-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 01/18/07 17:52, David Fetter wrote: > On Thu, Jan 18, 2007 at 05:42:54PM -0600, Ron Johnson wrote: >> >> On 01/18/07 17:22, Scott Ribe wrote: >>>> But this won't work if one had a text column of dates in various >>>> formats, right? >>> Right. In my case I have bad data from a source I didn't control, exported >>> via code that I do control which happens to output YYYY-MM-DD. Well, except >>> that I don't do what I need to when MM or DD are more than 2 digits, but I'm >>> going back to look at that again ;-) >> Why didn't the PG engine reject these bad-date records at INSERT >> time. This smacks of something that MySQL would do... > > I'm pretty sure it didn't accept these as bad dates, but as text > strings. As you point out, it's a MySQLism to take "we are all here > to go into space" as a valid date. Ah, the relevant snippet from OP: > I have a varchar column That would explain everything. Except why it's a VARCHAR instead of DATE. But that's a whole 'nother discussion. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (GNU/Linux) iD8DBQFFsAtRS9HxQb37XmcRArUSAJ9p519CtEa10tHeOGmr83lXoaRZ1ACcDpl5 VOHvbmhM+kiSARaXKPg5ZfU= =jz33 -----END PGP SIGNATURE-----
