[Chicago] python 3.0 hep

Pete pfein at pobox.com
Mon Dec 8 17:04:53 CET 2008


On Dec 4, 2008, at 10:54 PM, Martin Maney wrote:

> On Thu, Dec 04, 2008 at 11:42:48AM -0600, Michael Tobis wrote:
>> Not that this isn't pragmatically good advice, but didn't I just hear
>> the ghost of Edsger Dijkstra moaning in Dutch somewhere in the
>> distance?
>
> Ah, like this:
>
>  http://www.cs.utexas.edu/users/EWD/ewd10xx/EWD1036.PDF

Thanks, that's a really good piece.  His handwriting is beautiful too.

While I really appreciate his perspective, Dijkstra's argument for  
formal methods in programming is predicated on the ability to create a  
formal specification of the problem at hand.  While that's fairly  
straightforward to do for the domino covering problem he describes,  
the real world is often not so conducive to such approaches.

To put it simply: In all likelihood, Donna in accounting can barely do  
her job by hand, let alone provide a rigorous mathematical description  
of the sort Dijkstra seems to be assuming as a prerequisite.

As programmers in an ambiguous and ever-changing world, Dijkstra's  
approach is insufficient.  What would he have us do, restrict  
ourselves to those problems which can be so specified?  I, for one,  
would rather hack my way through.

--Pete


More information about the Chicago mailing list