Re: storing survey answers of different data types
Date: Fri, 24 Apr 2009 16:00:20 GMT
> On Apr 24, 6:01 am, paul c <toledobythe..._at_oohay.ac> wrote:
>> Joe Thurbon wrote:
> Well, you can get a lot of mileage out of the theory... I think that
> it's not a binary distinction between a model being 'broken' or
> 'unbroken', it's more of a spectrum between things that are easy and
> hard. You can get a lot of mileage out of the RDBMS, and if/when you
> get to a point where things start to look difficult ( "Boy, I don't
> immediately see a way to do this"), check to make sure that you really
> understand the theory, you're "doing it right" -- implementing a good
> solution for the problem, and if that fails, maybe you can fudge
> something other than the database, because the database is doing so
> much else for you.
> But if you decide that the view updating problem means that Codd's
> model is broken, therefore should be abandoned, what are you going to
> replace it with?
> In other words, if your car runs a little rough at 60 MPH, are you
> going to abandon the car and start walking? Are you improving the
> situation then?
Those aren't very pertinent comparisons. A better one would be what do you do if your car breaks down? Either you wait and do nothing or you walk home. One db school of thought says it's better to do nothing than give a wrong answer. Another says enforce requirements that make the answer right. Eg., the first school says you can't infer a value for A from the value of A JOIN B. I gather the reason is that is the situation in predicate logic. At the same time, the first school is quite happy to use UNION to augment a table, even though that is not quite the situation in predicate logic where an 'OR' operation doesn't produce a single value. The second school recognizes that relational results are not directly comparable to predicate logic results even though they have similar algebras. Received on Fri Apr 24 2009 - 18:00:20 CEST