

int main(int c, char *v[]) { int days[] = {31,29,31,30,31,30,31,31,30,31,30,31}; int m, y, w; struct tm tm; char buf[32];

if (c < 2 || !sscanf(v[1], "%d", &y)) return 1;

days[1] -= y % 4 || (y % 100 && ! (y % 400)); sprintf(buf, "%d-1-1", y); strptime(buf, "%Y-%m-%d", &tm); w = tm.tm_wday - 1; /* day of week for zeroth of Jan */

for(m = 0; m < 12; m++) { w = (w + days[m]) % 7; printf("%d-%02d-%d\n", y, m + 1, days[m] + (w < 5 ? -2 : 5) - w); }

return 0;

}</lang>

Building the C version under cygwin (gcc 3.4.4), I get:

<lang bash>$ make last_fridays cc last_fridays.c -o last_fridays

$ ./last_fridays 2011 2011-01-27 2011-02-24 2011-03-31 2011-04-28 2011-05-26 2011-06-30 2011-07-28 2011-08-25 2011-09-29 2011-10-27 2011-11-24 2011-12-29

$ ./last_fridays 2012 2012-01-27 2012-02-24 2012-03-30 2012-04-27 2012-05-25 2012-06-29 2012-07-27 2012-08-31 2012-09-28 2012-10-26 2012-11-30 2012-12-28

$ cc --version cc (GCC) 3.4.4 (cygming special, gdc 0.12, using dmd 0.125) Copyright (C) 2004 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.</lang>

The C entry currently shows different results for 2011, but I cannot reproduce them. --Rdm 23:25, 8 November 2011 (UTC)

I'll assume it's a cygwin bug for now. For the record, can you change two lines to
<lang>sprintf(buf, "%d-01-01 09:01:01", y);

strptime(buf, "%Y-%m-%d %H:%M:%S", &tm);</lang> and see how it does? --unsigned

I replaced the sprintf an strptime lines with these two lines and got the same result. Note also that might results are consistent (but this does not prove that the issue is not uninitialized memory). --Rdm 11:17, 9 November 2011 (UTC)
I have a similar problem with OpenBSD 4.9. The output changes from run to run, and might show the last Mondays, Thursdays, Fridays or Saturdays. After the call to strptime(), tm.tm_wday contains junk values like 50237728 or 239567136. (Legal values are 0 to 6.) Change to "%d-01-01 09:01:01" and "%Y-%m-%d %H:%M:%S" is no help. Checking my manuals, mktime(3) describes the fields of struct tm, but strptime(3) is vague; I assume that strptime() never fills tm_wday unless the string contains the weekday. --Kernigh 01:05, 9 November 2011 (UTC)

1582

4 Oct 1582 is followed by 15 Oct 1582 in the Gregorian Calendar. Java seems to take care of that. Rexx results differ:

25.09.1582
26.09.1582
27.09.1582
28.09.1582 Fri
29.09.1582
30.09.1582
01.10.1582
02.10.1582
03.10.1582
04.10.1582
15.10.1582 Fri
16.10.1582
17.10.1582
18.10.1582
19.10.1582
20.10.1582
21.10.1582
22.10.1582 Fri
23.10.1582
24.10.1582
25.10.1582
26.10.1582
27.10.1582
28.10.1582
29.10.1582 Fri
30.10.1582
31.10.1582
Java       Rexx
Jän 26     Friday 29 Jan 1582
Feb 23     Friday 26 Feb 1582
Mär 30     Friday 26 Mar 1582
Apr 27     Friday 30 Apr 1582
Mai 25     Friday 28 May 1582
Jun 29     Friday 25 Jun 1582
Jul 27     Friday 30 Jul 1582
Aug 31     Friday 27 Aug 1582
Sep 28     Friday 24 Sep 1582  <--- different
Okt 29     Friday 29 Oct 1582
Nov 26     Friday 26 Nov 1582
Dez 31     Friday 31 Dec 1582

--Walterpachl 06:05, 11 August 2012 (UTC)


No, there are no missing days in the Gregorian calendar.

October 4th, 1582, in the Gregorian calendar is followed by October 5th in the Gregorian Calendar.

October 4th, 1582, in the Julian calendar is followed by October 15th in the Gregorian calendar.

October 4th, 1582, in the Julian calendar is followed by October 5th in the Julian calendar for those countries (and places) that didn't convert to the Gregorian calendar.

October 4th, 1582, in the Julian calendar is followed by October 5th in the Julian calendar, even though it wasn't the "legal" (in use) calendar anymore.

It's like when someone leaves San Francisco, California and flies to New York, and adjusts (or adopts) their wristwatch to local time by jumping three hours ahead. There's no missing three hours. The traveler just adopted the "time" of a different time zone.

See: http://rosettacode.org/wiki/Talk:Calendar

The only "missing days" are when one switches from one calendar to another. The Gregorian calendar is proletic. October 5th follows 4 Oct in EVERY year in the Gregorian calendar (starting in January 1, year 1). -- Gerard Schildberger 06:18, 11 August 2012 (UTC)

October 4th in year 1582 was the date in the Julian calender (the day before the switchover in some countries). The next day, the Gregorian calendar was adopted (or put into effect) in various countries, but not everywhere. The new Gregorian calender says it was October 15th. If we switched to a Mayan calendar, would we be missing years (or whatever)? Of course not. Once the Gregorian calendar was adopted, it was like the calendar was in effect all along, with a continous calendar (proletic) backward to January 1st, year 1. No missing days. That is why there are references to O.S. (old style) for years previous to the adoption of the new Gregorian calendar for those people who were born before the switchover. If the Gregorian calendar wasn't proletic, there would be no need for O.S. type of dates.

See: http://en.wikipedia.org/wiki/George_Washington

-- Gerard Schildberger 07:02, 11 August 2012 (UTC)

See comments above, see comments above. -- Gerard Schildberger 07:13, 11 August 2012 (UTC)

One learns something every other day. Thanks!
Does that mean that Java is wrong here?--Walterpachl 17:16, 11 August 2012 (UTC)
proletic should be proleptic?!? --Walterpachl 17:34, 11 August 2012 (UTC)