[prog] Joel on Software: Leaky Abstractions

[Jimen Ching]

On Thu, 17 Apr 2003, Robert J. Hansen wrote:
>> Good point.  Does anyone know of any non-leaky abstraction?  Is that even
>> achievable?
>Goedel's Incompleteness Theorem suggests no; Claude Shannon says no.
>Any non-leaky abstraction would necessarily be of the same complexity as
>the problem, hence, it wouldn't be an abstraction.
>
>Leaks are the price you pay for abstractions.

Hm, if this is the case, then it brings up another question: were
abstractions meant to be non-leaky?  Is it possible that abstractions
weren't meant to hide all complexities of a concept.  But instead, hide
just enough of it to solve a problem?

--jc
-- 
Jimen Ching (WH6BRR)      jching at flex.com     wh6brr at uhm.ampr.org