Re: 4 the FAQ: Are Commercial DBMS Truly Relational?
Date: Sun, 10 Oct 2004 23:07:51 +0100
Message-ID: <ckcbs0$7i8$2_at_news.freedom2surf.net>
Laconic2 wrote:
> "Andrew McDonagh" <news_at_andrewcdonagh.f2s.com> wrote in message
> news:ckc0ch$3qm$1_at_news.freedom2surf.net...
>
>>Also, IMHO, premature optimisation tend to create designs that are >>fragile to future changes.
>
>
> Agreed.
>
> And also, it tends to create logical designs that are ... well... illogical.
>
> I like the term premature optimization.
Wish I could take credit, but I can't, its used a lot in the OO and XP community.
> In an earlier lifetime, when dealing with intelligent programs, we used to
> run into another phenomenon: premature binding.
>
> In my experience, if the design is simple, sound, and relevant, and if the
> hardware isn't underconfigured for the load, most of the resulting system
> will perform satisfactorily. You then focus in on those few parts of the
> system that need to be sped up.
>
> But often, when repairing the work of premature optimizers, you end up with
> a complete rat's nest, and one that performs lousy, to boot.
>
> Again, going back to an earlier lifetime: "I know we aren't supposed to use
> GOTO, but it's more efficient this way!" Sigh!
>
Yes, even today, I see people writing OO code and using large switch statements or lots of IFs, cause they are 'more efficient'...(than a virtual method invocation). But I honestly believe its because they don't see how to make that portion of code polymorphic, and the efficiency claim is just to hide this fact.
Andrew Received on Mon Oct 11 2004 - 00:07:51 CEST