Definitions of date arithmetic

Ken Pizzini ken at halcyon.com
Mon Oct 12 03:49:54 UTC 1998


On 11 Oct 1998 01:32:39 -0000, D. J. Bernstein <djb at cr.yp.to> wrote:
>This example certainly doesn't convince me that mktime()'s handling of
>invalid dates is a bad thing.

True, the examples we have been dealing with have not been crossing
the line that mktime() has trouble with.  Those mainly occur when
one has to deal with a discontinuity, such as imposed by a Summer Time
crossing.

But I'm prepared to drop the idea, because of the point that Joe Celko
raised: there is *no* well-accepted standard in the "real world" on
how to deal with this.  Each application will have to implement its
own rules.  I still wish there were some good way to provide applications
with the knowledge of the calendar which the time portion of the C
library has, so that we didn't have to have each application writer
make their own idiosyncratic mistakes in this realm, but I cannot think
of how to do this in a resonable manner.

		--Ken Pizzini



More information about the tz mailing list