Re: How to ensure data consistency?

From: Felix E. Klee <felix.klee_at_inka.de>
Date: Thu, 9 Sep 2004 16:29:24 +0200
Message-ID: <20040909162924.125b8a0a.felix.klee_at_inka.de>


On Thu, 9 Sep 2004 09:13:19 -0400 Laconic2 wrote:
> > Thus, using Joe Celko's solution, there still is the problem that
> > the main table may contain entries for which there are no subtables.

Did you read my correction? But to (hopefully) increase clarity, I'll rephrase the above sentence once again:

Thus, using Joe Celko's solution, there still is a problem: The main table may contain rows for which there are no rows in the corresponding subtable.

For example the table PERSORGS may contain a row with

    persorg_id = 10 (PK)
    persorg_type = "person" (PK)

but the corresponding row may be missing in the table PERSONS.

> It seems to me that you want to accept inputs from users with what
> amounts to uncontrolled inputs the the "Certificate Type" field. That
> is, some one comes along, fills out a form, when s/he gets to the
> Certificate Type field there's a drop down menu, where the last menu
> entry is "None of the Above".

Hey, that's a different issue. See the thread "Other or not?". Sorry for confusing you by not phrasing the above sentence correctly. But let's discuss the other topic nevertheless:

> [Initialization, use and clean up of certificate types]

Correct, that's what I want.

> I don't think you really want referntial integrity, at least for the
> initial entries.

Here I'm loosing you, but maybe this has to do with the initial misunderstanding (i.e. wrong thread). What do you mean by "I don't think you really want referntial integrity"? Of course, I want to make sure

  • that a user can either choose a certificate types that is in the *approved* list, or
  • that she/he can create a new certificate type (in this case duplicate entries are allowed - e.g. two users may have specified "high school diploma" as a new type).

But that's probably not what you were questioning.

Felix Received on Thu Sep 09 2004 - 16:29:24 CEST

Original text of this message