Re: quick FK question

  • From: Yong Huang <yong321@xxxxxxxxx>
  • To: gints.plivna@xxxxxxxxx
  • Date: Tue, 22 Dec 2009 14:45:42 -0800 (PST)

--- On Tue, 12/22/09, Yong Huang <yong321@xxxxxxxxx> wrote:

> Filling in a meaningless value 
> ... It's only bad that certain queries need to 
> be re-evaluated (unless the business case is such that the
> value is guaranteed to be meaningless),

Correction. I should say unless the value is guaranteed to be 
meaningless *and also* the usage of this column won't include this 
meaningless value. So if -1 is used in place of null, 
"quantity > 3" or "quantity = 3" doesn't need a rewrite. But 
"quantity < 3" does. So does "nvl(quantity,0)". Of course if this is 
designed at the beginning, we don't say *re*-write, just say write 
with extra care. This may be a dangerous requirement for business 
analysts that directly write SQLs. So education for them becomes 
more important (for "disciplined use" as said in Codd and Date's 
debate), or it's necessary to build GUI's for them that hides the 
SQL in the mid-tier.

Yong Huang


Other related posts: