From: "Dale Jr, William" <[log in to unmask]>

> I do not feel that licensing S/W engineers will help since they are not
used
> and if used ignored!  In fact, I think SE as a degree should be eliminated
> completely and SE concepts should be required to be taught to ALL
engineers.
> They end up writing the code and designing the system anyway -- so lets
get
> them trained to do a better job. Otherwise we will be dealing with MatLab
> and Java hackers trying to design reliable systems!

It's possible to make reasonably  reliable systems using Java. It's
capable - in the
right hands, with the right tools - of being not too bad, especially
compared
to C++ in any of its guises. But...

I wouldn't like to bet my life on such a system, even the best Java I've
ever written.
And I now have a reasonable number of KLoCs of both languages under my belt.

OTOH I have bet my life, and others lives, on Ada-83 code I've written.

It's marginally easier to implement some OO constructions in Java than
Ada-95,
though the difference may be a matter of personal taste. The various library
classes,
from JDOM to SWING provide functionality in a neat way, and the Javadoc
facilities
are great, as good or better than the DIANA or other Ada doc methods that
never
really caught on. If only there were *well publicised* Ada equivalents that
came with
every GNAT distribution. Sure there are some, but not many (that I know
of...)

But all you have to do is look at for (int i=0;i<foo;i++) and you'll see
that the C
foundation of Java cripples it. And as for any task involving numerical
analysis,
or interfacing with peculiar peripherals (no representation, no
roll-your-own integers
or floats, no fixed point types...) it just won't cut the mustard.