/prog/ is really fast

when you read it once a week

We need another ``Genius" sorting algorithm

Why is real OOP slow? Smalltalk is slow, CLOS is slow, Ruby is slow, Javashit is slow. Why can't we have OO and crunch billions of numbers in a millisecond?

>>3
Design a hardware architecture around the actor model, of which OOP is a special case

>>4
No-no-no. Design software, not hardware.

1. An object receives a message to screw some billions of numbers around.
2. The object runs the whole numberfuckery internally with efficient arithmetic primitive operations on unboxed numbers.
3. The object sends a message with the result.

See, it could be performant. What's the problem with this scheme besides the lunatic ideas that "in an OO language everything must be an object" and "you must be able to redefine CAR and CDR"?

>>3
It's not. I have no problem doing things in Java, C++ and C#.

>>5
But designing hardware is fun, writing software sucks.

>>7
Only if you do it as a hobby. Hardware/driver people routinely to our jobs telling me during the interview that they are bored out of their minds by bits and switches.

>>4
I prefer OOP where the actor model is orthogonal.

object message // sync message send "method call"
object @message // actor message send (returns a future)

Objects become actors on demand. Actor refs are scoped like object refs (because they're one and the same) eliminating the need to mess around with pids or mailbox ids or whatever.

>>6
Those aren't OOP, those are COP.

>>10
I don't understand. You'll have to define those terms for me as I believe they are OOP.

>>11
C+# and Java are Class-Oriented Programming languages, not Object-Oriented. You just go ahead and try to modify an object without modifying a class in those languages, and the diff should be obvious.