On 10/5/06, Carl Eastlund <cce_at_ccs.neu.edu> wrote:
> > > It is also misleading to programmers to treat completion of the
> > > program as an "exceptional" situation; program termination is a
> > > normal occurrence and not a problem to be dealt with exceptionally.
> >
> > Terminology doesn't matter if the semantics is appropriate.
>
> I disagree strongly.
I want to clarify this statement. I don't intend to focus on
terminology in the sense of giving things names. I do intend to focus
on the meaning expressed by a program. If we give something one name
or another (exception versus condition, or whatever) it's not
important if everyone agrees on its behavior. If, on the other hand,
we establish a convention for some program construct (e.g. raising
conditions for abnormal program behavior), then violate that
convention for a special case (e.g. raising conditions for program
termination), then the meaning of a program becomes muddied. When a
program raises the termination condition, should that mean abnormal
termination? When a program installs a condition handler, is it
implicitly aborting program termination? It becomes much harder to
reason about programming with that construct.
I don't know if that's best called an issue of "terminology" or
"semantics" or something else, but it's important and I still think
using "conditions" for program termination goes against it.
--
Carl Eastlund
Received on Thu Oct 05 2006 - 18:16:28 UTC