Re: Views for denomalizing

"DA Morgan" <damorgan_at_x.washington.edu> wrote in message news:1107441768.808277_at_yasure...
> Dawn M. Wolthuis wrote:
>
>>>1. There is no such thing as a database that confirms with SQL92.
>>>It just doesn't exist. Nor is SQL92 one thing.
>>
>> I only used SQL92 in the statement because I believe that the SQL3
>> specification allows for "nested relations" or "child tables" or
>> you-pick-the-language as I seem to be missing. I did think there was a
>> single SQL92 standard even if implemented differently by each vendor. Is
>> that not the case?
>
> You are still incorrect ... SQL92 is not one thing ... and there is not
> a single commercial RDBMS that ever met it nor would anyone expect that
> there would ever be. It appears that you are commented on something
> that you haven't actually read.

I have not read the SQL-92 standards document, but I did think there was one. What I have read is the ODBC SQL standards and I realize that they came up with something more standardized because no two DBMS's implementing SQL-92 did it in the same way. I thought that had more to do with the non-core (if there is such a think) SQL features, but OK, I'll take your word for it unless I hear from someone else that there really is a SQL-92 standards document (even if not implemented the same across vendors).

>>>2. There is no relationship between 3NF, or anyNF and views or
>>>denormalization.
>>
>> Views, like base tables, map to relations. As relations, we can discuss
>> whether they are in this or that normal form -- isn't that correct? If
>> not, what is it precisely that I am getting wrong.
>
> We can discuss anything. We can discuss invisible purple rhinos. But
> that doesn't make a relationship to denormalization exist.

I have always talked about relations in the "logical model" (which might be an old term at this point) as being what we normalize in our model. Is that correct?

In the logical model, I model both base relations and derived relations (aka views). I realize some people might narrow their model to only the base relations, at least at first, but this is legitimate, right?

Then why isn't it appropriate to discuss the degree of normalization of each of those relations (whether base or not)?

>>>3. Absolutely not. A view might, and I emphasize MIGHT have one for
>>>that purpose but most views written have nothing to do with end users.
>>
>> The users to whom I'm referring are you and me.
>
> I'm not a user. Not of drugs and not of views. Nor would I expect to
> find "end-users" posting at c.d.theory.

I am not the one who wrote "end users" above and if I used the term "end" to modify users (and I don't recall that I did, but we have found that my memory is not exactly running optimally) it would be because I was referring to the people at the end of some layers of software who are taking business models and implementing them, writing code to the API for a DBMS product, using whatever tools. I'm referrnig to this set of tools and APIs as the DBMS and surely SOMEONE is using them, else nothing is happening with that DBMS at all.

If you, Daniel, are not a user of any DBMS tools or APIs, then I can understand why you are missing my point. When developers in database companies are writing tools and API's for their DBMS products, it is the person who they see as their user that I am calling the DBMS user. Does that work for you?

>> A DBMS is a software application for a set of users -- the users I'm
>> referring to are any who use the DBMS software directly. That would
>> include any person or software issuing SQL commands, for example.
>
> But most RDBMS users are machines. Machines doing batch processing at
> 2am. Most are printing billing invoices. Most are doing a lot of things
> that do not involve human interaction. There are no "users."

Who is coding to the API's -- those are users of the DBMS.

>>>4. If I understood your intent I'd likely still disagree. But based
>>>on what I think you are saying ... no. I can build a view to do
>>>whatever I wish given the tools available.
>>
>> Perhaps you can eventually make the RDBMS tool do what you want it to do.
>
> Everything except my cat and politicians.
>
>> I can use Word instead of InDesign to lay out a poster (it's that
>> hammer/nail thing) I'm looking for improvements in software development
>> and maintainability of database applications and current RDBMS tools seem
>> to be one roadblock, from my perspective, to significant improvements.
>
> I can not imagine how. I think that it is your lack of actual knowledge
> of these products standing in the way.

Hmmmm. There is always that possibility, but at least I don't have problems imagining, eh? --dawn

> --
> Daniel A. Morgan
> University of Washington
> damorgan_at_x.washington.edu
> (replace 'x' with 'u' to respond)
Received on Fri Feb 04 2005 - 03:29:43 CET