Re: Real World (Re: Mixing OO and DB)

topmind wrote:

>I cannot speak for every domain, but in the domain I am most familiar
>with: custom biz apps, there is no "real world". It's mostly dealing
>with intellectual property (money, invoices, laws, etc.) Even a paper
>invoice is merely a representation and not THE "invoice" per se. Paper
>is simply an old-style implementation. Before that they maybe used
>rocks or animal bones.
>
>Thus, any operations used are artificial anyhow; all mental
>constructs.

OK, it's partly abstract and intangible, but in the end money buys you food, laws can make people go to jail, etc.

In the example I used (numbers of inhabitants of countries in the world) the operation of counting all people in a country is also "artificial": it doesn't really happen that way. Still, the numbers correspond to that operation. Most attributes in your biz apps are more abstract, but we can still give them concrete meaning in the same way. (This is not to say that we should write code to mimic such operations and consider that code to be a specification.)

>Now, I will agree that OOP generally models how a typical
>clerk might do it: one paper at a time and one operation at time via a
>cursor-oriented pencil. Manual labor has no real set operations.
>Manual labor is generally imperative and thus "navigational".
>
>However, that is NOT the same as "modeling the real world", but rather
>modeling the old fashioned *implementation*.

You have a point. I agree it would be terrible to define the meaning of, say, an attribute, in terms of how particular implementations of the operations that use it are executed in detail. But OO doesn't direct us to do that. It distinguishes between an operation and its implementation. E.g. we can abstract from the details of how collections and operations on them are implemented.

-- 
Reinier