Re: NSCalendarDate problem

[Richard Frith-Macdonald]

On Thursday, July 3, 2003, at 07:47  pm, Andreas Hoeschler wrote:

> Hi all,
>
> the two lines
>
>       NSCalendarDate *date = [[NSCalendarDate alloc] initWithYear:0 
> month:0 day:0 hour:18 minute:0 second:0 timeZone:nil];
>       NSLog(@"date %@", date);
>
> produce the following output.
>
> 2003-07-03 20:44:40.516 FBTest[18838] File NSCalendarDate.m: 1239. In 
> [NSCalendarDate -initWithYear:month:day:hour:minute:second:timeZone:] 
> invalid month given - 0
> 2003-07-03 20:44:40.526 FBTest[18838] File NSCalendarDate.m: 1244. In 
> [NSCalendarDate -initWithYear:month:day:hour:minute:second:timeZone:] 
> invalid day given - 0
> 2003-07-03 20:44:40.526 FBTest[18838] date 0000-01-01 06:00:00 +0100
>
> I assume this is a bug. It's a few months since I got a current 
> version from cvs. Has this problem been fixed in the meanwhile or 
> should I look into the sources myself?

GNUstep is obviously trying to be tolerant of programmer errors, and 
restrict the arguments to the valid range.
What behavior are you expecting?   Do you know if MacOS-X causes an  
exception to be raised or the object to be deallocated and nil 
returned?  The OpenStep spec and MacOS-X documentation don't specify 
the behavior, so technically there's nothing wrong with the current 
behavior.