|
|
Posted by Chung Leong on 08/09/06 18:25
Kenneth Downs wrote:
> Object-oriented forms frameworks appear to me to be the drome. They catch
> you with a dream and you starve to death (lost productivity and billing)
> eating the imaginary food (silver bullet of perfect programs).
It's easy to show an OO architecture in a diagram. That's where I think
people make a mental mistake. A good way to visually present
information is not necessarily a good way to create that information in
the first place. In fact, the opposite is likely to be true. It's hard
to fit a system of interacting parts into a pre-defined structure when
the rules governing the interactions are independent of those governing
the structure.
The recent fad with design pattern is another case of failing to see
the asymmetry between observation and creation. The fact that you can
identify patterns within a complex system does not mean you can
recreate a similiarly complex system using those patterns. The patterns
might give you some insights into the original system, but they are not
good starting points for a new one.
> All this being said, there is much value in having library routines that
> handle repetitive tasks or make your code more readable. We have found
> that simple routines like "hSpan($content)" make code readable, while
> comprehensive routines like "hInputsFromTable($table_id)" that create a
> complete set of form inputs are also very useful. But structuring it all
> into a nested hierarchy of objects just adds more labor with no payoff.
You're of course, not obliged to use those routines when they won't
work in a particular situation. They're like contractions we use when
we talk. We say "google" this and "google" that when the word actually
signifies a number of different actions: opening a web browser, typing
in www.google.com, etc. "To google" is a higher level function than "to
type", which in turn is a higher level function than "to press," so
there is a sort of hierarchy. We don't need to pay attention to it
however.
Navigation:
[Reply to this message]
|