Re: [Gnumed-devel] how to think like a computer scientist

[Karsten Hilbert]

On Wed, Apr 10, 2013 at 09:13:34PM +1000, Liz wrote:

> > Ever wondered how all this programming works ?
> > 
> > Why not get started today ? Here are two books to start with

> I've tried and tried
> It's hard after many years of learning to do pattern recognition and
> diagnostic heuristics which involve initially widening the search and
> then narrowing it
> but I don't think like a computer scientist

You do but you (probably) haven't experienced that.

I agree programming as in writing new software isn't like
practicing medicine.

Debugging, however, is very much so.

It's about asking the right questions, prodding the
user/patient for relevant answers, trying to elucidate
certain details or sequences of events, looking for clues in
the narrative, putting stuff in perspective, looking at
where in the big picture certain symptoms may fit - or not -
or what else they might mean if they don't quite fit. Based
on that devise a strategy, perhaps initiating a
well-selected probe into the system (lab test, tracking
variables), perhaps executing temporary diagnostic
interventions (instrumenting software, putting patients on a
medication) and evaluating their outcome, eventually finding
a root cause or determining the relative harmlessness of the
behaviour (declaring that a bug is a feature) or even facing
the incurable terminality of bit rot.

Karsten
-- 
GPG key ID E4071346 @ gpg-keyserver.de
E167 67FD A291 2BEA 73BD  4537 78B9 A9F9 E407 1346