[clug] [OT] Bjarne Stroustrop on

[Conrad Canterford]

On Thu, 2010-04-08 at 13:53 +1000, Alex Satrapa wrote:
> On 08/04/2010, at 12:32 , Steve McInerney wrote:
> > But I had studied computing theories at Uni. I knew the theory of how
> > memory worked and generically the how & why of how to code (structured)
> > in low level languages like assemblers; as well as higher like C,
> > Pascal, Modula2 and Cobol.

> These are more like trade skills, from my perspective {...}
> Scientific discipline is good, but test-build-test is a specific
> example of it that we can teach to folks rather than having every
> programmer trying to solve problems in an industrial setting by
> performing rigorous experimentation and inventing new computing
> models.

I'm with Steve on this one (who, quite co-incidentally, started in
Defence Manpower at the same time I stared in Defence Logistics - also
on Unisys 2200 series mainframes).

If all you want is programmers who know how to slot pre-defined
solutions ("pipes") together, the trade approach is fine. If, however,
you want a programmer who can look at a problem and determine the best,
practical solution to the problem, then a university-trained programmer
is what you need. Personally, I think the second kind are the better -
but I may well be biased.

I might add, I think pure engineers (especially software engineers) are
a bit prone to being too far removed from the real world. They come up
with great conceptual ideas that spectacularly fail to live up to
expectations when introduced to real environments.

Conrad.