Re: Database design

JOG schrieb:
> Alexandr Savinov wrote:

>> JOG schrieb:
>>> Mark Johnson wrote:
>>>> "x" <x_at_not-exists.org> wrote:
>>>>
>>>>> "Roy Hann" <specially_at_processed.almost.meat> wrote in message
>>>>> news:3--dnYnbkrrCfmTenZ2dnUVZ8qadnZ2d_at_pipex.net...
>>>>>> "x" <x_at_not-exists.org> wrote in message news:dtcjfn$f87$1_at_nntp.aioe.org...
>>>>> Well, the slippery part is not that amusing after a while.
>>>>>> I am more inclined to read it as just the usual witless gaff of noticing
>>>>>> that the bounding box of a printed representation of a table has length
>>>>>> width and leaping to the conclusion that a table is therefore
>>>>>> two-dimensional; planar: flat.
>>>> Then I certainly stand to be corrected. I thought the relation was
>>>> thought to be essentially an unordered set or list of entities, and
>>>> nothing more.
>>> A tuple does not equate to an entity, in fact far from it.
>> So what is then an entity?
>>

I am afraid that both terms (entity and tuple) have wider scope than entity-relational model and normal forms. And this is precisely why it is interestingly to compare them. In particular, (if you insist that they are different things) why entity is NOT a tuple and vice versa why tuple is NOT an entity? Of course, we might add to this comparison list some other terms like object, row, record etc. But since we cannot even formulate the difference between the above two terms I think adding more complexity does not make sense. Another problem is that you cannot define terms indefinitely - you must stop somewhere. So it is not a solution. Such terms as the above two are precisely a kind of such terms that are very difficult to define. We can only characterize them in great extent informally.

> This whole 'flat' debate is nonsense too. Write a database down in its
> mathematical form, devoid of tables, and tell me how on earth it can be
> flat (which semantically means two-dimensional of course), deep, fat,
> thin, whatever. If you mean it doesn't support composite types say
> that. If you mean it contains no explicit links, say that. Calling it
> flat is semantically redundant and doesn't aid any real discussion.

I have seen at least two definitions of the term "flat" in this discussion:

• a structure is said to be flat if it consists of columns and rows, that is, any element can be retrieved by specifying a column and a row.
• a structure is said to be flat if it is equivalent to n-dimensional space (that is, with only one level where points are characterized by their coordinates along n dimensions).

It is an answer on your question "how on earth it can be flat".

You can also define a structure as deep. For example, if it is equivalent to a hierarchical space. It is also not excluded that some other term could be helpful such as fat, thin etc. Notice that these terms are far from the most exotic ones in science. (Fuzziness and roughness of sets, charmness of quarks etc.)

-- 
http://conceptoriented.com