On Fri, 3 Nov 2000, Dave Thomas wrote:

> You may be right, but it isn't a Ruby bug...
> 
>      dave[rubyweb/raa 8:39:48] date -d "Fri Nov  3 06:20:59 CST 2000"
>      Fri Nov  3 06:20:59 CST 2000
>      dave[rubyweb/raa 8:39:55] date -d "Fri Nov  3 06:20:60 CST 2000"
>      Fri Nov  3 06:21:00 CST 2000
> 

I'd still consider that suspect, but this is proof by "select isn't
broken".  What does date -d do with 31-DEC-2000 23:59:60 which is where
a leap second would be inserted in real life?  It might be employing
subtle checks to Do The Right Thing (by rolling over) in most cases.
We don't have -d on our date, so I cannot test this.

I certainly hope the people who write and test the date command know more
than me about this sort of thing!
> 
> 
> Dave
> 
	Hugh
	hgs / dmu.ac.uk