|
|
Posted by SUKRU on 10/01/38 11:40
Hello Celko,
Thanks for the reply!
Your right I'm a beginner, but the example above is ficticious!
In fact there is no checked table! I made that up. I choose the checked
name just for practice sake. The original tables are in Dutch, i
wouldn't want to bother you with Dutch ;).
Further more i'm not very familiar with sql-server coding conventions.
I will look it up on the internet.
the main issue is that I couldn't get the original trigger to work
because if my trigger returned more then one row i would get an error.
I've got an example in postgreSQL where the trigger works with the "for
each row function". but SQL-server 2000 doesn't support the "for each
row" statement!
But thank you for your advise, i will apply your advise the next time i
post an message!
--CELKO-- wrote:
> Please post DDL, so that people do not have to guess what the keys,
> constraints, Declarative Referential Integrity, data types, etc. in
> your schema are. Sample data is also a good idea, along with clear
> specifications.
>
> Your design sounds flawed. You have a customer_id which has a huge
> size, and no way to validate it (check digit, anything??) Next, we
> seldom use binary flags in SQL; instead we have a history with (start,
> finish, status) triplets and/or an encodeing scheme for the status.
>
> When I see a table named "CHECKED", I get scared. That is a status,
> not an entity or relationship!! It means that you have split an
> attribute out of an entity and made it into a separate table. If you
> had a "MalePersonnel" and a "FemalePersonnel", you see the flaw of a
> split on "sex" immediately.
>
> Knowing that uppercase names are almost 10 times more often mis-read or
> mis-typed, why did you use them? We had no choice in the days of
> punchcards. In fact, you whoel design seems to be a punch card system
> done in SQL.
>
> You need to learn BASIC RDBMS, and how to use VIEWs.
Navigation:
[Reply to this message]
|