Python vs. C++ Builder - speed of development

Brandon Van Every vanevery at 3DProgrammer.com
Thu Jan 30 22:35:41 CET 2003


John Ochiltree wrote:
>
> You seem to be suggesting that when you discover a problem to be
> solved you approach it from mathematical first principles?

Yes... my problems right now are all 3D math problems, and need to get
solved by 3D math first principles.

> And what
> sort of mathematical pattern could be useful in implementing a
> singleton for instance (leaving aside the obvious distinction
> required between none and unity)?

This is backwards.  We don't condition our implementations to fit abstract
patterns for their own sake!  The abstract pattern is supposed to help me
with my implementation.  For 3D math problems, it does not help.

> Or the notion of the decorator, or
> abstract factory or just about any common day to day programming
> problem?

You're missing the basic point: not everyone's day-to-day programming
problems are "Patterns" in the sense discussed by books like "Design
Patterns."  My day-to-day problems are 3D math problems.  They have
absolutely nothing to do with Patterns.  Really!  I end up thinking things
like "do I need a more abstract container for this thingy?" maybe once every
2 months.  Day-to-day it's computation, correctness of test cases, and
optimization.

> You also appear to think that patterns provide some sort of
> direct solution. Implementation is mostly not supplied (esp. in GoF)
> and you have to implement the pattern in whatever language you're
> working in. The solution that a pattern provides is at the design
> level, probably best illustrated by the UML's class and sequence
> diagrams (amongst others).

I can only reiterate: there are many kinds of design that are *not*
Patterns.

> Or perhaps you have an aversion to acyclic graphs as well?

I don't *need* an acyclic graph right now.  When I do need it, I am jolly
well not going to implement a general purpose, bloated,
solve-everyone's-problems kind of acyclic graph.  I'm going to do something
rather domain specific for my game's AI.  Why?  Because we don't have the
computing horsepower to afford the recombinations of general purpose
approaches.  When we do, in 20 years or so, maybe I'll use acyclic graph
patterns to solve some of my problems.

--
Cheers,                         www.3DProgrammer.com
Brandon Van Every               Seattle, WA

20% of the world is real.
80% is gobbledygook we make up inside our own heads.





More information about the Python-list mailing list