Re: Mixing OO and DB
Date: Fri, 29 Feb 2008 21:04:11 -0800 (PST)
Message-ID: <62413116-9b62-4d5e-9aa8-5ccd0729bf7c_at_e10g2000prf.googlegroups.com>
On Feb 29, 3:33 pm, Patrick May <p..._at_spe.com> wrote:
> Marshall <marshall.spi..._at_gmail.com> writes:
>
> The fact that you don't understand it doesn't make it
> impossible. One approach is the Dependency Inversion Principle. The
> application code depends on an interface and one implementation of
> that interface uses a relational database. The database schema can
> change without impact to the application code because only the
> implementation of the interface is affected. The relational database
> implementation can be replaced with some completely different
> mechanism that implements the same interface and the application
> remains unaffected. Decoupling.
Ten years ago I thought that sort of thing seemed a promising approach. Today, after having built many systems using this and a variety of other approaches, it is obvious to me why what you suggest requires throwing away most of the value of the RDBMS. For anyone who doesn't understand where most of the value of the RDBMS lies, it will be hard to understand my opposition. In fact I am sure that the me of ten years ago would think my opposition baseless!
> The general ledger schema is not the schema for the content
> management application.
Agreed. The general ledger schema and the general ledger application cannot be decoupled.
> I never said "allow any arbitrary schema to
> be used with any arbitrary application." You're being ridiculous.
Some claims:
- It is possible to decouple X and Y where X is an application and Y is a schema.
- It is possible to decouple general-ledger-application and and general-ledger-schema.
You claim 1. I claim 2 is a straightforward existential instantiation of your universally quantified claim 1. You say not, but give no argument, reasoning, or details, and yet it is clear on its face that 2 is simply an instantiation of 1.
Naturally you object that 2 is ridiculous! And I agree: it is indeed ridiculous. One might even say that argument 2 reduces argument 1 to an absurdity.
Marshall Received on Sat Mar 01 2008 - 06:04:11 CET