Re: THe OverRelational Manifesto (ORM)

By the way.

>Hardly. I can write a program as a single procedure without factoring
>out any reusable code, without identifying any cohesive subunits all the
>while reusing variables of compatible types where-ever possible. That
>program will require fewer names because I did not bother to name any
>subroutines and because I maximized variable name reuse.

>The resulting program would be a complete mess that would be impossible
>to debug or maintain.

>Alternatively, I might decompose a program into small cohesive units
>that in many cases will be provably correct and (suitably named) will
>facilitate human comprehension. In each of those small cohesive
>subroutines, I might minimize variable name reuse by limiting each local
>variable to a single task with a name describing that task.

>The resulting program would be well organized and relatively easy to
>debug or maintain.

I don't mean what you can do using the same language (I'm sure all variables you use stared only with "stupid" - (eg "stupid_counter", "stupid_string" etc) and types only with "shit"). I think that's your life.

For example in polimorphous languages only one name can be bound with different implementation. We can't do this in monomorphous languages, so we have to use many different names and remember what each name means. You can write as you wish, but using a polimorphous system anyway seems to be easy? or not? Waiting for your next bark :) Received on Tue Apr 18 2006 - 11:38:21 CEST