Databasen - NULL dates

[Nigel Rantor]

Neil de Carteret wrote:
> On 18/10/06, Nigel Rantor <wiggly at wiggly.org> wrote:
>> We got into a discussion about whether or not to use NULL values for
>> dates that are as-yet-unknown.
> 
> I think we're unanimous.  So go on then, what reasons were cited for
> not using NULL in this case?

Yes, well, I would just use NULLs, I think that's what they're for.

The main 'argument' was that NULL is not a valid date.

My response that some date in the far future is not logically valid 
either in this case, and that you're going to get some really weird 
results initially fell on deaf ears until we moved beyond the example we 
were currently on and discussed systems that might actually need to 
model dates that are in the far future.

The fact that different systems have different maximal dates before they 
wrap was also brought up.

The candidate still said that he would only consider using NULLs if you 
  really had to though, and generally would use dates in the far future 
because it had worked for him so far.

Anyway, this is not the main, or even one of the big reasons, that he 
didn't get the job.

My question was a sanity check to see if there were any really good 
reasons not to use NULLs for date values that aren't yet known, so 
thanks for the replies :-)

   n