Re: A creepy story about dates. How to prevent it?

On Thu, 2003-06-19 at 01:06, Joel Rees wrote:
> (Comments from the peanut gallery here)
>
> > > > IMHO it is a bug.  We don't let postgresql "guess" about a lot of more
> > > > obvious things (i.e. int4 to int8 casting, etc...) and letting it guess
> > > > about dates makes it non-ACID compliant.
> > >
> > > How do you arrive at that conclusion?
> >
> > The same way I come to all my conclusions, logic. :-)  but seriously...
> >
> > Why not accept a date of 04/44/2003 and just wrap it into May?  It's
> > the same kind of thing.
>
> Is it? Similar, perhaps, but at least you can be pretty sure that 44 and
> 2003 are not valid months. (Not that I want the database fixing that for
> me, either.)
>
> > I told my database where I live, and expect it to
> > only accept dates that are valid in my locale.
>
> I wouldn't suggest that. Locale is not dependable because there is
> simply no dependable way of mapping, for example, IP address to a
> physical location, much less to a cultural location.

The locale specified by the SysAdmin should be canononical.

[snip]

> Good question. Another good question is how the database would implement
> the check.

Other databases do it.  It can't be *that* hard to do.

OTOH, Andrew Snow's method (alway use ANSI standard YYYY-MM-DD)
is guaranteed to work.  Have your app convert to that format before
inserting, and then PostgreSQL is guaranteed to puke if there's
a problem.

--
+-----------------------------------------------------------+
| Ron Johnson, Jr.     Home: ron.l.johnson@cox.net          |
| Jefferson, LA  USA   http://members.cox.net/ron.l.johnson |
|                                                           |
| "Oh, great altar of passive entertainment, bestow upon me |
|  thy discordant images at such speed as to render linear  |
|  thought impossible" (Calvin, regarding TV)               |
+-----------------------------------------------------------