My news-service was sick from Saturday to Tuesday; the following was
probably not available to Demon users, and possibly not to anyone.
There is some new material after the repeat.
<repeat>
JRS: In article <ax*******************@twister.nyroc.rr.com>, seen in
news:comp.lang.javascript, Mick White <mw******@BOGUSrochester.rr.com>
posted at Sat, 21 Feb 2004 02:39:02 :-
According to the Safari browser the world began on "Fri Dec 13 1901
15:45:52 GMT-0500", but I need to be able to get around this limitation.
Evaluating new Date(-Math.pow(2,31)*1000)
gives me Fri Dec 13 20:45:52 UTC 1901
which is the same time.
In <URL:http://www.merlyn.demon.co.uk/critdate.htm> you can read that :
* 1901-12-13 Fri - 20:45:52 is where UNIX should go to from 2038-01-19;
but I have read that systems can show Mon Jan -17 1902,
-03:-14:-08 or 17:00:00.
So it seems that Safari may do date arithmetic using signed 32-bit
time_t, in which case it might fail to go ahead to or past
* 2038-01-19 Tue - 03:14:08 GMT
For both UNIX/C time_t and Javascript, zero is 1970-01-01 00:00:00 GMT.
See what date range
<URL:http://www.merlyn.demon.co.uk/js-clndr.htm#Ctrls> gives you; it is
programmed for 0100-01-01 to 9999-12-31.
If you can go past 2038, then for Gregorian work just add a multiple of
400 to the year number, since (ignoring Easter) the Gregorian Calendar
repeats every 400 years.
For Julian, since the calendar repeats every 28 years, you can use a
hand-calculated look-up table indexed with Year%28.
Otherwise, you can calculate day-of-week by Zeller's Congruence; see
<URL:http://www.merlyn.demon.co.uk/zeller-c.htm> and the link to any of
his papers; Gregorian and Julian routines are in
<URL:http://www.merlyn.demon.co.uk/zel-incl.js>
and visible in <URL:http://www.merlyn.demon.co.uk/js-nclds.htm>.
I am interested in dates from 1500 to 1901, as far as I can determine,
there are 14 possible calendar variations.
Year starts on Sun, Mon.....
Leap year starts on Sun, Mon..
I can label these early year "types" as a number between 0 and 13.
Let's say 2005 is type 5, and that 1655 is too. (In both years Jan 1
falls on a Saturday)
Gregorian 1655 Jan 1 was a Friday.
Julian 1655 Jan 1 was a Monday.
I'm trying to create a function that will identify the "type" of year.
function getYearType(year){
return Number // number between 0 and 13
}
Any suggestions?
Javascript will need assistance for that, since 1500 was a Leap Tear
everywhere that used the Christian Calendar, and 1700 was not Leap in
the British Empire, to which you then belonged. And also because, while
the succession of days of the week was unbroken, eleven days of 1752
were omitted for us. // Be thankful you're not Swedish.
ISTM that the simplest way will be to do something like
X = new Date(Y, 0, 64) // 64th day of year
Week = X.getDay() // 64 = Jan 1 + 9 weeks
Leap = 5 - X.getDate()
Number = Leap*7 + Week
but that uses the Proleptic Gregorian Calendar. For the Julian you'll
need to implement suitable routines.
The following may well set a Date Object from an ISO-8601-style Julian-
Calendar Greenwich date string :
function JCDtoDObj(A) { var Yr, Y, D // A = [Y, M, D]
Yr = +A[0] ; Y = Yr%4 ; D = (Yr-Y)/4 * 1461
return new Date(Date.UTC(204+Y, A[1]-1, A[2]-74511+D)) }
function DObjToJCD(DOb) { return '?' } // may be done now, needs test
function JulDatTry() { var DObj, F = document.forms.Frm8
F.GD.value = DObj = JCDtoDObj(F.XX.value.split(/\D+/)) // OK so far
F.QQ.value = DObjToJCD(DObj) }
Testable at foot of <URL:http://www.merlyn.demon.co.uk/js-tests.htm>, at
least for the meanwhile.
</repeat>
// above URL is now <URL:http://www.merlyn.demon.co.uk/js-date8.htm>
It has always seemed to me a pity that an Object does not invariably
have a Duplicate method, with a parameter for shallow or deep copy.
In particular, one sometimes wants to take a Date Object, and generate
another object representing the same instant and capable of being
fiddled with independently.
ISTM that D2 = new Date(D1) is commonly used for this.
In testing Gregorian <-> Julian date conversions, I got an unexpected
NaN result from such code. Admittedly D1 was 0000-02-29, i.e. BC 1 Feb
29, proleptic Gregorian Astronomical. But the following, executed in
MSIE 4, gives me NaN in the middle result for any date in the important
Year -68..+99 region :
D = new Date(0) ; D.setFullYear(-19,0,29)
x = [D, new Date(D), new Date(+D)]
/* setFullYear seems to do the sensible thing with any year, unlike
setYear, and so is better than new Date(Y, M-1, D). Using new Date(0)
gets midnight, or at least GMT midnight. /*
Of course, many coders don't care about dates so long ago.
However, on considering what is actually happening, ISTM likely that in
new Date(D) D is converted to string, which is re-converted to be
stored in the new Object.
In new Date(+D) ISTM that the unary + is taken as "don't convert to
string" rather than "convert to Number.
Therefore, new Date(+D) seems likely to be quicker than new
Date(D), and test confirm over 10% improvement in my system.
--
© John Stockton, Surrey, UK. ?@merlyn.demon.co.uk Turnpike v4.00 MIME. ©
Web <URL:http://www.merlyn.demon.co.uk/> - w. FAQish topics, links, acronyms
PAS EXE etc : <URL:http://www.merlyn.demon.co.uk/programs/> - see 00index.htm
Dates - miscdate.htm moredate.htm js-dates.htm pas-time.htm critdate.htm etc.
