What do you mean, exactly? If I use the calculator in kstars, it can calculate any 
date in the Gregorian calendar. 
 
It should be noted, however, that the widget is limited by the start of the Gregorian 
era (actually, 160 years after the beginning of the Gregorian Calendar). If you enter 
any date before 15 September 1752 (which is the date of introduction of the 
calendar in the UK), it'll reset to that date. 

Ahh, I see. 
I didn't think of the limitations of the Calendar. Is there any possibility to use a 
different one, for the ancient data? 

Hi, 
 
Thiago's absolutely right, this is a limitation of the QDate class, which is undefined 
for dates before 1752 and after about 8000.  We have discussed using an additional 
date scheme for more remote dates.  The problem is, it gets very complicated to try 
to continue to use calendar dates.  We will likely try to just use Julian Days for 
remote dates, and leave it to the user to decide how the simple count of days maps 
onto the more complicated calendar system. 
 
BTW, this is covered in the "What Time is it?" section of chapter 2 of the Handbook.  
I know, I know, no one reads documentation ;) 
 

I'd just like to take the opportunity to reiterate my disagreement with the 1782 date in 
QDate. The Gregorian calendar was introduced in 4 October 1582 (the next day 
being the 15 October), by Pope Gregory XIII. 
 
If an arbitrary date is to be chosen, the date it was first introduced would suit best. 
Not the date it was introduced in any given country. 
 

This bug has been fixed.  We have written our own date/time classes to replace QDateTime, KDatePicker, KDateEdit, etc.  The replacement classes can be found at kdeedu/libkdeedu/extdate, if anyone is interested.

In KStars, you can now set the date to anything between years -50000 and +50000.  This range may eventually be extended (there's no theoretical limit to the range, but the extrapolation of object positions becomes more and more meaningless at very remote dates).