| Joe Zitzelberger 2004-06-28, 3:55 pm |
| In article <10doir79ojc8cad@corp.supernews.com>,
"Rick Smith" <ricksmith@mfi.net> wrote:
> Joe Zitzelberger <joe_zitzelberger@nospam.com> wrote in message
> news:joe_zitzelberger-8EB06C.08082925062004@corp.supernews.com...
>
> Mr Zitzelberger, your statement seems to be an absolute not
> supportted by anything I have posted or read lately. Stroustrup,
> for example, stated that "existing techniques developed over
> decades are adequate"; which, to me, suggests that benefit is
> relative. A similar sense comes from McConnell.
I never want to just be "adequate". I want people to be amazed at how
easy it is to enhance, extend and maintain systems that I write.
McConnell also has some not-so-nice things to say about "existing
techniques developed over decades" in other areas. I think I cited him
extensively in a thread about source formatting and indention where he
shot down decades of block structure source format style as a
code-horror (with the hair-raised icon next to it).
>
> While this would make it easier to accomodate some changes,
> but what you suggest is little more than a procedural server for a
> procedural client, effectively modular progrmming. Consider that
> field definitions (usage and picture) must still be consistent
> elsewhere in the application. That is, changes in the server will
> show up somewhere in its clients.
New technologies often build on the current state-of-the-art. Much of
OO is simply compiler support for techniques that have been found to be
helpful (data hiding, polymorphism, et al). For example,
variable-length records with different data lengths or a record-type
indicator represent an early form of polymorphism -- processing was
similar for some parts, different for other parts of different records.
Of course it was completely up to the programmer to enforce this. With
OO, you get compile time checks that it was enforced.
|