Re: The 1980 ACM Turing Award Lecture The Emperor's Old Clothes

[Stephen Leake <stephen.a.leake.1@gsfc.nasa.gov>]

Hyman Rosen <hyrosen@mail.com> writes:

> Oh, good lord! Do you program in the real world?

Yes.

> In the real world, remote CORBA servers go down. 

I've never used a CORBA server.

> In the real world, remote databse servers go down. 

I've never used a remote database server

> Most of the time they don't, so when they do, you throw an exception
> when you lose connectivity so you don't blithely ignore the problem,
> but you also don't litter every step of the way with checks.

Sounds like a good partial plan. How do you handle the exceptions at
the top level?

In the part of the real world that I work in (hard real time embedded
systems), we need to show that _every_ exceptional situation has been
anticipated and appropriately handled. Sometimes that means allowing
for switching to redundant hardware. That will typically be a huge
"mode change". It is not something that is triggered by some
exception, but rather by a certain pattern of failures, possibly with
confirmation by an external supervisory system (ground control, in the
case of satellites).

In this world, exceptions are used during the debug and test phases,
and sometimes the compiler checks are turned off for the final build,
because we've proven that they will never fail.

-- 
-- Stephe