X-no-Archive:yes
This would allow programmers to bypass this validation. As a result, you
still could not quite trust that any datum had been validated.

A good case in point for DBMS-level validation is leap day. How many times
have we either seen this mishandled, or handled by way more code than
actually necessary merely to determine whether or not 2000 was a leap year,
and thus February 29 a valid date or 366 a valid DOY?

And yet, if I am moving validated data from one "table" to another in the
same engine, perhaps to history or to a data warehouse, I really do not need
to revalidate the already validated date. Now, a mechanism to handle that
case would be nice.

And yet, great argument can be made for validating dates or two letter state
abbreviations or a host of historically stable ranges of allowable value in
the presentation layer, rather than sending them across some potentially
vast network, to be validate on the server. But I guess that's not what we
are discussing.

Greg Stigers
http://www.cgiusa.com

-----Original Message-----
From: Frank Gribbin [mailto:[log in to unmask]]
Sent: Friday, March 09, 2001 10:49 AM
To: [log in to unmask]
Subject: Re: Expensive RDBM Systems (Oracle)


The concern about overhead can be allied by placing a binary flag in the
IMAGE call, eg
1 = validate using imbedded rules and
0 = skip validation

Frank Gribbin
Potter Anderson & Corroon LLP