Re: Practical considerations of dealing with two meanings of NULLs

sinister wrote:

> One simplistic method of trying to distinguish multiple meanings possibly
> associated with NULL is to associate particular non-NULL meanings with those
> values.
>
> For example, I might want to store both "unknown" and "known, but not yet
> filled in by data entry clerk." I could associate NULL to one and a
> particular non-NULL value to the other, or non-NULL values to both.
>
> If the data at issue are e.g. strings or something categorical, this is OK.
> (The only problem I can think of is the case where (in the interface---I'm
> using a webserver for that) I give users a selectbox of predetermined
> choices, and there's an associated textbox where they can put in "Other"
> (non-predetermined values)---what's to stop them from overlooking the
> predetermined choice "Unknown" and typing in "don't know" in the textbox?)
>
> But if the data are numerical, it's messy. It's easy if the data are known
> to be nonnegative, because then one can reserve particular nonnegative
> values for this (e.g. "-1 means not filled in by user," "-2 means known").
> But this method just seems ugly and kludgey to me.
>
> Is there any clean method of dealing with this?

Use multiple relations and no NULL. Received on Wed Aug 08 2007 - 14:27:16 CEST