On Wed, Dec 19, 2007 at 11:54:41AM -0500, Greg Troxel wrote:
> I concur.  As a human parser I see 12/2007 (more often) and also
> 2007/12, but 2007-12 is unambigious.  Really one needs to check number
> to parse N/M correctly, and that's a bit scary.

Of the former two, only one passes the range check, so it shouldn't be a
great deal. 2007/12 was the original thought and I find that more
natural than the other order. I would disadvertise is in preference of
2007-12. Of course, many cal implementation behave stupidly like ours
did before, so it should not be assuemd to be portable.

Joerg