Re: 3 value logic. Why is SQL so special?

From: J M Davitt <jdavitt_at_aeneas.net>
Date: Thu, 21 Sep 2006 00:53:46 GMT
Message-ID: <uolQg.7335$L15.3409@tornado.ohiordc.rr.com>

Roy Hann wrote:
> "Chris Lim" <blackcap80_at_hotmail.com> wrote in message
> news:1158642255.069101.30980_at_h48g2000cwc.googlegroups.com...
>

>>Marshall wrote:
>>
>>>In comp.lang.java.programmer, it was not uncommon to have
>>>someone describe a problem that would trivially be solved
>>>by adding a new class, but they would reject that solution because
>>>"too many classes."
>>
>>The key difference here is 'trivially'. Adding lots (and we are talking
>>about lots when you consider the number of nullable columns) of tables
>>makes it much hard to query that data (especially if you also disallow
>>outer joins!). And if you are going to prove me wrong with an example,
>>at least use a table with lots of nullable columns instead of just one
>>or two.

>
>
> You are incorrectly assuming that each nullable column gives rise to a
> separate table. Based on the three or four dozen databases that I work
> with regularly that wouldn't usually happen. In most cases a table will
> have 5, 10, maybe more nullable columns because it conflates two or three
> different entity types. Thus just one or two additional tables are often
> sufficient.

Shh!! Next thing you know you'll have folks thinking that life is simpler with two or three more "tables" than it is with a dozen or so more "columns" of ambiguous... content. (What? You thought I was going to say, "data?" Or, "values?" Ha!)

> But if more were required, so what? You've got to write the code to sort it
> out somewhere, so why not in the query?
>
> Roy
>
>
Received on Wed Sep 20 2006 - 19:53:46 CDT